mar19 medpac entirereport sec

Transcript

1 MARCH 2019 Report to the Congress: Medicare Payment Policy REPOR G RESS T TO THE CON Medicare Payment Policy | March 2019 Washington, DC 20001 425 I Street, NW • Suite 701 • (202) 220-3700 • Fax: (202) 220-3759 • www.medpac.gov

2 The Medicare Payment Advisory Commission (MedPAC) is an independent congressional agency established by the Balanced Budget Act of 1997 (P.L. 105–33) to advise the U.S. Congress on issues affecting the Medicare program. In addition to advising the Congress on payments to health plans participating in the Medicare Advantage program and providers in Medicare’s traditional fee-for-service program, MedPAC is also tasked with analyzing access to care, quality of care, and other issues affecting Medicare. The Commission’s 17 members bring diverse expertise in the financing and delivery of health care services. Commissioners are appointed to three-year terms (subject to renewal) by the Comptroller General and serve part time. Appointments are staggered; the terms of five or six Commissioners expire each year. The Commission is supported by an executive director and a staff of analysts, who typically have backgrounds in economics, health policy, and public health. MedPAC meets publicly to discuss policy issues and formulate its recommendations to the Congress. In the course of these meetings, Commissioners consider the results of staff research, presentations by policy experts, and comments from interested parties. (Meeting transcripts are available at www.medpac.gov.) Commission members and staff also seek input on Medicare issues through frequent meetings with individuals interested in the program, including staff from congressional committees and the Centers for Medicare & Medicaid Services (CMS), health care researchers, health care providers, and beneficiary advocates. Two reports—issued in March and June each year—are the primary outlets for Commission recommendations. In addition to annual reports and occasional reports on subjects requested by the Congress, MedPAC advises the Congress through other avenues, including comments on reports and proposed regulations issued by the Secretary of the Department of Health and Human Services, testimony, and briefings for congressional staff.

3 MARCH 2019 REPORT TO THE CONGRESS Medicare Payment Policy 425 I Street, NW • Suite 701 • Washington, DC 20001 (202) 220-3700 • Fax: (202) 220-3759 • www.medpac.gov

4

5 601 New Jersey A venue, NW • Suite 9000 425 I Street, NW • Suite 701 Wa shington, DC 20001 Washington, DC 20001 - - - - 202-220-3700 • Fax: 202-220-3759 220 202 3700 • Fax: 202 3759 220 www.medpac.gov medpac.gov www. Glenn M. Hackbar n th, J.D., Chairma Francis J. Crosson, M.D., Chairman Rober t A. Berenson, M.D., F. A.C.P ., Vice Chairma n Jon Christianson, Ph.D., Vice Chairman Mark E. Miller , Ph.D., Executive Director James E. Mathews, Ph.D., Executive Director March 15, 2019 The Honorable Michael R. Pence President of the Senate U.S. Capitol Washington, DC 20510 The Honorable Nancy Pelosi Speaker of the House U.S. House of Representatives U.S. Capitol Room H-232 Washington, DC 20515 Dear Mr. President and Madam Speaker: Report to the Congress: I am pleased to submit the Medicare Payment Advisory Commission’s March 2019 Medicare Payment Policy . This report fulfills the Commission’s legislative mandate to evaluate Medicare payment issues and make recommendations to the Congress. The report contains 16 chapters: • a chapter that provides a broader context for the report by documenting Medicare and total health care spending and their impacts on federal spending; • a chapter that describes the Commission’s analytic framework for assessing payment adequacy; nine chapters that describe the Commission’s recommendations on fee-for-service (FFS) payment rate updates • and related issues; • a chapter on increasing the equity of Medicare’s payments within post-acute care settings; a chapter that updates the trends in enrollment, plan offerings, and payments in Medicare Advantage (MA) plans; • • a chapter that updates the trends in enrollment and plan offerings for plans that provide prescription drug coverage; a chapter that recommends development of a hospital value incentive program; and • • a chapter responding to a congressional mandate on incentives for prescribing opioids in certain Medicare payment systems and monitoring their use. In this report, we continue to make recommendations aimed at finding ways to provide high-quality care for Medicare beneficiaries while giving providers incentives to constrain their cost growth and thus help control program spending.

6 In light of our payment adequacy analyses, we recommend positive payment updates in 2020 for three FFS payment systems (hospital, long-term care hospital, and dialysis); zero updates for three systems (physician, skilled nursing facility, and ambulatory surgical center); and negative updates for three systems (home health, inpatient rehabilitation facility, and hospice). For two of these sectors, we include additional recommendations to the Secretary of Health and Human Services to improve payment accuracy by: • requiring ambulatory surgical centers to report cost data and • continuing to revise the skilled nursing facility prospective payment system and annually recalibrate it as needed. In addition, in the Commission’s continuing effort to move payments from volume to value, we recommend the replacement of Medicare’s four current hospital quality programs with a single hospital value incentive program. Significantly, this recommendation would provide hospitals with higher aggregate payments than they would receive under current law. However, these additional payments would not be distributed across the board but, instead, would be distributed based on the quality of care hospitals provide. I hope you find this report useful as the Congress continues to grapple with the difficult task of controlling the growth of Medicare spending while preserving beneficiaries’ access to efficiently delivered, high-quality care and providing equitable payment for providers. Sincerely, Francis J. Crosson, M.D. Enclosure

7 Acknowledgments time and knowledge. They include Rochelle Archuleta, This report was prepared with the assistance of many Cristina Boccuti, Tom Bradley, Anna Cook, James people. Their support was key as the Commission Cosgrove, Juliette Cubanski, Akin Demehin, William considered policy issues and worked toward consensus on Dombi, Theresa Forster, Jane Galvin, Bruce Gans, David its recommendations. Gifford, Zinnia Harrison, Doug Johnson, Joanna Kim, Despite a heavy workload, staff members of the Centers Lane Koenig, Tom Kornfield, Ashley McGlone, Jennifer for Medicare & Medicaid Services and the Department McLaughlin, Kara Newbury, Daria Pelech, Judi Lund of Health and Human Services were particularly Person, Danielle Pierotti, William Prentice, Erica Rogan, helpful during preparation of the report. We thank Carol Sherri Smith, Steve Speil, Aaron Tripp, Cori Uccello, Blackford, Erick Chuang, Mitali Dayal, Kadie Derby, Rebecca Yip, Carolyn Zollar, and Steve Zuckerman. Elizabeth Goldstein, Kate Goodrich, Steve Heffler, Michele Hudson, John Kane, Michelle Ketcham, Diane Once again, the programmers at Social and Scientific Kovach, Jana Lindquist, Larry Liu, Hillary Loeffler, Systems provided highly capable assistance to Commission Cindy Massuda, Joshua McFeeters, Gil Ngan, Rebecca staff. In particular, we appreciate the hard work of Michael Paul, Blake Pelzer, Monica Reed-Asante, Cheri Rice, Brown, Po-Lun Chou, Daksha Damera, Darya Leyzarovich, Tiffany Swygert, Scott Talaga, Donald Thompson, David Sravani Mallela, Sanee Maphungphong, Shelley Mullins, Vance, and Laurence Wilson. Lorena Ortiz, Cindy Saiontz-Martinez, and Susan Tian. Finally, the Commission wishes to thank Hannah Fein, The Commission also received valuable insights and Mary Gawlik, and Melissa Lux for their help in editing assistance from others in government, industry, and and producing this report. the research community who generously offered their ■ March 2019 | Report to the Congress: Medicare Payment Policy v

8

9 Table of contents Acknowledgments v ... ... Executive summary xiii Chapters 1 Context for Medicare payment policy ... 3 ... 7 Introduction National health care spending ... 7 11 Medicare spending ... Medicare’s financing challenge ... 18 ... 23 Health care spending consumes growing shares of state and family budgets Recent trends in life expectancy, morbidity, and mortality ... 27 The relationship between Medicare spending and quality ... 32 Baby boomers will make up the next generation of Medicare beneficiaries 34 ... Inefficient spending suggests Medicare could spend less without compromising care, but not without challenges ... 35 Conclusion ... 41 2 Assessing payment adequacy and updating payments in fee-for-service Medicare ... 51 Background 55 ... Are Medicare payments adequate in 2019? ... 55 What cost changes are expected in 2020? ... 59 How should Medicare payments change in 2020? ... 60 Payment adequacy in context ... 61 ... 3 Hospital inpatient and outpatient services 65 Background ... 69 ... Are Medicare payments adequate in 2019? 70 How should Medicare payment rates change in 2020? ... 89 4 Physician and other health professional services ... 97 ... 99 Background Are Medicare fee schedule payments adequate in 2019? ... 100 ... 119 How should Medicare payments change in 2020? 5 Ambulatory surgical center services 127 ... Background ... 129 Are Medicare payments adequate in 2019? ... 131 How should Medicare payments change in 2020? ... 144 Outpatient dialysis services ... 155 6 ... 157 Background Are Medicare payments adequate in 2019? ... 161 How should Medicare payments change in 2020? ... 174 7 Cross-cutting issues in post-acute care ... 183 Medicare’s payments remain high, and revisions to the SNF and HHA payment systems need to be implemented ... 185 Quality measures should focus on claims-based outcome measures ... 186 ... Conclusion 187 Report to the Congress: Medicare Payment Policy March 2019 | vii

10 8 Skilled nursing facility services 193 ... ... Background 197 Are Medicare payments adequate in 2019? ... 199 How should Medicare payments change in 2020? ... 215 ... 217 Medicaid trends 9 Home health care services 227 ... ... 231 Background 236 Are Medicare payments adequate in 2019? ... How should Medicare payments change in 2020? ... 245 10 Inpatient rehabilitation facility services ... 251 Background ... 255 260 Are Medicare payments adequate in 2019? ... ... 273 How should Medicare payments change in 2020? Long-term care hospital services 11 281 ... Background ... 285 Are Medicare payments adequate in 2019? ... 286 How should Medicare payments change in 2020? 300 ... 12 Hospice services ... 311 Background ... 313 Are Medicare payments adequate in 2019? ... 315 How should Medicare payments change in 2020? ... 335 13 The Medicare Advantage program: Status report 343 ... ... Background 349 Trends in enrollment, plan availability, and payments ... 349 Medicare Advantage risk adjustment and coding intensity ... 359 Quality in Medicare Advantage is difficult to evaluate ... 368 ... 378 Future direction of MA payment policy 14 The Medicare prescription drug program (Part D): Status report 385 ... ... 389 Background ... 395 Enrollment, plan choices in 2018, and benefit offerings for 2019 Plan sponsors and their tools for managing benefits and spending ... 400 Drug pricing ... 404 Program costs ... 408 ... 415 Beneficiaries’ access to prescription drugs Quality in Part D 416 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Redesigning Medicare’s hospital quality incentive programs ... 429 Background ... 431 Design of a hospital value incentive program ... 432 Scoring methodology ... 435 Converting HVIP points to payment adjustments using peer grouping ... 436 Comparison of HVIP model to existing hospital quality programs ... 441 444 Recommendation to redesign hospital quality incentive programs ... Table of contents viii

11 16 Mandated report: Opioids and alternatives in hospital settings—Payments, incentives, and Medicare data ... 451 Introduction ... 455 How Medicare pays for opioids and non-opioid alternatives in hospital settings ... 455 Incentives for prescribing opioids and non-opioid alternatives in hospital settings ... 459 Medicare monitoring of opioid use through claims and other data ... 463 Policy options for tracking opioid use in hospital settings ... 468 Appendix Commissioners’ voting on recommendations ... 479 A ... Acronyms 485 More about MedPAC Commission members ... 491 493 Commissioners’ biographies ... 497 ... Commission staff March 2019 | Report to the Congress: Medicare Payment Policy ix

12

13 Executive summary

14

15 Executive summary coordinate services across time and care settings. To By law, the Medicare Payment Advisory Commission address these problems directly, two approaches must be reports to the Congress each March on the Medicare pursued. First, payment reforms need to be implemented fee-for-service (FFS) payment systems, the Medicare more broadly, coordinated across settings, and pursued as Advantage (MA) program, and the Medicare prescription expeditiously as possible. Second, delivery system reforms drug program (Medicare Part D). In this year’s report, we: that have the potential to encourage high-quality care, consider the context of the Medicare program in terms • better care transitions, and more efficient provision of care of the effects of its spending on the federal budget and need to be enhanced and closely monitored, and successful its share of national gross domestic product (GDP). models need to be adopted on a broad scale. • evaluate payment adequacy and make In the interim, it is imperative that the current FFS recommendations concerning Medicare FFS payment payment systems be managed carefully and continuously policy in 2020 for acute care hospital, physician and improved. Medicare is likely to continue using its current other health professional, ambulatory surgical center, FFS payment systems for some years into the future. outpatient dialysis facility, skilled nursing facility, This fact alone makes unit prices—their overall level, home health care, inpatient rehabilitation facility, the relative prices of different services in a sector, and long-term care hospital, and hospice services. the relative prices of the same service across sectors—of critical importance. Constraining unit price increases can • review the status of the MA program (Medicare Part create pressure on providers to control their own costs and C) through which beneficiaries can join private plans to be more receptive to new payment methods and delivery in lieu of traditional FFS Medicare. system reforms. r • eview the status of the Medicare program that provides For each recommendation, the Commission presents its prescription drug coverage (Medicare Part D). rationale, the implications for beneficiaries and providers, and how spending for each recommendation would recommend that a hospital value incentive program be • compare with expected spending under current law. developed. The spending implications are presented as ranges over • as mandated by the Congress, report on incentives one-year and five-year periods. Unlike official budget for prescribing opioid and non-opioid pain treatment estimates used to assess the impact of legislation, these under Medicare’s hospital inpatient and outpatient estimates do not take into account the complete package payment systems and how opioid use in the hospital of policy recommendations or the interactions among setting is monitored by Medicare. them. Although we include these budgetary estimates, our recommendations are not driven by any single budget The goal of Medicare payment policy is to obtain good target, but instead reflect our assessment of the payment value for the program’s expenditures, which means rate needed to ensure adequate access to appropriate care maintaining beneficiaries’ access to high-quality services balanced with preserving the fiscal sustainability of the while encouraging efficient use of resources. Anything Medicare program. less does not serve the interests of the taxpayers and beneficiaries who finance Medicare through their taxes In Appendix A, we list all recommendations and the and premiums. Commissioners’ votes. The Commission recognizes that managing updates and Context for Medicare payment policy relative payment rates alone will not solve what have Part of the Commission’s mandate is to consider the effect historically been fundamental problems with Medicare of its recommendations on the federal budget and view FFS payment systems to date—that providers are paid Medicare in the context of the broader health care system. more when they deliver more services, without regard To help meet this mandate, Chapter 1 examines health care to the value of those additional services, and that these spending growth—for the nation at large and Medicare in systems do not include incentives for providers to March 2019 | Report to the Congress: Medicare Payment Policy xiii

16 Some health care spending is inefficient. For Medicare, particular—and considers its effect on federal and state if such spending could be identified and eliminated, the budgets as well as the budgets of individuals and families. efficiencies achieved could result in improved beneficiary The chapter also reviews recent mortality and morbidity health, greater fiscal sustainability for the program, and trends; profiles the health status of the next generation of reduced federal budget pressures. Certain structural Medicare beneficiaries; and reviews evidence of inefficient features of the Medicare program pose challenges for health care spending, structural features of the Medicare targeting inefficient spending; however, the Commission program that contribute to inefficient spending, and the has made multiple recommendations to the Congress and Commission’s approach to combating those challenges. the Secretary that have the potential to improve the quality In 2017, total national health care spending was $3.5 of care and move the Medicare program toward paying for trillion, or 17.9 percent of GDP. Private health insurance value. spending was $1.2 trillion, or 6.1 percent of GDP. Medicare spending was $705.9 billion, or 3.6 percent of Assessing payment adequacy and updating GDP. payments in fee-for-service Medicare As required by law, the Commission annually makes Health care spending growth has fluctuated recently, first payment update recommendations for providers paid with several years of historic lows, followed by a period of under FFS Medicare. An update is the amount (usually accelerated growth, and most recently a return to modest expressed as a percentage change) by which the base growth. From 2009 to 2013, growth in total health care payment for all providers in a payment system is changed spending and Medicare spending slowed to average annual relative to the prior year. As described in Chapter 2, to rates of 3.7 percent and 4.3 percent, respectively, and then determine an update, we first assess the adequacy of increased to rates of 5.5 percent and 4.9 percent from 2013 Medicare payments for providers in the current year to 2015 before declining to a rate of 4.2 percent (of both (2019) by considering beneficiaries’ access to care, the total and Medicare spending) from 2016 to 2017. quality of care, providers’ access to capital, and Medicare The aging of the baby-boom generation will continue to payments and providers’ costs. Next, we assess how have a profound impact both on the Medicare program those providers’ costs are likely to change in the year the and taxpayers, who primarily finance it. Over the next update will take effect (the policy year, 2020). As part of 15 years, as Medicare enrollment surges, the number the process, we examine payments to support the efficient of taxpaying workers per beneficiary is projected to delivery of services, consistent with our statutory mandate. decline. By 2029 (when most boomers will have aged into Finally, we make a judgment about what, if any, update is Medicare), the Medicare Trustees project there will be just needed. 2.4 workers for each Medicare beneficiary, down from This year, we consider recommendations in nine FFS 4.6 around the time of the program’s inception and 3.0 in sectors: acute care hospitals, physicians and other health 2018. Those demographics create a financing challenge professionals, ambulatory surgical centers, outpatient not only for the Medicare program but also for the entire dialysis facilities, skilled nursing facilities, home health federal budget. By 2041, under federal tax and spending care agencies, inpatient rehabilitation facilities, long-term policies specified in current law, Medicare spending care hospitals, and hospices. Each year, the Commission combined with spending on other major health care looks at all available indicators of payment adequacy programs, Social Security, and net interest on the national and reevaluates any assumptions from prior years debt will exceed total projected federal revenues and will using the most recent data available to make sure its thus either increase federal deficits and debt further or recommendations accurately reflect current conditions. We crowd out spending on all other national priorities. may also consider recommending changes that redistribute The growth in health care spending also affects state payments within a payment system to correct any biases budgets and the budgets of individuals and families. that may make patients with certain conditions financially States pay for a significant portion of Medicaid spending, undesirable, make particular procedures unusually increases in private insurance premiums have outpaced profitable, or otherwise result in inequity among providers. the growth of individual and family incomes over the past Finally, we may also make recommendations to improve decade, and out-of-pocket costs for Medicare beneficiaries program integrity. have grown faster than Social Security benefits. Executive summary xiv

17 approximately 8 percent on average in 2017. After The Commission also examines payment rates for services declining over the last several years, inpatient use per that can be provided in multiple settings. Medicare often beneficiary in 2017 increased by 0.7 percent. Outpatient pays different amounts for similar services across settings. visits per beneficiary also increased by 0.7 percent, a Basing the payment on the rate in the most efficient setting slower pace of outpatient volume growth than in recent would save money for Medicare, reduce cost sharing for years. beneficiaries, and reduce the financial incentive to provide services in the higher paid setting. The Commission Quality of care— From 2013 to 2017, hospital mortality has recommended equalizing rates for evaluation and and readmission rates improved slowly. Patient satisfaction management office visits and additional services provided also improved somewhat: The share of patients who rated in hospital outpatient departments and physicians’ offices their hospital a 9 or 10 on a 10-point scale increased from and recommended consistent payment between acute care 71 percent to 73 percent. hospitals and long-term care hospitals for certain classes of patients. We have also recommended elements of a Providers’ access to capital— Access to bond markets single prospective payment system (PPS) for all post- has been strong, with hospital bond offerings in 2015, acute care to replace the four independent PPSs in use 2016, and 2017 ranging from $24 billion, to $38 billion, today (the skilled nursing facility, inpatient rehabilitation to $35 billion, respectively. While some hospitals struggle facility, long-term care hospital, and home health PPSs) to with low occupancy and limited access to capital, most make payments across all of the post-acute care payment hospitals have good access to capital because of strong all- settings comparable. The Commission will continue to payer profit margins. All-payer margins were 7.1 percent analyze opportunities for applying this principle to other in 2017, only 0.1 percentage point below their all-time services and settings. high of 7.2 percent in 2013. Hospital inpatient and outpatient services In 2017, Medicare payments and providers’ costs— hospitals’ aggregate Medicare margin was −9.9 percent, In 2017, the Medicare FFS program paid 4,700 hospitals down slightly from –9.7 percent in 2016. The profit $190 billion consisting of $119 billion for about 10 margin for relatively efficient providers was about –2 million Medicare inpatient admissions, $66 billion for percent. We project that the overall Medicare margin will about 200 million outpatient services, and $6 billion for decline to about –11 percent in 2019. uncompensated care provided to patients who are not Medicare beneficiaries. On net, between 2016 and 2017, For 2020, the Commission recommends that the Congress overall hospital spending increased $7 billion and hospital update Medicare inpatient and outpatient payment rates spending per FFS beneficiary rose 4.3 percent, increasing by 2 percent. This update recommendation is based from $4,992 to $5,208. on indicators of beneficiaries’ access to hospital care, hospitals’ access to capital, hospital quality, and the As discussed in Chapter 3, most payment adequacy relationship between Medicare payments and hospital indicators (including access to care, quality of care, costs. As we discuss in Chapter 15, the Commission also and access to capital) are positive. Average Medicare recommends a new hospital value incentive program margins continue to be negative, although hospitals with (HVIP) that aligns with our principles for quality excess capacity still have an incentive to see Medicare measurement and replaces the current quality incentive beneficiaries because Medicare payment rates remain programs. The difference between the 2 percent update about 8 percent higher than the variable costs associated and the update amount specified in current law should with Medicare patients. be used to increase payments in the new HVIP. Together, In 2017, the average Beneficiaries’ access to care— these recommendations are expected to increase hospital hospital occupancy rate was 62.5 percent, suggesting payments 2.8 percent by increasing the base payment hospitals have excess inpatient capacity in most markets. rate and the average rewards hospitals receive under the Because Medicare payments exceed the marginal cost proposed Medicare HVIP. In addition, we recommend of providing services, hospitals with excess capacity eliminating the penalties associated with the current have a financial incentive to increase services provided quality incentive programs, which will have the effect to Medicare beneficiaries. Marginal profits were of increasing payments by about 0.5 percent. On net, March 2019 | Report to the Congress: Medicare Payment Policy xv

18 Medicare payments and providers’ costs— CMS currently hospital payment rates would be expected to increase by projects that the increase in 2020 in the Medicare an average of 3.3 percent under our combined update and Economic Index (which measures input prices) will HVIP recommendation. be 2.4 percent. In 2017, Medicare FFS payment rates Physician and other health professional for physician and other health professional services services were 75 percent of commercial rates for preferred provider organizations, unchanged from 2016. Median Physicians and other health professionals deliver a compensation in 2017 was much lower for primary care wide range of services—including office visits, surgical physicians than for physicians in certain specialties, such procedures, and diagnostic and therapeutic services—in as radiology and nonsurgical, procedural specialties, a variety of settings. In 2017, Medicare paid $69.1 billion continuing to raise concerns about fee schedule mispricing for physician and other health professional services. About and its impact on the future availability of primary care 985,000 clinicians billed Medicare: roughly 596,000 services for beneficiaries. physicians and 389,000 nurse practitioners, physician assistants, therapists, chiropractors, and other practitioners. The evidence suggests that Medicare payments for physicians and other health professionals are adequate. Medicare pays for the services of physicians and other Therefore, the Commission recommends that the 2020 health professionals using a fee schedule. Under current payment rate for physicians and other health professional law, there is no update to Medicare’s conversion factor for services be updated by the amount specified in current law. the fee schedule on January 1, 2020. As discussed in Chapter 4, our payment adequacy Ambulatory surgical center services indicators for physicians and other health professionals are Ambulatory surgical centers (ASCs) provide outpatient generally positive. procedures to patients who do not require an overnight stay after the procedure. In 2017, 3.4 million FFS Overall, beneficiary access Beneficiaries’ access to care— Medicare beneficiaries were treated in the 5,603 ASCs to physician and other health professional services is certified to provide services to Medicare beneficiaries. comparable with prior years. Most beneficiaries continue Medicare program and beneficiary spending on ASC to report that they are able to find a new doctor without services was about $4.6 billion. a problem. A small number of beneficiaries report more difficulty, with a higher share reporting problems obtaining Our results, described in Chapter 5, indicate that a new primary care doctor than reporting problems beneficiaries’ access to ASC services is adequate. Most obtaining a new specialist. The number of physicians per of the available indicators of payment adequacy for ASC beneficiary declined slightly, the number of advanced services, discussed below, are positive. practice registered nurses and physician assistants per beneficiary rose, and the share of providers enrolled in Our analysis of facility Beneficiaries’ access to care— Medicare’s participating provider program remains high. supply and volume of services indicates that beneficiaries’ In 2017, across all services, volume per beneficiary grew access to ASC services has generally been adequate. by 1.6 percent. From 2012 to 2016, the number of ASCs increased by an average annual rate of 1.0 percent. In 2017, the number CMS assesses the quality of Medicare- Quality of care— of ASCs increased 2.4 percent. Almost all new ASCs in billing physicians and other health professionals based on 2017 (about 94 percent) were for-profit facilities. From clinician-reported individual quality measures. We report 2012 through 2016, the volume of services per beneficiary three population-based measures: patient experience, increased by an average annual rate of 1.2 percent. In avoidable hospitalizations for ambulatory care–sensitive 2017, volume increased by 1.7 percent. conditions, and rates of low-value care in Medicare. Patient experience scores in FFS Medicare remain high, The first four years of ASC-reported Quality of care— and rates of avoidable hospitalizations for ambulatory quality data show improvement in performance, but care–sensitive conditions continue to decline modestly the measures used within the ASC Quality Reporting from prior years, but there is substantial use of low-value (ASCQR) Program will change substantially in the next care. few years. Among the 11 quality measures for which data Executive summary xvi

19 have a financial incentive to continue to serve Medicare were available through 2016, performance among the beneficiaries. ASCs that reported data improved for most measures. Between 2012 and 2017, mortality, Quality of care— Providers’ access to capital— Because the number of hospitalization, and 30-day readmission rates declined, ASCs has continued to increase and hospital systems though the proportion of FFS dialysis beneficiaries using and others have significantly incorporated ASCs into the emergency department increased. With regard to their business strategies, access to capital appears to be anemia management, negative cardiovascular outcomes adequate. associated with the use of high levels of erythropoiesis- Medicare payments and providers’ costs— From 2012 stimulating agents declined, and blood transfusions, which to 2016, Medicare payments for ASC services per FFS initially increased under the PPS, have trended downward beneficiary increased by an average annual rate of 3.5 since 2013. Between 2012 and 2017, beneficiaries’ use of percent. By contrast, in 2017, payments for ASC services home dialysis, which is associated with improved patient increased by 7.7 percent. ASCs do not submit data on the satisfaction and quality of life, increased from 9.5 percent cost of services they provide to Medicare beneficiaries. to 11 percent of dialysis beneficiaries. The first-year Therefore, we cannot calculate a Medicare margin as (2016) results of the accountable care organization model we do for other provider types to help assess payment specific to dialysis providers, the ESRD Seamless Care adequacy. Organization model, were positive; for example, there were fewer inpatient admissions for beneficiaries, and all On the basis of these indicators, the Commission 13 organizations in the model produced savings relative to concludes that ASCs can continue to provide Medicare their benchmarks. It is not clear if this trend will continue; beneficiaries with access to ASC services with no update the results for 2017 and 2018 are not yet available. to the payment rates for 2020. In addition, the Commission continues to recommend that the Secretary of Health and Access to capital for Providers’ access to capital— Human Services collect cost data from ASCs without dialysis providers continues to be strong. The number further delay. of facilities, particularly for-profit facilities, continues to increase. Under the dialysis PPS, the two largest dialysis Outpatient dialysis services organizations have grown through acquisitions and Outpatient dialysis services are used to treat the majority mergers with midsized dialysis organizations. of individuals with end-stage renal disease (ESRD). In Between Medicare payments and providers’ costs— 2017, nearly 395,000 beneficiaries with ESRD on dialysis 2016 and 2017 cost per dialysis treatment increased by 2 were covered under FFS Medicare and received dialysis percent, while Medicare payment per treatment increased from approximately 7,000 dialysis facilities. In 2017, by 0.6 percent. We estimate that the aggregate Medicare Medicare expenditures for outpatient dialysis services margin was –1.1 percent in 2017, and the 2019 Medicare were $11.4 billion, a 0.4 percent increase over 2016 margin is projected at –0.4 percent. expenditures. In light of these findings, the Commission recommends Our payment adequacy indicators for outpatient dialysis that for 2020, the Congress update the ESRD PPS base services, described in Chapter 6, are generally positive. rate by the amount determined under current law. Measures of the capacity Beneficiaries’ access to care— and supply of providers, beneficiaries’ ability to obtain Cross-cutting issues in post-acute care care, and changes in the volume of services suggest Post-acute care (PAC) providers offer important payments are adequate. Dialysis facilities appear to have recuperation and rehabilitation services to Medicare the capacity to meet demand. Between 2016 and 2017, beneficiaries. PAC providers include skilled nursing the number of dialysis treatment stations grew faster than facilities (SNFs), home health agencies (HHAs), inpatient the number of FFS dialysis beneficiaries, and the growth rehabilitation facilities (IRFs), and long-term care in the number of FFS dialysis beneficiaries and total hospitals (LTCHs). In 2017, FFS program spending on number of treatments was relatively flat. The 17 percent PAC services totaled $58.5 billion. marginal profit in 2017 suggests that dialysis providers March 2019 | Report to the Congress: Medicare Payment Policy xvii

20 The Commission has previously discussed the challenges a stay in an acute care hospital. In 2018, about 15,000 to increasing the accuracy of Medicare’s payments SNFs furnished 2.3 million Medicare-covered stays to 1.6 and overcoming the shortcomings of the separate FFS million FFS beneficiaries. Medicare FFS spending on SNF payment systems for PAC. Over more than a decade, the services was $28.4 billion in 2017, about 1 percent less Commission has worked extensively on PAC payment than in 2016. Just over 4 percent of beneficiaries used SNF reform, pushing for closer alignment of costs and services. payments and more equitable payments across different As discussed in Chapter 8, most of our payment adequacy types of patients. measures for SNFs are positive. As discussed in Chapter 7, despite some actions by the Access to SNF services Beneficiaries’ access to care— Secretary and the Congress, Medicare’s payments remain remains adequate for most beneficiaries. The number of too high relative to the costs of treating beneficiaries SNFs participating in the Medicare program has been in three of the four settings (SNF, HHA, and IRF). In stable. The vast majority (89 percent) of beneficiaries addition, the current HHA and SNF payment systems live in a county with three or more SNFs or swing bed create inequities across patients with different care needs facilities (rural hospitals with beds that can serve as and the providers that treat them. These overpayments either SNF beds or acute care beds), and less than 1 and misalignments threaten the long-run sustainability percent live in a county without one. Between 2016 of the program and create incentives for providers to and 2017, the median occupancy rate declined slightly treat some types of cases over others. Furthermore, they but remained high (85 percent). Medicare-covered affect the benchmarks for Medicare Advantage plans admissions per FFS beneficiary decreased 2 percent and alternative payment models. However, after years between 2016 and 2017. Lengths of stay also declined of research and recommendations by the Commission, by 2 percent. Both contributed to fewer covered days in the Secretary is poised to make substantial changes 2017 compared with 2016. Lower SNF use reflects the to the payment systems Medicare uses to pay HHAs growing presence of alternative payment models, not and SNFs that will increase the equity of Medicare’s the adequacy of Medicare’s payments. An indicator of payments within each of these settings. These changes whether freestanding SNFs have an incentive to treat more are consistent with longstanding recommendations made Medicare beneficiaries—marginal profit—averaged 19 by the Commission. percent for freestanding facilities in 2017. A uniform payment system for all PAC would increase Quality of care— Since 2011, SNF quality measures the equity of payments across patients and providers in have shown mixed performance. The average rate of all PAC settings, but its implementation is on a longer discharge to the community increased; the average rate of timetable. Until a unified PAC PPS is in place, Medicare readmission during the SNF stay improved; the average must continue to improve its setting-specific payment rate of readmissions after the SNF stay worsened; and the systems. measures of mobility remained the same. Changes in the measures between 2016 and 2017 were similarly mixed. To assess the quality of post-acute care, there has been progress in defining common outcome measures across Because most SNFs are part Providers’ access to capital— PAC providers and establishing value-based purchasing of nursing homes, we examine nursing homes’ access to policies for HHAs (on a demonstration basis) and SNFs. capital. Despite relatively low total margins (a measure However, the Commission is increasingly concerned of the total financial performance across all payers and that trends in some provider-reported quality measures lines of business), lending and investment activities remain raise questions about the accuracy and reliability of this robust. Access to capital was adequate in 2018 and is information. The Commission has work underway to expected to remain so in 2019. Lending wariness reflects examine the accuracy of the patient assessment–based broad changes in post-acute care, not the adequacy of quality measures. Medicare’s payments. Medicare is regarded as a preferred payer of SNF services. Skilled nursing facility services Skilled nursing facilities (SNFs) provide short-term skilled Medicare’s Medicare payments and providers’ costs— nursing and rehabilitation services to beneficiaries after spending in 2017 decreased 1 percent to $28.4 billion. In Executive summary xviii

21 2017, the average Medicare margin for freestanding SNFs Home health care services was 11.2 percent—the 18th year in a row that the average Home health agencies (HHAs) provide services to was above 10 percent. Margins varied greatly across beneficiaries who are homebound and need skilled facilities, reflecting differences in costs and shortcomings nursing or therapy. In 2017, about 3.4 million Medicare in the SNF PPS that favor treating rehabilitation patients beneficiaries received care, and the program spent $17.7 over medically complex patients. billion on home health care services. In that year, almost 12,000 HHAs participated in Medicare. Consistent with our previous years’ recommendations, the Commission recommends that the Secretary proceed As we discuss in Chapter 9, the indicators of payment with his plans to implement a revised SNF PPS. Further, to adequacy for home health care are generally positive. keep the relative costs of stays aligned with payments, the Commission recommends that the relative weights of the Access to home health Beneficiaries’ access to care— case-mix groups be recalibrated annually. care is adequate: Over 98 percent of beneficiaries lived in a ZIP code where an HHA operated in 2017, and 84 To address the high level of Medicare’s payments, the percent lived in a ZIP code with five or more HHAs. The Commission recommends that the Congress eliminate the number of HHAs fell slightly (by 3 percent) in 2017, but fiscal year 2020 update to the Medicare base rates. While this decline follows a long period of growth in prior years. the level of payments indicates a reduction to payments From 2004 to 2016, the number of HHAs increased by is needed to more closely align aggregate payments and 60 percent. The decline in 2017 was concentrated in areas costs, the SNF industry is likely to undergo considerable that experienced sharp increases in supply in prior years. changes as it adjusts to the redesigned PPS. Given From 2002 to 2016, home health utilization increased the impending changes, the Commission will proceed substantially, with the number of episodes rising nearly 60 cautiously in recommending reductions to payments. A percent and the episodes per home health user climbing zero update would begin to align payments with cost while from 1.6 to 1.9 episodes. In 2017, volume dropped 3.1 exerting pressure on providers to keep their cost growth percent, the total number of FFS users also fell slightly, low. and the average number of episodes per home health user declined by 1.4 percent. Episodes not preceded by Medicaid trends a hospitalization accounted for most of the growth since As required by the Patient Protection and Affordable Care 2002, increasing from about half of episodes in 2002 to Act of 2010, we report on Medicaid use and spending two-thirds of episodes in 2017. In 2017, freestanding and non-Medicare (private-payer and Medicaid) margins. HHAs’ marginal profit—that is, the rate at which Medicaid finances most long-term care services provided Medicare payments exceed providers’ marginal cost—was in nursing homes, but also covers the copayments on SNF 17.5 percent, suggesting a significant financial incentive care for low-income Medicare beneficiaries (known as for HHAs to serve Medicare patients. dual-eligible beneficiaries) who stay more than 20 days Quality of care— In 2017, the rate of home health in a SNF. The number of Medicaid-certified facilities has patients who were hospitalized or received treatment in declined slightly since 2013, by less than 1 percent, but the emergency room during an episode did not change remains close to 15,000. CMS reports total FFS spending significantly, while measures of functional status, such on nursing home services declined 1.6 percent between as improvement in walking and transferring, increased. 2016 and 2017 but projects small increases for 2019. However, the functional status measures should be In 2017, the average total margin—reflecting all payers interpreted cautiously because these measures are based (including managed care, Medicaid, Medicare, and on provider-reported data and could be affected by agency private insurers) and all lines of business (such as hospice, coding practices. ancillary services, home health care, and investment Providers’ access to capital— Access to capital is a less income)—was 0.5 percent, down from 2016 (0.7 percent). important indicator of Medicare payment adequacy The average non-Medicare margin (which includes all for home health care because this sector is less capital payers and all lines of business except Medicare FFS SNF intensive than other health care sectors. The major publicly services) was –2.4 percent, the same as in 2016. traded for-profit home health companies had sufficient March 2019 | Report to the Congress: Medicare Payment Policy xix

22 access to capital markets for their credit needs. Several profit under Medicare’s IRF prospective payment system acquisitions to increase capacity and expansion of capacity suggest that capacity remains adequate to meet demand. by publicly traded home health care firms indicate After declining for several years, the number of IRFs adequate access to capital. In 2017, the average all-payer increased in 2014 and continued to grow through 2016, margin for HHAs was 4.5 percent. reaching 1,188 facilities nationwide. In 2017, the number of IRFs declined slightly, to 1,178 facilities. Over time, the In 2017, Medicare payments and providers’ costs— number of hospital-based and nonprofit IRFs has declined, Medicare spending for home health care declined by 1.6 while the number of freestanding and for-profit IRFs percent. However, between 2002 and 2016, spending has increased. In 2017, the average IRF occupancy rate increased by over 88 percent. For more than a decade, remained at 65 percent, indicating that capacity is more payments under the home health PPS have consistently than adequate to meet demand for IRF services. From and substantially exceeded costs. In 2017, Medicare 2016 to 2017, the number of Medicare FFS cases going margins for freestanding agencies averaged 15.2 percent. to IRFs declined 2.7 percent, falling to 380,000 cases The projected margin for 2019 is 16.0 percent. after having experienced small annual growth every year since 2010. The marginal profit, an indicator of whether The high margins of freestanding HHAs have led the IRFs with excess capacity have an incentive to treat more Commission to recommend that the 2020 home health Medicare beneficiaries, was 19.4 percent for hospital- PPS base payment rate be equal to the 2019 level reduced based IRFs and 38.8 percent for freestanding IRFs—a by 5 percent. However, this reduction will likely be very positive indicator of patient access. inadequate to align Medicare payments with providers’ actual costs, and further reductions through rebasing Quality of care— The Commission tracks three broad will likely be necessary. In past years, the Commission categories of IRF quality indicators: risk-adjusted facility- has recommended that payments be rebased in the year level change in patients’ functional and cognitive status following a payment rate reduction, but this year’s during the IRF stay, rates of discharge to the community recommendation is complicated by the changes to home and to skilled nursing facilities, and rates of readmission health payment set for 2020. A rebased payment rate to an acute care hospital. Most measures were steady or should reflect the mix and level of services HHAs provide improved between 2012 and 2017. under the new payment policies because the mix of services and number of visits provided in an episode will Providers’ access to capital— The parent institutions likely change. These data will not be available until mid- of hospital-based IRFs continue to have good access 2021. to capital. The major freestanding IRF chain, which accounted for almost half of freestanding IRFs in 2017 Inpatient rehabilitation facility services and about a quarter of all Medicare IRF discharges, also has good access to capital. We were not able to determine Inpatient rehabilitation facilities (IRFs) provide intensive the ability of other freestanding facilities to raise capital. rehabilitation services to patients after illness, injury, Access to capital in large part depends on total (all-payer) or surgery. Rehabilitation programs are supervised by profitability. In 2017, the all-payer margin was 10.4 rehabilitation physicians and include services such as percent for freestanding IRFs and was 7.0 percent for physical and occupational therapy, rehabilitation nursing, hospitals with IRF units. speech–language pathology, and prosthetic and orthotic services. In 2017, Medicare spent $7.9 billion on IRF The aggregate Medicare payments and providers’ costs— care provided to FFS beneficiaries in about 1,180 IRFs Medicare margin for IRFs has grown steadily since 2009 nationwide. About 340,000 beneficiaries had around and in 2017 stood at 13.8 percent. In 2017, Medicare 380,000 IRF stays. On average, the Medicare FFS margins in freestanding IRFs were 25.5 percent. In 2017, program accounted for 58 percent of IRF discharges. hospital-based IRF margins were comparatively low at 1.5 percent, but one-quarter of hospital-based IRFs had As described in Chapter 10, our indicators of Medicare Medicare margins greater than 11 percent, indicating that payment adequacy for IRFs are positive. many hospitals can manage their IRF units profitably. Beneficiaries’ access to care— Our analysis of IRF supply To the extent that hospital-based IRFs routinely assess and volume of services provided and of IRFs’ marginal their patients as less disabled than do their freestanding Executive summary xx

23 counterparts, their payments—and margins—will be Beneficiaries’ access to care— We have no direct systematically lower. For 2019, we project an aggregate measures of beneficiaries’ access to needed LTCH Medicare margin of 11.6 percent for all IRFs. services. While we consider the capacity and supply of LTCH providers and changes over time in the volume This year, the Commission for the first time examined the of services they furnish, we expect reductions in these financial performance of relatively efficient IRFs. Our metrics since the implementation of the new dual analysis found that relatively efficient IRFs performed payment-rate structure that began in fiscal year 2016. better on quality metrics and had costs 18 percent lower The number of LTCHs began to decrease in 2013, but the than other IRFs. Relatively efficient IRFs were on average decline has been more rapid since the implementation larger and had higher occupancy rates, contributing to of the dual payment-rate structure. We estimate that the greater economies of scale and lower costs. number of LTCHs decreased by 4.1 percent from 2016 to 2017 and by an additional 2.3 percent from 2017 to On the basis of these factors, the Commission 2018. However, the average LTCH occupancy rate was recommends a 5 percent reduction to the IRF payment rate 64 percent in 2017, suggesting that LTCHs have adequate for fiscal year 2020. In addition, the Commission reiterates capacity in the markets they serve. From 2016 to 2017, its March 2016 recommendations that (1) the high-cost the number of LTCH cases decreased by 7.3 percent, outlier pool be expanded to further redistribute payments continuing a four-year trend that began in 2013. The in the IRF payment system and reduce the impact of number of LTCH cases per FFS beneficiary also declined misalignments between IRF payments and costs and that during this period (2016 to 2017) by 7 percent. However, (2) the Secretary conduct focused medical record review from 2016 to 2017, the number of LTCH cases that met of IRFs that have unusual patterns of case mix and coding the criteria per 10,000 FFS beneficiaries increased by 3.6 and conduct other research as necessary to improve the percent. In 2017, marginal profit, an indicator of whether accuracy of payments and protect program integrity. LTCHs with excess capacity have an incentive to admit Medicare patients, averaged about 14 percent across all Long-term care hospital services LTCHs and 16 percent for LTCHs with a high share (85 Long-term care hospitals (LTCHs) provide care to percent or more) of cases meeting the new criteria. beneficiaries who need hospital-level care for relatively extended periods. To qualify as an LTCH for Medicare Quality of care— Consistent with prior years, non-risk- payment, a facility must meet Medicare’s conditions adjusted rates of direct LTCH to acute care hospital of participation for acute care hospitals and, for certain readmission, death in the LTCH, and death within 30 days Medicare patients, have an average length of stay greater of discharge were stable across all LTCH cases. than 25 days. In 2017, Medicare spent $4.5 billion on Providers’ access to capital— LTCHs have begun altering care provided in LTCHs nationwide. About 103,000 their cost structures and referral patterns in response to the FFS beneficiaries had roughly 116,000 LTCH stays. On dual payment-rate structure. This transition, coupled with average, Medicare FFS beneficiaries accounted for about payment reductions to annual updates required by statute two-thirds of LTCHs’ discharges. and moratoriums in effect for most of the past decade, In fiscal year 2016, CMS began implementing a dual have limited opportunities for growth in the near term and payment-rate structure for LTCHs that decreased payment reduced the industry’s need for capital. rates for certain cases not meeting the criteria specified The aggregate Medicare payments and providers’ costs— in the Pathway for SGR Reform Act of 2013. The extent Medicare margin for LTCHs was 3.9 percent across to which LTCHs change their admission patterns to all cases in 2016. In 2017, the first year that all LTCHs admit more cases meeting the criteria (cases that will began transitioning to the dual payment-rate structure, the thus be paid the standard LTCH PPS rate) will ultimately aggregate Medicare margin was –2.2 percent. However, determine the industry’s financial performance under when we consider a cohort of LTCHs with a high share Medicare. In Chapter 11, we focus some analyses on of cases that met the criteria, and thus admission patterns LTCHs with a high share (85 percent or more) of cases consistent with the goals of the dual payment-rate meeting the criteria in 2017, consistent with the goals of structure, the Medicare margin remained positive. Indeed, the dual payment-rate policy. in 2017, LTCHs with 85 percent or more of their Medicare March 2019 | Report to the Congress: Medicare Payment Policy xxi

24 ® cases meeting the criteria had a Medicare margin of Assessment of Healthcare Providers and Systems ® 4.6 percent. We expect continued changes in admission (CAHPS ) survey data for individual providers became patterns and cost structures of LTCHs in response to available for the first time in 2018. Scores on the eight the implementation of the dual payment-rate structure. CAHPS measures were generally high; however, there We project that LTCHs’ aggregate Medicare margin is more variation and potential for improvement with the for facilities with more than 85 percent of Medicare CAHPS measures than with the process measures. discharges meeting the criteria will be 1.2 percent in 2019. Hospices are not as capital Providers’ access to capital— On the basis of these indicators, and in the context of intensive as some other provider types because they do recent changes in payment policy, for fiscal year 2020 not require extensive physical infrastructure. Continued the Commission recommends that the Secretary increase growth in the number of for-profit providers (5 percent the 2019 LTCH payment rate by 2 percent. This update increase in 2017) suggests capital is available to these supports LTCHs in their provision of safe and effective providers. Less is known about access to capital for care for Medicare beneficiaries meeting the criteria for the nonprofit freestanding providers, for which capital may standard LTCH PPS rate. be more limited. Hospital-based and home health–based hospices have access to capital through their parent Hospice services providers. The Medicare hospice benefit covers palliative and support Medicare payments and providers’ costs— The aggregate services for beneficiaries who are terminally ill with a 2016 Medicare margin was 10.9 percent, up from 9.9 life expectancy of six months or less if the illness runs its percent in 2015. The projected Medicare margin is 10.1 normal course. When beneficiaries elect to enroll in the percent in 2019. Medicare hospice benefit, they agree to forgo Medicare coverage for conventional treatment of their terminal Given the margin in the industry and our other positive illness and related conditions. In 2017, nearly 1.5 million payment adequacy indicators, the Commission Medicare beneficiaries (including more than half of recommends that hospice payment rates be reduced by decedents) received hospice services from 4,488 providers, 2 percent in 2020. This recommendation would bring and Medicare hospice expenditures totaled about $17.9 payment rates closer to costs, would lead to savings for billion. beneficiaries and taxpayers, and would be consistent with the Commission’s principle that it is incumbent on As discussed in Chapter 12, the indicators of payment Medicare to maintain financial pressure on providers to adequacy for hospices are positive. constrain costs. Beneficiaries’ access to care— In 2017, hospice use The Medicare Advantage program: Status increased across almost all demographic and beneficiary report groups examined. In 2017, the number of hospice Chapter 13 provides a status report on the Medicare providers increased by about 2.4 percent due to growth in Advantage (MA) program. In 2018, the MA program the number of for-profit hospices, continuing a more than included about 3,100 plan options offered by 185 decade-long trend of substantial market entry by for-profit organizations, enrolled over 20 million beneficiaries providers. In 2017, the proportion of beneficiaries using (33 percent of all Medicare beneficiaries), and paid MA hospice services at the end of life continued to grow, and plans about $233 billion (not including Part D drug plan length of stay among decedents increased. For hospice payments). To monitor program performance, we examine providers, Medicare payments exceeded marginal costs by MA enrollment trends, plan availability for the coming roughly 14 percent in 2016, suggesting that providers have year, and payments for MA plan enrollees relative to an incentive to treat Medicare patients. spending for FFS Medicare beneficiaries. We also provide Quality of care— Limited quality data are available updates on risk adjustment, risk coding practices, and for hospice providers. In 2017, hospices’ performance current quality indicators in MA. on seven quality measures related to processes of care The MA program gives Medicare beneficiaries the option at hospice admission was very high, but most of the of receiving benefits from private plans rather than from measures appear to be topped out. Hospice Consumer Executive summary xxii

25 the traditional FFS Medicare program. The Commission 2019 MA benchmarks (including quality bonuses), bids, strongly supports the inclusion of private plans in the and payments will average 107 percent, 89 percent, and Medicare program; beneficiaries should be able to choose 100 percent of FFS spending, respectively. Adjusting for between the traditional FFS Medicare program and the uncorrected coding intensity differences would increase extra benefits and alternative delivery systems that private the ratio of MA payments to FFS spending by 1 percent plans often provide. Because Medicare pays private plans to 2 percent; hence, MA payments would average about a risk-adjusted per person predetermined rate rather than 101 percent to 102 percent of FFS spending. On average, a per service rate, plans have greater incentives than FFS quality bonuses in 2019 will add 4 percent to the average providers to innovate and use care-management techniques plan’s base benchmark and will add 2.4 percent to plan to deliver more efficient care. payments. Lower benchmarks have led to more competitive bids from plans: Bids have dropped from roughly 100 The Commission has emphasized the importance of percent of FFS before the Patient Protection and Affordable imposing fiscal pressure on all providers of care to Care Act of 2010 to 89 percent of FFS in 2019. improve efficiency and reduce Medicare program costs and beneficiary premiums. For MA, the Commission Risk adjustment and coding intensity— Medicare previously recommended that payments be brought down payments to MA plans are enrollee specific, based on from prior levels, which were generally higher than FFS, a plan’s payment rate and an enrollee’s risk score. Risk and be set so that the payment system is neutral and scores account for differences in expected medical does not favor either MA or the traditional FFS program. expenditures and are based in part on diagnoses that Legislation has reduced the inequity in Medicare spending providers code. Most claims in FFS Medicare are paid between MA and FFS nationally, even as plans have using procedure codes, which offer little incentive for received increased payments because of higher risk coding providers to record more diagnosis codes than necessary and quality bonus rules. As a result, over the past few to justify ordering a procedure. In contrast, MA plans have years, plan bids and payments have come down in relation a financial incentive to ensure that their providers record to FFS spending while MA enrollment continues to grow. all possible diagnoses: Higher enrollee risk scores result in The pressure of lower benchmarks has led to improved higher payments to the plan. efficiencies and more competitive bids that enable MA Our updated analysis for 2017 shows that higher plans to continue to increase enrollment by offering diagnosis coding intensity resulted in MA risk scores benefits that beneficiaries find attractive. that were 7 percent higher than scores for similar FFS Enrollment— Between November 2017 and November beneficiaries. By law, CMS makes a minimum across- 2018, enrollment in MA plans grew by 8 percent—or 1.6 the-board adjustment to MA risk scores to make them million enrollees—to 20.5 million enrollees. Among plan more consistent with FFS coding. In 2017, the adjustment types, HMOs continued to enroll the most beneficiaries. reduced MA risk scores by 5.66 percent, leaving MA risk Special needs plan enrollment grew by 13 percent, and scores and payments about 1 percent to 2 percent higher employer group enrollment grew by 12 percent. than they would have been if MA enrollees had been treated in FFS Medicare. The adjustment for 2019 will be Plan availability— Access to MA plans remains high in 5.9 percent. The Commission previously recommended 2019; 99 percent of Medicare beneficiaries have access that CMS change the way diagnoses are collected for use to an MA plan and 97 percent have an HMO or local in risk adjustment and calculate a new coding adjustment preferred provider organization (PPO) plan operating in that improves equity across plans and eliminates the their county of residence. Regional PPOs are available impact of differences in MA and FFS coding intensity. to 74 percent of beneficiaries. Plan availability continues to grow; the average beneficiary in 2019 has 23 available Quality in MA— Chapter 13 summarizes our concerns plans. Compared with 2007, MA enrollment in 2018 was with the MA star rating system that is the basis for plan more heavily concentrated. The top 10 MA organizations bonuses and public reporting of plan quality. Because (ranked by enrollment) had 74 percent of total enrollment of the way the system has been implemented, it is not in 2018, compared with 61 percent in 2007. possible to accurately compare quality among plans or track changes in MA quality over time, and plans Using the 2019 plan bid data, before Plan payments— can receive quality bonus payments when they are not adjusting fully for coding intensity, we estimate that March 2019 | Report to the Congress: Medicare Payment Policy xxiii

26 warranted. In addition, we continue to lack information benefits have remained around $30 per month for many that would permit a comparison of MA quality with the years. More than 8 in 10 Part D enrollees report they are quality of care in FFS. satisfied with the program. MA star ratings are determined at the contract level, with However, changes to Part D’s coverage gap and many quality results determined based on a small sample manufacturer discounts combined with the expanding role of medical records. Because contracts can cover wide of high-cost medicines may be eroding plans’ incentives geographic areas and because of the sample-size issue, for cost control. Over time, as more enrollees have reached contract-level reporting does not capture geographic the catastrophic phase of the benefit, a growing share variation in quality and is unable to adequately identify of Medicare’s payments to plans have taken the form of variation among subgroups of the Medicare population. cost-based reinsurance subsidies rather than capitated Using encounter data as the source of quality metrics in payments. In addition, beginning in 2019, brand-drug MA and moving to market areas as the reporting unit manufacturers must provide a 70 percent discount in the would address this concern. Moving to encounter-based coverage gap (an increase from 50 percent). This change metrics in MA would also permit comparisons between correspondingly decreases what plan sponsors must cover MA and claims-based metrics in FFS. in benefits and likely weakens sponsors’ incentives to manage spending. A separate concern is that Part D’s LIS MA plans receive quality bonuses if they have a star rating may lead to plan and beneficiary incentives that increase of at least 4 stars on a 5-star scale. An issue of concern to program costs. the Commission has been the practice of plan sponsors consolidating contracts so that nonbonus contracts acquire Policymakers are taking steps to give plan sponsors new the star rating of the “surviving” contract. At the end flexibilities to manage drug spending. For example, of 2018, about 550,000 beneficiaries were moved from CMS now allows for certain midyear formulary changes nonbonus plans to bonus-level plans through contract without prior approval. However, measures to increase consolidations, and the sponsors will receive unwarranted the financial risk that sponsors bear (such as those bonus payments for those enrollees. This concern has recommended by the Commission in 2016) are also been partly addressed through recent legislation, which needed so that plan sponsors have greater incentive to use provides that, starting at the end of 2019, the star rating the new management tools and keep Part D financially for consolidated contracts will be based on an enrollment- sustainable for beneficiaries and taxpayers. weighted average of the results of each contract that is Enrollment in 2018 and benefit offerings for 2019— In being consolidated. 2018, 73.3 percent of Medicare beneficiaries were enrolled in Part D plans. An additional 2.5 percent obtained drug The Medicare prescription drug program coverage through employer-sponsored plans that received (Part D): Status report Medicare’s retiree drug subsidy. The remaining 24.2 Chapter 14 provides a status report on Part D plans. In percent were divided roughly equally between those who 2018, Part D plans were the primary source of outpatient had creditable drug coverage from other sources and those prescription drug coverage for 43.9 million Medicare with no coverage or coverage less generous than Part D. beneficiaries. Medicare subsidizes about three-quarters of the cost of basic benefits. Part D also includes a low- Between 2007 and 2018, enrollment grew faster in MA– income subsidy (LIS) that provides assistance with Prescription Drug plans (MA–PDs) compared with stand- premiums and cost sharing to 12.5 million individuals alone prescription drug plans (PDPs). In 2018, 42 percent with low income and assets. In 2017, Part D expenditures of enrollees were in MA–PDs compared with 30 percent totaled $93.9 billion. Enrollees paid $14.0 billion of that in 2007. Over the same period, the share of enrollees who amount in plan premiums, in addition to what they paid in received the LIS fell from 39 percent to 28 percent. cost sharing. For 2019, beneficiaries continue to have a broad choice Part D has been a success in many respects. It has of plans. Sponsors are offering 15 percent more PDPs improved beneficiaries’ access to prescription drugs. and 21 percent more MA–PDs than in 2018. MA–PDs Generic drugs now account for nearly 90 percent of the continue to be more likely than PDPs to offer enhanced prescriptions filled. Enrollees’ average premiums for basic Executive summary xxiv

27 benefits. In 2019, 215 premium-free PDPs are available Redesigning Medicare’s hospital quality to enrollees who receive the LIS. With the exception of 1 incentive programs region (Florida), all regions have at least 3 and as many as The quality of hospital care has improved in recent years, 10 PDPs for LIS enrollees at no premium. in part due to Medicare’s four hospital quality incentive programs: the Hospital Inpatient Quality Reporting Part D program costs— Between 2007 and 2017, Program, Hospital Readmissions Reduction Program Medicare payments to Part D plans and employers (HRRP), Hospital-Acquired Condition Reduction Program increased from about $46 billion to about $80 billion (HACRP), and hospital value-based purchasing (VBP) (average annual growth of 5.6 percent). Medicare’s program. Nevertheless, the Commission has several reinsurance (which covers 80 percent of enrollees’ concerns about the design of these programs, which we spending in the catastrophic phase of the benefit) grew at discuss in Chapter 15. an average annual rate of nearly 17 percent and continues to be the fastest growing component of program spending. The Commission asserts that quality measurement Also in this period, the portion of the benefits paid to plans should be patient oriented, encourage coordination, through capitated direct subsidies fell from 55 percent and promote delivery system change. Based on our to 21 percent, while the portion paid through Medicare’s principles for quality measurement, in our June 2018 reinsurance grew from 25 percent to 54 percent. Enrollees report to the Congress we examined the potential to who incur spending high enough to reach the catastrophic create a single, outcome-focused, quality-based payment phase of the benefit (high-cost enrollees) continued program for hospitals—the hospital value incentive to drive Part D spending. In 2016, high-cost enrollees program (HVIP). Initially, the HVIP can incorporate accounted for 58 percent of all Part D spending, up from existing quality measure domains such as readmissions, about 40 percent before 2011. Among high-cost enrollees, mortality, spending, patient experience, and hospital- nearly all growth in spending was due to increases in acquired conditions (or infection rates). The HVIP uses the average price per prescription filled. In 2016, nearly clear, prospectively set performance standards to translate 360,000 enrollees filled a prescription that was so hospital performance on these quality measures to a expensive that their cost-sharing for a single fill would reward or penalty. have been sufficient to meet their out-of-pocket threshold, Adjusting measure results for social risk factors can up from just 33,000 in 2010. mask disparities in clinical performance. Therefore, In 2019, the average star rating among Quality in Part D— the HVIP that we modeled accounts for differences Part D plans decreased somewhat for PDPs and remained in providers’ patient populations by incorporating a about the same for MA–PDs. However, the trend among peer-grouping methodology, in which quality-based MA–PD sponsors of consolidating contracts to achieve payments are distributed to hospitals separated into 10 higher star ratings leads us to question the validity of peer groups, defined by the share of fully dual-eligible MA–PD ratings and the comparison between PDPs and beneficiaries treated (using full Medicaid eligibility as MA–PDs. It is not clear that current quality metrics a proxy for income). The HVIP redistributes pools of help beneficiaries make informed choices among their dollars to hospitals in the peer groups based on their plan options. In the past, the Commission has expressed quality performance. The pools of dollars are funded by concerns about the effectiveness of plans’ medication a payment withhold from all hospitals in the peer group therapy management (MTM) programs to improve the (e.g., 5 percent) and a portion of the current-law hospital quality of pharmaceutical care due to the lack of financial payment update. Under the Commission’s HVIP model, incentives for sponsors of stand-alone PDPs. In 2017, the use of peer grouping of hospitals that serve different CMS implemented the enhanced MTM program that populations makes payment adjustments more equitable rewards PDPs for reducing medical spending. Initial compared with the existing quality payment programs. results indicate that half of the participating plans Consistent with the Commission’s principles, the HVIP successfully reduced medical spending by 2 percent or links payment to quality of care to reward hospitals for more, qualifying them for a higher premium subsidy in efficiently providing high-quality care to beneficiaries. 2019. We are encouraged by the initial results. Accordingly, the Commission recommends that the March 2019 | Report to the Congress: Medicare Payment Policy xxv

28 Congress replace Medicare’s current hospital quality providers to choose opioids over non-opioid alternatives. programs with this new HVIP that includes a small set The IPPS and OPPS payment bundles create a financial of population-based outcome, patient experience, and incentive for hospitals to be cost conscious in selecting value measures; scores all hospitals based on the same goods and services. Medicare’s quality measurement absolute and prospectively set performance targets; and and reporting programs, along with providers’ clinical accounts for differences in patients’ social risk factors by expertise and professionalism, are designed to balance distributing payment adjustments through peer grouping. this financial incentive. Ideally, these balanced incentives As we discuss in Chapter 3, the Commission recommends result in high-quality outcomes at the best prices for that payments in the HVIP be increased by the difference beneficiaries and other taxpayers. However, if opioids between the Commission’s update recommendation for were systematically cheaper than non-opioid alternatives, acute care hospitals and the amount specified in current providers might be more inclined to opt for them, law. The increased payment in the HVIP will better especially if doing so did not affect performance on quality reward hospitals providing higher quality care. In addition, measures. We analyzed publicly available prices for opioid eliminating the existing penalty-only programs (i.e., the and non-opioid alternatives commonly used in the hospital HRRP and HACRP) would have the effect of removing setting and found that both opioids and non-opioids are about $1 billion in penalties that hospitals currently pay available at a range of list prices, including expensive each year. and inexpensive options for both. Thus, there is no clear indication that Medicare’s IPPS or OPPS discriminates Mandated report: Opioids and alternatives against non-opioids. Indeed, hospitals that select more in hospital settings—Payments, incentives, expensive options for clinical reasons have tools available and Medicare data to them, such as reducing length of stay, to partially or Chapter 16 is the Commission’s response to the fully offset these costs. mandate in the Substance Use-Disorder Prevention that Our study is not intended to be an assessment of the Promotes Opioid Recovery and Treatment for Patients clinical appropriateness of the use of opioids versus non- and Communities Act of 2018 for the Commission to opioid alternatives. Clinicians’ decisions about which describe how Medicare pays for both opioid and non- analgesic drugs to prescribe are based on a multitude of opioid pain management treatments in hospital inpatient patient-specific factors. Furthermore, we recognize that and outpatient settings, incentives under the inpatient and there are incentives in addition to financial ones that may outpatient prospective payment systems for prescribing have an even greater influence on clinicians’ choice of opioids and non-opioids, and how opioid use is monitored pain treatments, such as effects on patient experience, through Medicare claims data. length of stay, need for additional nursing services, Medicare uses bundled payments to pay for pain and—most important—the management of potential risks management drugs and services in both the inpatient and clinical efficacy. However, these motivations are not and outpatient settings. Bundled payments are applied unique to the Medicare IPPS and OPPS, so to comply with differently in the two settings. The inpatient prospective the mandate’s due date, we focused on the extent to which payment system (IPPS) assigns stays to categories these payment systems introduce financial incentives. (Medicare severity–diagnosis related groups) based on CMS monitors opioid use through claims and other data in patients’ conditions and sets payment bundles that reflect the Part D program. The tools used in the Part D program the average costs of providing all goods and services include the Medicare Part D Overutilization Monitoring supplied during the stay. The outpatient prospective System, which ensures that Part D plan sponsors payment system (OPPS) also groups services into implement the opioid overutilization policy effectively; categories (ambulatory payment classifications), but on quality measures to track trends in opioid overuse across the basis of clinical and cost similarity, and sets payment the Medicare Part D program and to drive performance bundles to cover the costs of providing integral goods improvement among plan sponsors; and the publicly and services and items along with the primary service. available Medicare Part D opioid prescribing mapping Additional goods and services are paid separately or are tool. not paid under the OPPS. Medicare does not operate similar tracking programs in Some observers have questioned whether Medicare’s Part A or Part B. Given concerns about the opioid crisis, hospital payment systems create financial incentives for Executive summary xxvi

29 policymakers may wish to direct CMS to track opioid use a Part A and Part B opioid tracking program: (1) require in hospital inpatient and outpatient settings. If Medicare prescription drug event–type reporting, (2) include all were to undertake an opioid monitoring program in Part A pain management drugs in Part A and Part B claims, and and Part B, there are structural differences from Part D that (3) link Part D opioid use to hospitals responsible for its would require adaptation of CMS’s current monitoring initiation. ■ program. There are at least three options for implementing March 2019 | Report to the Congress: Medicare Payment Policy xxvii

30

31 CHAPTER 1 Context for Medicare payment policy

32

33 CHAPTER 1 Context for Medicare payment policy Chapter summary In this chapter Part of the Commission’s mandate is to consider the effect of its National health care spending • recommendations on the federal budget and view Medicare in the context Medicare spending • of the broader health care system. To help meet this mandate, this chapter examines health care spending growth—for the nation at large and Medicare Medicare’s financing • in particular—and considers its effect on federal and state budgets as well challenge as the budgets of individuals and families. The chapter also reviews recent Health care spending • mortality and morbidity trends; profiles the health status of the next generation consumes growing shares of of Medicare beneficiaries; and reviews evidence of inefficient health care state and family budgets spending, structural features of the Medicare program that contribute to Recent trends in life • inefficient spending, and the Commission’s approach to combating those expectancy, morbidity, and challenges. mortality • The relationship between In 2017, total national health care spending was $3.5 trillion, or 17.9 Medicare spending and percent of gross domestic product (GDP) according to the National Health quality Expenditure Accounts (NHEA) official estimates of total health care spending • Baby boomers will make in the United States (Centers for Medicare & Medicaid Services 2018a). up the next generation of Private health insurance spending was $1.2 trillion, or 6.1 percent of GDP. Medicare beneficiaries Medicare spending was $705.9 billion, or 3.6 percent of GDP. • Inefficient spending suggests Health care spending growth has fluctuated recently, first with several years of Medicare could spend less without compromising care, historic lows, followed by a period of accelerated growth, and most recently but not without challenges with a return to modest growth. For decades—from 1975 to 2009—total Report to the Congress: Medicare Payment Policy March 2019 | 3

34 health care spending and Medicare spending grew robustly, annually averaging 9.0 percent and 10.6 percent, respectively. Then, from 2009 to 2013, growth in total health care spending and Medicare spending slowed to average annual rates of 3.7 percent and 4.3 percent, respectively. The causes of the system-wide slowdown are still a matter of speculation. A variety of factors could have contributed—weak economic conditions, payment and delivery system reforms, lower Medicare payment rates for most types of providers as mandated by the Patient Protection and Affordable Care Act of 2010 (PPACA), and the increased use of generic drugs as top-selling brand drugs lost patent protection (Boards of Trustees 2016, Centers for Medicare & Medicaid Services 2015, Cutler and Sahni 2013, Holahan et al. 2017). However, spending increased from 2013 to 2015. Medicare actuaries estimate that national health care spending grew at an average annual rate of 5.5 percent and that Medicare spending grew at an average annual rate of 4.9 percent. The increase in the national health care spending growth rate was largely due to the continued effects of coverage expansions for health insurance that commenced in 2014 under PPACA; higher growth in spending for private health insurance (driven largely by price growth, hospital care, and physician and clinical services); and the rapid growth in retail prescription drug spending. The aging of the baby-boom generation will continue to have a profound impact both on the Medicare program and taxpayers, who primarily finance it. Over the next 15 years, as Medicare enrollment surges, the number of taxpaying workers per beneficiary is projected to decline. By 2029 (when most boomers will have aged into Medicare), the Medicare Trustees project there will be just 2.4 workers for each Medicare beneficiary, down from 4.6 around the time of the program’s inception and 3.0 in 2018. Those demographics create a financing challenge not only for the Medicare program but also for the entire federal budget. By 2041, under federal tax and spending policies specified in current law, Medicare spending combined with spending on other major health care programs, Social Security, and net interest on the national debt will exceed total projected federal revenues and will thus either increase federal deficits and debt further or crowd out spending on all other national priorities. The growth in health care spending also affects state budgets and the budgets of individuals and families. States pay for a significant portion of Medicaid spending (funded jointly by states and the federal government for health care services provided to state residents with low incomes). Under PPACA, the Medicaid population is expanding; however, under current law, the federal government Context for Medicare payment policy 4

35 will pay for most of the costs associated with the expansion. Increases in private insurance premiums have outpaced the growth of individual and family incomes over the past decade, and out-of-pocket costs for Medicare beneficiaries have grown faster than Social Security benefits. Some health care spending is inefficient. For Medicare, if such spending could be identified and eliminated, the efficiencies achieved could result in improved beneficiary health, greater fiscal sustainability for the program, and reduced federal budget pressures. Certain structural features of the Medicare program pose challenges for targeting inefficient spending; however, the Commission has made multiple recommendations to the Congress and the Secretary that, if implemented, have the potential to improve the quality of care and move the Medicare program toward paying for value. ■ March 2019 | Report to the Congress: Medicare Payment Policy 5

36

37 that share slowed. From 2009 through 2013, total health Introduction care, private health insurance, and Medicare spending as a share of GDP remained relatively constant. But beginning The Medicare program lies at the junction between the in 2014, spending as a share of GDP for all three began national health care system as a whole and the federal rising again (Centers for Medicare & Medicaid Services government. For this reason, this chapter reviews the 2017). following key areas to help explain the Medicare payment The recent slowdown in the rate of health care spending policies discussed in the rest of this report: growth has not been fully explained. Contributing factors • national health care spending and Medicare spending; could include weak economic conditions, payment and delivery system reforms, lower Medicare payment rates • the impact of health care spending on federal and state for most types of providers as mandated by the Patient budgets; Protection and Affordable Care Act of 2010 (PPACA), and the increased use of generic drugs as top-selling brand effects of health care spending on individuals and • drugs lost patent protection (Boards of Trustees 2016, families; Centers for Medicare & Medicaid Services 2015, Cutler 1 • recent trends in life expectancy, morbidity, and and Sahni 2013, Holahan et al. 2017). mortality; Medicare actuaries estimate that spending growth from the impact of Medicare spending on the quality of • 2016 to 2017 slowed compared with 2015 to 2016, both health care; for private health insurance and for Medicare (Martin et al. 2018). From 2016 to 2017, spending growth both for • the next generation of Medicare beneficiaries; and private health insurance and Medicare was 4.2 percent. Yet from 2015 to 2016, spending growth for private health evidence of inefficient health care spending. • insurance was 6.2 percent and for Medicare was 4.3 This chapter also reviews the challenges that Medicare percent. This recent increase followed a brief period of in particular faces and the Commission’s principles high growth from 2013 through 2015. From 2013 through for constructing recommendations to address those 2015, growth for private health insurance averaged 6.3 challenges. percent per year and averaged 4.9 percent per year for Medicare. By 2017, total health care spending accounted for 17.9 percent of GDP. Overall, the slower growth from 2016 to 2017 was due largely to the lower use and National health care spending intensity of medical goods and services, including hospital and clinician services and retail prescription drugs. Spending growth Over the next decade, Medicare actuaries project that The relationship between health care spending growth growth in national health expenditures will be driven and the nation’s economic growth serves as a gauge by increases in prices for medical goods and services, for assessing spending trends. For decades, health care including drugs, and growth in the volume and intensity spending rose as a share of gross domestic product (GDP). of services. In addition, enrollment will continue to shift That general trend was true both for private health insurance from private health insurance to Medicare because of spending and Medicare (Figure 1-1, p. 8). From 1975 to the continued aging of the baby-boom generation into 2009, health care spending as a share of GDP more than eligibility. Thus, growth rates for health care spending will doubled, from 7.9 percent to 17.3 percent ($133 billion average 5.5 percent annually, outpacing average growth to $2.5 trillion, respectively). Private health insurance in GDP by 1.0 percentage point (Centers for Medicare spending as a share of GDP more than tripled over that & Medicaid Services 2018b). By 2026, total health care period, from 1.8 percent to 5.8 percent ($31 billion to spending as a share of GDP will grow to 19.7 percent $833 billion). Medicare spending as a share of GDP also (Cuckler et al. 2018). In that year, private health insurance more than tripled over that period, from 1.0 percent to 3.5 spending and Medicare spending are projected to reach 6.2 percent ($16 billion to $499 billion, respectively). But in percent and 4.7 percent of GDP, respectively (Centers for the recent past (from 2009 to 2013), the rate of increase in Medicare & Medicaid Services 2018b). March 2019 | Report to the Congress: Medicare Payment Policy 7

38 FIGURE Health care spending... 1-1 FIGURE Health care spending has grown as a share of GDP 1–1 25 Total health care spending Historical Projected Private health insurance spending $5.7T Medicare spending 20 Medicaid spending 15 10 $133B Share of GDP (in percent) $1.8T $1.4T 5 $31B $996B $16B $13B 0 1995 2025 1975 1985 2015 2005 Note: GDP (gross domestic product), B (billion), T (trillion). First projected year is 2018. Beginning in 2014, private health insurance spending includes federal subsidies for both premiums and cost sharing for the health care exchanges created by the Patient Protection and Affordable Care Act of 2010. Source: MedPAC analysis of National Health Expenditure Accounts from CMS, historical data released December 2018, projected data released February 2018. Note: Note and Source are in InDesign. Source: Notes about this graph: (Figure 1-2). During this period, out-of-pocket spending • Data is in the datasheet. Make updates in the datasheet. Personal health care spending (e.g., cost sharing, deductibles, and health care services not To better understand who is paying for health care, we • I deleted the years from the x-axis and put in my own. covered by insurance) as a share of total personal health examine a subset of total national health expenditures, • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. care spending declined from 31 percent to 12 percent, namely personal health care spending—all medical goods while the shares accounted for by private health insurance, • The dashed line looked ok here, so I didn’t hand draw it. and services provided for an individual’s treatment. In Medicare, and Medicaid all increased. At the same time, 2017, personal health care spending (which excludes • I can’t delete the legend, so I’ll just have to crop it out in InDesign. Medicare has remained the single largest purchaser of spending on government public health activities (e.g., • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph health care in the United States (Centers for Medicare & epidemiological surveillance and disease prevention 2 default when you change the data. Medicaid Services 2018a). programs); administration of private and public health • Use paragraph styles (and object styles) to format. insurance; and investments in medical research, Despite the decline in the share of health care spending equipment, and structures) accounted for 85 percent paid directly out of pocket by individuals and the increase of total health care spending (Centers for Medicare & in the share of health care spending paid by private and Medicaid Services 2018a). public insurance, people generally have not experienced real declines in the share of health care costs they pay. Over the past four decades, total personal health care One reason is that in the commonly defined health care spending increased from $0.1 trillion to $3.0 trillion Context for Medicare payment policy 8

39 FIGURE Medicare population FIGURE 1-2 Out-of-pocket spending as a share of personal health care spending 1–2 declined, while the share of spending by payers—private, Medicare, and Medicaid—increased, 1977 and 2017 2017 1977 Total = $3.0 trillion Total = $0.1 trillion 4% 5% 15% CHIP, DoD, and VA DoD and VA 12% Medicare ($126B) ($7B) 22% Out of pocket ($22B) Medicare ($366B) ($660B) 31% 11% Out of pocket Medicaid ($45B) ($17B) 12% 18% Other third-party Medicaid payers ($521B) 35% ($17B) Private health insurance 9% 27% ($1,040B) Other third-party Private health payers insurance ($248B) ($39B) DoD (Department of Defense), VA (Department of Veterans Affairs), B (billion), CHIP (Children’s Health Insurance Program). “Personal health care” is a subset Note: Note: Note and Source are in InDesign. of national health expenditures. It includes spending for all medical goods and services that are provided for the treatment of an individual and excludes other spending, such as government administration, the net cost of health insurance, public health, and investment. Spending is in nominal dollars. “Out-of-pocket” Source: spending includes cost sharing for both privately and publicly insured individuals. Only the portion of premiums used to pay for benefits are included in the shares of each program (e.g., Medicare and private insurance) rather than in the out-of-pocket category. “Other third-party payers and programs” includes work-site health care, other private revenues, Indian Health Service, workers’ compensation, general assistance, maternal and child health, vocational rehabilitation, other federal programs such as the Substance Abuse and Mental Health Services Administration, other state and local programs, and school health. Totals may not sum to 100 percent due to rounding. MedPAC analysis of National Health Expenditure Accounts historical data from CMS, released December 2018. Source: spending categories, the premiums people pay (which have Some people have coverage from more than one source. grown over time) are not included in the out-of-pocket For example, about 10 million people are dually enrolled (OOP) category but, rather, in the private health insurance in both Medicare and Medicaid (Boards of Trustees 2018). and Medicare categories. Second, people receive lower Medicaid pays for either a portion or all of the Medicare salaries and reduced benefits in exchange for employer- premium and OOP health care expenses for those sponsored health insurance (Baicker and Chandra 2006, enrollees who qualify for dual enrollment based on limited Gruber 2000, Milusheva and Burtless 2012). income and resources. Enrollees in public health insurance programs may also have private health insurance. For CMS actuaries estimate that, in 2017, Medicare covered example, Medicare beneficiaries typically also have about 57 million people and Medicaid covered about supplemental insurance sold by private companies to pay 73 million people (Martin et al. 2018). Private health some of the health care costs that Medicare does not cover, insurance covered 197 million people, and 30 million such as copayments, coinsurance, and deductibles. people were uninsured. March 2019 | Report to the Congress: Medicare Payment Policy 9

40 FIGURE FIGURE 1–3 Hospital care and physician and clinical services accounted for the largest XXXXX 1-3 shares of personal health care spending in 1977 and 2017 1977 2017 Total = $0.1 trillion Total = $3.0 trillion 6% 7% 8% 7% Nursing Nursing Other Other care care health care health care and CCR 2% and CCR 2% ($247B) ($11B) facilities Durable facilities Durable medical ($166B) medical 39% ($10B) equipment 46% equipment Hospital ($3B) Hospital ($54B) ($1,143B) ($67B) 6% 11% Retail Retail prescription drugs prescription drugs ($9B) ($333B) 1% 3% Home health Home health care care 8% ($1B) ($97B) Other professional 8% ($226B) 23% Other 23% Physician and professional Physician and clinical services ($12B) clinical services ($33B) ($694B) Note: CCR (continuing care retirement), B (billion). “Personal health care” is a subset of national health expenditures. It includes spending for all medical goods and Note: Note and Source are in InDesign. services that are provided for the treatment of an individual and excludes other spending, such as government administration, the net cost of health insurance, public health, and investment. “Other health care” includes expenditures on nondurable medical products and other health, residential, and personal care. “Other Source: professional” includes expenditures on dental and other professional services. “Nursing care facilities” includes nursing care facilities and continuing care retirement communities. “Hospital” includes all services provided in hospitals to patients: room and board, ancillary services such as operating room fees, inpatient and outpatient care, services of resident physicians, inpatient pharmacy, hospital-based nursing home care, hospital-based home health care, and fees for any other services billed by the hospital, such as hospice. “Physician and clinical services” includes services provided in physician offices, outpatient care centers, and in hospitals, if the physician bills independently for those services, plus the portion of medical laboratories services that are billed independently by the laboratories. Components may not sum to totals due to rounding. MedPAC analysis of National Health Expenditure Accounts historical data from CMS, released December 2018. Source: In 2017 as well as in 1977, the largest shares of personal percent, or $97 billion) (see text box on prescription drug health care spending were for hospital care and physician spending trends). Between 1977 and 2017, the share of and clinical services (Figure 1-3). In 2017, hospital care spending on hospital care declined (from 46 percent to 39 accounted for 39 percent of spending ($1,143 billion), and percent), while the share of spending for retail prescription physician and clinical services accounted for 23 percent drugs increased (from 6 percent to 11 percent) (Centers for ($694 billion). Smaller shares went to spending on retail Medicare & Medicaid Services 2018a). prescription drugs (11 percent, or $333 billion), nursing In 2017, Medicare accounted for 22 percent of spending care and continuing care retirement (CCR) facilities (6 for personal health care services (Figure 1-2, p. 9), but percent, or $166 billion), and home health care services (3 Context for Medicare payment policy 10

41 Prescription drug spending trends (Martin et al. 2018). However, retail drugs made pending on prescription drugs has increased up a greater share of all Medicare spending—14 significantly compared with other sectors, nearly percent. Medicare’s retail spending in 2016 reflects doubling as a share of personal health care S Part D program spending and prescription drugs billed spending, from 6 percent in 1977 to 11 percent in 2017 separately under Part B. (see Figure 1-3). The Commission developed estimates of Medicare CMS’s Office of the Actuary projects that national drug spending that include not only retail drug spending on prescription drugs will grow faster than spending, which is the typical metric used to describe spending on other health care goods and services at an the magnitude of drug spending, but also spending average annual rate of 6.3 percent from 2017 to 2026 for drugs and pharmacy services used as inputs at (Cuckler et al. 2018). The Office explains that “this health care facilities, which is not typically included trend primarily reflects faster anticipated growth in in measures of drug spending. These estimates are drug prices, which is attributable to a larger share of based on Medicare cost reports, Medicare claims, drug spending being accounted for by specialty drugs and estimates of program spending from the Trustees over the coming decade.” The American Academy of reports. Ultimately, the estimates are all in terms of Actuaries attributes prescription drugs spending growth what the Medicare program paid. In comparison with to both price and utilization, specifically driven by Medicare’s retail spending, the Commission estimates “delays in introducing generics, higher cost inflation total drug and pharmacy services, that, in 2016, in the United States for pharmaceuticals relative to including those provided at health care facilities, other nations, and the compensation of numerous accounted for 23 percent of total Medicare spending stakeholders throughout the pharmacy supply chain” (excluding beneficiary cost sharing). That total share (Hanna and Uccello 2018). was 20 percent in 2007. ■ In 2016, across all payers, retail drug spending made up 10 percent of all national health expenditures its share varied by type of service, with a slightly higher Medicare spending share of spending on hospital care (25 percent) and retail prescription drugs (30 percent) and a much higher Medicare spending can be divided into three program share of spending on home health services (40 percent) components: the traditional fee-for-service (FFS) program, (Figure 1-4, p. 12). Medicare’s share of spending on the Medicare Advantage (MA) program, and the Part D nursing care facilities was smaller than Medicaid’s share prescription drug program. because Medicare’s benefit pays for skilled nursing or rehabilitation services only, whereas Medicaid pays for • Medicare’s traditional FFS program. In FFS, custodial care (assistance with activities of daily living) Medicare pays health care providers directly for health provided in nursing homes for people with limited income care goods and services furnished to FFS Medicare and assets. Medicare’s share of spending varies for other beneficiaries at prices set through legislation and service categories included in personal health care that regulation. In 2017, Medicare spent $394 billion, or are not shown in Figure 1-4, namely, other professional $10,206 per beneficiary in traditional FFS (Boards of services; dental services; other health, residential, and Trustees 2018). personal care; and other nondurable medical equipment. March 2019 | Report to the Congress: Medicare Payment Policy 11

42 FIGURE FIGURE Medicare’s share... 1-4 Medicare’s share of spending on personal health care varied by type of service, 2017 1–4 100 Other 90 24 Medicaid and 80 all CHIP 47 Medicare 70 58 60 71 66 60 36 50 40 30 10 17 30 spending (in percent) 11 14 20 40 Share of personal health care 30 25 23 23 10 15 0 Nursing care Durable medical Home Physician Hospital Retail health prescription and and equipment ($1,143B) clinical services ($97B) drugs CCR facilities ($54B) ($333B) ($166B) ($694B) CHIP (Children’s Health Insurance Program), B (billion), CCR (continuing care retirement). “Personal health care” is a subset of national health expenditures. Note: It includes spending for all medical goods and services that are provided for the treatment of an individual and excludes other spending such as government administration, the net cost of health insurance, public health, and investment. “Hospital” includes all services provided in hospitals to patients: room and board, ancillary services such as operating room fees, inpatient and outpatient care, services of resident physicians, inpatient pharmacy, hospital-based nursing home care, hospital-based home health care, and fees for any other services billed by the hospital, such as hospice. “Physician and clinical services” includes services provided in physician offices, outpatient care centers, and in hospitals, if the physician bills independently for those services, plus the portion of medical laboratories services that are billed independently by the laboratories. “Nursing care and CCR facilities” includes freestanding facilities primarily engaged in providing inpatient nursing, rehabilitative, and continuous personal care services to persons requiring nursing care and continuing-care retirement communities with on-site nursing care facilities. “Other” includes private health insurance, out-of-pocket spending, and other private and public spending. Other service categories included in personal health care that are not shown here include other professional services; dental services; other health, residential, and personal care; and other nondurable medical equipment. MedPAC analysis of National Health Expenditure Accounts from CMS, historical data released December 2018. Source: MA program. Beneficiaries can choose, as an • insurance policies from private stand-alone drug plans alternative to FFS, to enroll in MA, which consists of or MA prescription drug plans. Medicare heavily private health plans that receive capitated payments subsidizes the premiums established by those plans. (or per enrollee payments) for providing health care In 2017, Medicare spent $80 billion, net of Part D coverage for enrollees. MA plans pay health care premiums (mostly premiums paid by beneficiaries), or providers for health care goods and services furnished $1,797 per beneficiary in Part D. to their enrollees at prices negotiated between the Growth in per beneficiary spending tends to differ across plans and providers. In 2017, Medicare spent $209 the three program components. From 2009 to 2013, billion, or $10,571 per beneficiary in MA. growth was fairly slow across all three (Figure 1-5). More • Medicare Part D prescription drug program. mixed trends emerged between 2013 and 2017. The lower Through Part D, beneficiaries can obtain subsidized growth rates were generally because of decreased use of prescription drug coverage by voluntarily purchasing health care services and restrained payment rate increases. Note: Note and Source are in InDesign. Source: Context for Medicare payment policy 12

43 FIGURE FIGURE Health care spending... Growth in per beneficiary Medicare spending was slow between 1–5 1-X 2009 and 2013 and mixed between 2013 and 2017 12 MA Percent change Part D $209B total spending 10 Percent change FFS $10,571 per beneficiary Percent change MA 8 6 FFS 4 $394B total spending $10,206 per beneficiary 2 0 Part D –2 Percent change in spending per beneficiary $80B total spending $1,797 per beneficiary –4 2015 2014 2011 2016 2017 2010 2009 2012 2013 FFS (fee-for-service), MA (Medicare Advantage), B (billion). Spending is on an incurred basis. Part D spending excludes total premiums paid to Part D plans by Note: enrollees. We calculate per beneficary spending by dividing total spending for each category reported in the Trustees report by the appropriate enrollment number (i.e., for Part A, Part B, or Part D) reported in the Trustees report. Source: MedPAC analysis of data from the 2018 annual report of the Boards of Trustees of the Medicare trust funds. Note: Note and Source are in InDesign. Source: From 2013 to 2017, FFS per beneficiary spending growth payments and plans’ increased coding of beneficiaries’ averaged 1.5 percent annually. PPACA lowered payment medical conditions (payments to MA plans are higher Notes about this graph: rate updates in FFS for many types of providers (other when beneficiaries have more medical conditions, all other • Data is in the datasheet. Make updates in the datasheet. than physicians) beginning in 2011. However, beginning in things being equal). 2014, FFS spending gradually grew because of an increase • I deleted the years from the x-axis and put in my own. Part D per beneficiary spending growth has fluctuated in per beneficiary spending on a wide range of outpatient • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. the most of the three program components over the services, including services received in hospital outpatient past decade. However, from 2010 to 2013, average per • The dashed line looked ok here, so I didn’t hand draw it. departments and physician services. beneficiary spending was somewhat constant, growing • I can’t delete the legend, so I’ll just have to crop it out in InDesign. 3 From 2013 to 2017, MA per beneficiary spending growth from $1,605 to $1,626 per year. The low growth for those • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph averaged 1.6 percent annually. Historically, Medicare years was in part due to the increase in low-priced generic default when you change the data. generally has spent more for a beneficiary enrolled in MA drugs on the market and plans’ efforts to encourage than if that same beneficiary had been enrolled in FFS. beneficiaries to use generics and other low-priced drugs. • Use paragraph styles (and object styles) to format. To bring payments more in line with FFS, PPACA began However, in both 2014 and 2015, per beneficiary spending lowering payments to plans in 2011. MA’s growth rate growth in excess of 6 percent caused Part D spending to would therefore have been lower, but the PPACA payment spike to $1,868 per beneficiary. Increased spending on reductions were offset somewhat by quality bonus March 2019 | Report to the Congress: Medicare Payment Policy 13

44 FIGURE FIGURE Cumulative change... Per beneficiary FFS spending growth remained high in some settings 1–6 1-X despite 2009–2013 slowdown in growth of health care spending, 2008–2017 12 9.9 10 7.7 7.6 7.3 8 6.7 5.2 6 4.1 3.9 3.9 3.2 4 1.9 1.9 1.1 2 1.3 0.8 0.2 0.2 0.0 0.0 0 –0.2 –2 –2.4 –4 –3 .3 Average annual percent change for: Average annual percent change –5.1 –6 2008–2009 2009–2013 2013–2017 –6.2 –8 Labs performed in Inpatient Physician Outpatient Home Hospice Skilled Durable nursing medical fee schedule health hospital and physician offices hospital facility and independent lab services equipment laboratories FFS (fee-for-service). We calculate per beneficary spending by dividing total spending for each category reported in the Trustees report by the appropriate Note: enrollment number (i.e., for Part A, Part B, or Part D) reported in the Trustees report. Outpatient hospital services and outpatient lab services are combined in the figure because a large portion of outpatient laboratory services were bundled into the outpatient prospective payment system effective January 1, 2014. MedPAC analysis of data from the 2018 annual report of the Boards of Trustees of the Medicare trust funds. Source: the period from 2008 to 2009 to 0.2 percent in the period high-priced specialty drugs to treat hepatitis C mainly from 2009 to 2017. Even the fastest growing categories accounts for this jump. After the high spending of 2015, experienced some reductions. For example, the average the surge of hepatitis C drug spending tapered off while annual per beneficiary spending growth in outpatient Part D enrollment continued to grow, which contributed hospital and lab services was lower between 2009 and to per Part D enrollee spending declining by 1.9 percent 2013 (6.7 percent) than between 2008 and 2009 (7.6 per year to $1,797 by 2017 (Boards of Trustees 2018, percent) but bounced back to 7.7 percent between 2013 Boards of Trustees 2017). The Medicare Trustees project and 2017 annually, in part because of shifts in site of care the annual growth in per beneficiary Part D spending from both the inpatient hospital setting and physician from 2018 to 2026 to remain higher than growth in other 4 offices to the outpatient hospital setting. categories of spending, averaging 3.9 percent per year As a reference (Boards of Trustees 2018). point, average annual growth in GDP between 2008 and 2017 was about 3.1 percent (data not shown). Figure 1-6 provides a more detailed look at FFS spending growth over the past decade. Generally, all settings Despite the recent slowing of growth rates, cumulative experienced a slowdown in per beneficiary spending growth in per beneficiary FFS spending over the past growth; however, the impact was not uniform. For decade has increased in almost all settings and increased example, for inpatient hospital care, the average annual substantially in some settings. Per beneficiary spending Note: Note and Source are in InDesign. growth in per beneficiary spending fell from 1.1 percent in on outpatient hospital and lab services, hospice, and Source: Context for Medicare payment policy 14

45 FIGURE FIGURE Title here... 1-XX Cost of employer-sponsored commercial insurance has grown 1–7 more than twice as fast as Medicare costs, 2008–2017 60 Employer-sponsored HMO premium growth for a single person Employer-sponsored PPO premium growth for a single person 50 Medicare per capita FFS spending growth (Part A, Part B, Part D) 40 30 Percent 20 10 0 2016 2008 2009 2010 2011 2012 2013 2014 2017 2015 HMO (health maintenance organization), PPO (preferred provider organization), FFS (fee-for-service). Medicare spending is reported including the effects of the Note: sequester, which reduced program spending for most benefits by 2 percent beginning in 2013. Source: Employer-sponsored premium data from Kaiser Family Foundation surveys, 2007 through 2017. Medicare spending figures from MedPAC analysis of data from the 2018 annual report of the Boards of Trustees of the Medicare trust funds. Note: Note and Source are in InDesign. Source: labs performed in physician offices and independent Comparison of private sector and Medicare laboratories all grew faster than per capita GDP. In spending trends contrast, during this time, per beneficiary spending on From 2010 to 2016, per capita spending on health care in Notes about this graph: durable medical equipment fell by an average of 4.4 the private sector grew (Centers for Medicare & Medicaid • Data is in the datasheet. Make updates in the datasheet. percent per year. That decline was primarily due to the Services 2018a). Increased prices were largely responsible phasing in of a competitive bidding program for durable for spending growth, which occurred despite a decline • I deleted the years from the x-axis and put in my own. medical equipment in which suppliers submit bids to in service use (Health Care Cost Institute 2018, Health • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. provide services to beneficiaries. Care Cost Institute 2016, Health Care Cost Institute • The dashed line looked ok here, so I didn’t hand draw it. 2015). One key driver of the private sector’s higher prices Prior Commission reports have explored the relationship was provider market power (Baker et al. 2014a, Baker et • I can’t delete the legend, so I’ll just have to crop it out in InDesign. between inpatient, outpatient, and physician services and al. 2014b, Cooper et al. 2018, Gaynor and Town 2012, found that growth in outpatient services in part reflects • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph Medicare Payment Advisory Commission 2017, Robinson hospitals purchasing freestanding physician practices and default when you change the data. and Miller 2014, Scheffler et al. 2018). Hospitals and billing these services through the higher paying hospital • Use paragraph styles (and object styles) to format. physician groups have increasingly consolidated, in outpatient prospective payment system (Martin et al. 2018, part to gain leverage over insurers in negotiating higher Medicare Payment Advisory Commission 2015, Medicare payment rates. For the private sector, that consolidation Payment Advisory Commission 2014b, Medicare Payment contributed to per capita spending growth from 2010 to Advisory Commission 2013, Medicare Payment Advisory 2016 of 3.7 percent annually. By comparison, over that Commission 2012). | Report to the Congress: Medicare Payment Policy March 2019 15

46 FIGURE FIGURE Cumulative change... 1-X Despite recent slowdown in per beneficiary spending growth, 1–8 total Medicare spending growth rate is projected to rise 14 11.7 Spending per beneficiary 12 Medicare enrollment 10 9.0 7.6 7.5 8 7.2 9.6 6 7.0 4.9 4.5 4.8 4 5.6 1.5 2 2.9 Average annual change (in percent) 2.6 2.7 1.9 1.9 1.5 0 1980s Trustees CBO 1990s 2000s 2010–2017 Projected, 2018–2026 Historical Note: CBO (Congressional Budget Office). Components of average annual changes may not sum to totals due to rounding. Trustees numbers are reported by calendar year; CBO numbers are reported by fiscal year. Source: 2018 annual report of the Boards of Trustees of the Medicare trust funds and CBO’s Medicare April 2018 baseline. same period, Medicare spending per beneficiary increased per capita spending on Part A, Part B, and Part D. Over the by 1.4 percent annually (Centers for Medicare & Medicaid period from 2008 to 2017, combined Medicare per capita Services 2018a). This difference suggests that the costs grew by about 16 percent. If FFS Medicare spending effectiveness of the tools private plans have to constrain had followed growth in commercial pricing, Medicare service use has been counteracted by the higher prices costs would have grown substantially more. plans pay relative to the lower Medicare payment rates Regulators and researchers have noted concerns about under its administered pricing system. increased consolidations and their effect on prices. In 2015, the number of hospital mergers increased 18 On average, since 2008, commercial insurance prices percent from the prior year and 70 percent from 2010 have grown faster than Medicare’s prices (Health (Ellison 2016). Consolidation of clinician practices Care Cost Institute 2016, Medicare Payment Advisory has also increased; a study of available data found a 47 Commission 2017). The faster growth in provider prices percent jump from 2014 (Irving Levin Associates Inc. from 2008 to 2017 contributed to HMO premiums for 2016). The American Medical Association’s survey of a single person growing by 48 percent and preferred physicians indicates that, over time, physicians have provider organization premiums for a single person by 45 shifted from solo and small practices to larger practices percent (Figure 1-7, p. 15). (Kane 2015). The Government Accountability Office To compare employer-sponsored plans’ premium growth (GAO) found that, between 2007 and 2013, the number with Medicare cost growth, we examined per capita Note: Note and Source are in InDesign. of physicians in “vertically consolidated” practices— spending for beneficiaries with FFS Medicare, including hospital-acquired physician practices, physicians hired as Source: Context for Medicare payment policy 16

47 salaried employees, or both—nearly doubled (Government The Commission is concerned that these market Accountability Office 2015). In addition, the Federal concentration effects will lead to higher Medicare Trade Commission observed that “providers increasingly spending if commercial prices are “imported” into pursue alternatives to traditional mergers such as Medicare. The Commission has tried to counteract these affiliation arrangements, joint ventures, and partnerships, effects by recommending restrained payment updates and all of which could also have significant implications for by recommending site-neutral payments (paying the same competition” (Federal Trade Commission 2016). Increased for a service regardless of the setting of care). Medicare consolidation has an inflationary effect on prices paid beneficiaries have robust access to hospital and physician in the private sector. A recent study found that disparity services in most markets. And with respect to hospital in hospital prices within regions is the primary driver of services, given the low occupancy rates and the marginal variation in health care spending for the privately insured profits of taking a Medicare patient, access to care is (Cooper et al. 2015). The study shows that hospitals that unlikely to be of concern in the near term (Medicare face fewer competitors have substantially higher prices; Payment Advisory Commission 2017). hospital prices in monopoly markets are more than 15 Over time, private sector trends can influence Medicare percent higher than those in areas with four or more trends. If the private sector is unable to constrain price competitors. It also found that, where hospitals face only growth, the profitability of caring for commercially one competitor, prices are over 6 percent higher; where insured patients will increase relative to the profitability they face two, almost 5 percent higher. of caring for Medicare beneficiaries. Eventually, the In recent work on the effect of provider consolidation difference between commercial rates and Medicare rates on private prices and the pressure that has created for will grow so large that more hospitals would have an Medicare to increase FFS payment rates (Medicare incentive to focus primarily on patients with commercial Payment Advisory Commission 2017), the Commission insurance, which will exert pressure on the Medicare presented the following key findings: program to increase its payment rates. Thus, in the long term, Medicare beneficiaries’ access to care may in part • Markets with greater physician practice consolidation depend on commercial payers restraining rates paid to have had greater increases in physician prices. hospitals (Medicare Payment Advisory Commission 2009, Stensland et al. 2010, White and Wu 2014). • Commercial insurers pay small independent physician practices at rates similar to Medicare for standard Medicare spending projections office visits. However, physicians in large practices What do these current trends portend for Medicare? The and hospital-affiliated practices (who have stronger growth in Medicare’s per beneficiary spending has fallen market power) receive higher rates from insurers for from average annual rates of 9.6 percent in the 1980s those visits. and 5.6 percent and 7.0 percent in the 1990s and 2000s Commercial insurers also pay higher rates to hospitals • (respectively) to 1.5 percent over the past seven years with greater market power. Gaynor and colleagues (Figure 1-8). report that “mergers between rival hospitals are likely For the next 10 years, the Trustees and the Congressional to raise the price of inpatient care and these effects Budget Office (CBO) project that growth in per are larger in concentrated markets. The estimated beneficiary spending will be higher than the recent lows magnitudes are heterogeneous and differ across but lower than the historical highs, with an average annual market settings, hospitals, and insurers” (Gaynor et al. growth rate of almost 5 percent (Boards of Trustees 2018, 2014). Congressional Budget Office 2018b). • Commercial prices vary widely by individual hospital At the same time, the aging of the baby-boom generation and individual insurer. On average, commercial prices is continuing to boost enrollment. Since 2010, the are about 50 percent higher than average hospital costs enrollment growth rate rose from about 2 percent per and are often far more than 50 percent above Medicare year historically to almost 3 percent and is projected to payment rates (Congressional Budget Office 2016a, continue growing faster than historical rates throughout Cooper et al. 2015, Health Care Cost Institute 2014, the next decade. So, despite the slowdown in spending Medicare Payment Advisory Commission 2014a, Selden et al. 2015). March 2019 | Report to the Congress: Medicare Payment Policy 17

48 FIGURE FIGURE Health care spending... 1-X Trustees and CBO project Medicare annual spending 1–9 to exceed $1 trillion by fiscal year 2022 1,600 Historical Projected 1,400 1,200 1,000 800 600 Dollars (in billions) 400 Trustees 200 CBO 0 2026 2018 2016 2014 2012 2020 2022 2010 2008 2006 2024 Fiscal year CBO (Congressional Budget Office). Note: Source: 2018 annual report of the Boards of Trustees of the Medicare trust funds and CBO’s Medicare April 2018 baseline. Note: Note and Source are in InDesign. Source: are deposited into the general fund of the Treasury. The per beneficiary (relative to historical standards), growth number of workers per Medicare beneficiary has already in total spending over the next decade is projected by the Notes about this graph: declined from about 4.6 around the program’s inception Trustees and CBO to average about 7.5 percent annually, • Data is in the datasheet. Make updates in the datasheet. to 3.0 in 2018 (Figure 1-10). Over the next dozen years, which outpaces the projected average annual GDP growth as Medicare enrollment surges, the number of workers of about 4 percent. At those rates, Medicare annual • I deleted the years from the x-axis and put in my own. per beneficiary is projected to decline further. By 2029, spending would rise from $707 billion in fiscal year 2017 • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. the Medicare Trustees project just 2.4 workers for each to $1 trillion by fiscal year 2022 under either projection 5 Medicare beneficiary. (Figure 1-9) (Boards of Trustees 2018, Congressional • The dashed line looked ok here, so I didn’t hand draw it. Budget Office 2018a). • I can’t delete the legend, so I’ll just have to crop it out in InDesign. These demographics create a financing challenge for the Medicare program. Since payroll tax revenues are not • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph growing as fast as Part A spending, the Trustees project default when you change the data. Medicare’s financing challenge that Medicare’s Hospital Insurance (HI) Trust Fund will • Use paragraph styles (and object styles) to format. become depleted and unable to pay its bills in full by The aging of the baby-boom generation will have a 2026—three years earlier than predicted in the 2017 profound impact both on the Medicare program and report—but that date does not tell the whole story (Boards on the taxpayers who support it. Workers pay for the of Trustees 2018). The HI Trust Fund covers less than half Medicare program through payroll taxes and taxes that of Medicare spending (42 percent in 2017), and that share Context for Medicare payment policy 18

49 FIGURE FIGURE Title here... X-X Medicare enrollment is rising while number of workers per HI beneficiary is declining 1–10 Figure 1-10b. Workers per HI beneficiary Figure 1-10a. Medicare HI enrollment 100 5 Number Historical Projected Historical Projected 4 80 60 3 40 2 Workers per HI beneficiary 20 1 Medicare HI enrollment (in millions) 0 0 2020 1990 2010 2020 2030 2040 1980 1970 2040 2030 2000 2010 2000 1990 1980 1970 Note: HI (Hospital Insurance). Hospital Insurance is also known as Medicare Part A. 2018 annual report by the Boards of Trustees of the Medicare trust funds. Source: Note: Note and Source are in InDesign. If the projections reflected such payment is projected to fall to 39 percent by 2024 (Figure 1-11, p. reductions, then any imbalances between 20). The Supplementary Medical Insurance (SMI) Trust Source: payments and revenues would be automatically Fund covers the remainder. The HI Trust Fund pays for eliminated, and the [Trustees] report would not Medicare Part A services, such as inpatient hospital stays, Notes about this graph: serve its essential purpose, which is to inform skilled nursing facilities, and hospice, and is largely (87 policymakers and the public about the size of any percent in 2017) funded through a dedicated payroll tax • Data is in the datasheet. Make updates in the datasheet. 6 trust fund deficits that would need to be resolved (i.e., a tax on wage earnings). • I deleted the years from the x-axis and put in my own. to avert program insolvency. To date, lawmakers To keep the HI Trust Fund solvent over the next 25 years, • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. have never allowed the assets of the Medicare the Trustees estimate that either the payroll tax would HI Trust Fund to become depleted (Boards of • The dashed line looked ok here, so I didn’t hand draw it. need to be increased immediately by 24 percent, rising Trustees 2018). • I can’t delete the legend, so I’ll just have to crop it out in InDesign. from its current rate of 2.90 percent to 3.61 percent, or The rest of Medicare benefit spending is covered by SMI. Part A spending would need to be reduced immediately • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph 7 It covers services under Part B (physician services and by 16 percent (Boards of Trustees 2018). (For projection default when you change the data. other ambulatory care received in hospital outpatient periods of 50 years and 75 years, see Table 1-1, p. 20.) • Use paragraph styles (and object styles) to format. departments) and Part D (prescription drug coverage). SMI Under current law, once the HI Trust Fund is depleted, is a trust fund in name only; it is not funded exclusively payments to providers would be reduced to levels that through dedicated taxes like the HI Trust Fund is. could be covered by incoming tax and premium revenues. Specifically, Part B and Part D are financed by premiums However, the Trustees note that: March 2019 | Report to the Congress: Medicare Payment Policy 19

50 FIGURE Health care spending... FIGURE 1-X The HI Trust Fund covers a declining share of total Medicare spending 1–11 60 Historical Projected 50 50 48 47 47 46 46 44 43 42 42 42 41 41 40 40 40 39 39 39 39 40 30 20 HI Trust Fund spending share of 10 total Medicare spending (in percent) 0 2022 2020 2018 2016 2014 2012 2010 2008 2027 2026 2024 Note: HI (Hospital Insurance). Hospital Insurance is also known as Medicare Part A. The rest of Medicare spending is covered by the Supplementary Medical Insurance Trust Fund, which comprises Part B and Part D. Source: 2018 annual report of the Boards of Trustees of the Medicare trust funds. paid by beneficiaries (covering 25 percent of spending) as SMI spending rises, premiums and transfers from the and general tax revenues plus federal borrowing (covering nation’s Treasury to the Medicare program also grow, Note: Note and Source are in InDesign. 75 percent of spending), which are reset each year to increasing deficits, the debt, and the strain on household Source: 8 match expected Part B and Part D spending. budgets both of workers and retirees, and—assuming no other policy or legislative interventions—reducing Since premiums and transfers are set to grow at the same Notes about this graph: the resources available to make investments that expand rate as Part B and Part D spending, the SMI Trust Fund future economic output (e.g., investments in education, • Data is in the datasheet. Make updates in the datasheet. is expected to remain solvent by construction. However, transportation, and research and development). • I deleted the years from the x-axis and put in my own. • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. • The dashed line looked ok here, so I didn’t hand draw it. TABLE Increase in payroll tax or decrease in HI spending needed to • I can’t delete the legend, so I’ll just have to crop it out in InDesign. 1–1 maintain HI Trust Fund solvency for specific time periods • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph Or decrease HI spending by: Increase 2.9 percent payroll tax by: To maintain HI Trust Fund solvency for: default when you change the data. • Use paragraph styles (and object styles) to format. 24% 16% 25 years (2018–2042) 28 50 years (2018–2067) 18 75 years (2018–2092) 28 17 HI (Hospital Insurance). Hospital Insurance is also known as Medicare Part A. Note: MedPAC summary of 2018 annual report of the Boards of Trustees of the Medicare trust funds. Source: Context for Medicare payment policy 20

51 FIGURE Health care spending... FIGURE 1-14 General revenue is paying for a growing share of Medicare spending 1–12 7 Total Medicare spending Historical Projected 6 Part A deficit 5 General revenue transfers 4 State transfers and drug fees 3 Premiums Share of GDP (in percent) 2 Tax on benefits 1 Payroll taxes 0 2076 1976 1986 1996 2006 2016 2026 2036 2046 2056 1966 2066 GDP (gross domestic product). These projections are based on the Trustees intermediate set of assumptions. “Tax on benefits” refers to the portion of income Note: taxes that higher income individuals pay on Social Security benefits that is designated for Medicare. “State transfers” (often called the Part D “clawback”) refers to payments from the states to Medicare, required by the Medicare Prescription Drug, Improvement, and Modernization Act of 2003, for assuming primary Note: Note and Source are in InDesign. responsibility for prescription drug spending. “Drug fees” refers to the fee imposed by the Patient Protection and Affordable Care Act of 2010 on manufacturers and importers of brand-name prescription drugs. These fees are deposited in the Part B account of the Supplementary Medical Insurance Trust Fund. Source: 2018 annual report of the Boards of Trustees of the Medicare trust funds. Source: Notes about this graph: • Data is in the datasheet. Make updates in the datasheet. • I deleted the years from the x-axis and put in my own. current law, payments to Part A providers would be For a more complete financial picture, consider the reduced to levels that could be covered by incoming tax combined spending and sources of income from the • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. and premium revenues when the HI Trust Fund becomes two trust funds. The top line of Figure 1-12 depicts • The dashed line looked ok here, so I didn’t hand draw it. depleted. Thus, as Medicare actuaries and others have total Medicare spending as a share of GDP; the layers • I can’t delete the legend, so I’ll just have to crop it out in InDesign. observed, total Medicare spending would be shifted down below the line represent sources of Medicare income. from the total projected spending by an amount equal Medicare’s three primary sources of income are payroll • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph to the Part A deficit (Aaron 2015, Spitalnic 2016). As taxes, premiums paid by beneficiaries, and general revenue default when you change the data. noted by the actuaries, if the projections reflected such transfers. The white space below the total Medicare • Use paragraph styles (and object styles) to format. payment reductions, any imbalances between payments spending line in Figure 1-12 represents the Part A deficit and revenues would be automatically eliminated. To date, created when payroll taxes fall short of Part A spending. lawmakers have never allowed the assets of the Medicare Figure 1-12 reflects projections in the Medicare Trustees HI Trust Fund to become depleted (Boards of Trustees report, which are based on current law with the exception 2018). of disregarding payment reductions that would result from the projected depletion of the HI Trust Fund. Under March 2019 | Report to the Congress: Medicare Payment Policy 21

52 FIGURE Health care spending... FIGURE 1-14 Spending on Medicare, other major health programs, Social Security, 1–13 and net interest is projected to exceed total federal revenues by 2041 30 Year 2041 Total federal spending 25 All other federal spending 20 Total federal revenues Net interest 15 Social Security Percent of GDP 10 Medicaid, CHIP, and exchange subsidies 5 Medicare 0 2048 2018 2038 2028 Note: GDP (gross domestic product), CHIP (Children’s Health Insurance Program). (published June 2018) from the Congressional Budget Office. The 2018 Long-Term Budget Outlook Source: Note: Note and Source are in InDesign. Source: Notes about this graph: Undeniably, the Part A deficit is a financing challenge, but must be covered by federal borrowing. For most years • Data is in the datasheet. Make updates in the datasheet. so too is the large and growing share of Medicare spending over the past several decades, the federal government • I deleted the years from the x-axis and put in my own. funded through general revenues. General revenues has spent more than it collects in revenues, increasing account for 43 percent of Medicare funding today and, the federal debt to levels not seen since World War II. • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. under current law, are projected to grow to 48 percent by Federal revenues have remained relatively constant even • The dashed line looked ok here, so I didn’t hand draw it. 2030; notably, in this context, general revenues include though the federal government has taken responsibility both general tax revenue as well as federal borrowing for a broader array of services (e.g., the Children’s Health • I can’t delete the legend, so I’ll just have to crop it out in InDesign. since, with few exceptions, federal spending has exceeded Insurance Program). • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph federal revenues since the Great Depression. default when you change the data. The layers below the top line in Figure 1-13 depict federal To understand why the growing reliance on general spending by program. Under current law, Medicare • Use paragraph styles (and object styles) to format. revenues presents a financing challenge, consider the spending is projected to rise from 2.9 percent of our situation from the perspective of the federal budget. The economy in 2018 to about 6 percent of our economy in 9 line at the top of Figure 1-13 represents total federal 2048 (Congressional Budget Office 2018a). In fact— spending as a share of GDP; the line below spending assuming no other policy or legislative interventions— represents total federal revenues. The difference between spending on Medicare, Medicaid, the other major health these two lines represents the budget deficit, which programs, Social Security, and net interest payments Context for Medicare payment policy 22

53 Under baseline assumptions, which reflect current law, are projected to reach almost 20 percent of the nation’s CBO projects the debt will reach 90 percent of GDP economy by 2041 and, by themselves, will exceed total 10 in 2024 and 152 percent of GDP in about 30 years (or federal revenues. by 2048). However, the CBO baseline assumes that per Moreover, the projection assumes that federal revenues beneficiary spending for Medicare and Medicaid will will rise above 19 percent of GDP, above the historical increase more slowly in the future than it has during the average of 17 percent of GDP. The increase in revenues is past several decades. On the one hand, if per beneficiary projected to occur mainly because income is projected to spending growth were 1 percentage point higher than that grow more rapidly than inflation, pushing more income of the baseline, the federal debt would be 206 percent into higher inflation-indexed tax brackets over time. of GDP by 2048. On the other hand, if per beneficiary However, if federal revenues continue at their historical spending growth were 1 percentage point lower, the average of 17 percent of GDP, spending on these major federal debt would be 110 percent of GDP by 2048. programs and net interest payments would exceed total federal revenues even sooner. The trends shown in Figure 1-13 reflect CBO’s budget Health care spending consumes growing projections based on the Tax Cuts and Jobs Act of shares of state and family budgets 2017. According to CBO, the Act will increase the total projected deficit over the 2018 to 2028 period by about Part of the Commission’s mandate is to view Medicare in $1.9 trillion, primarily because of reduced federal revenues the context of the broader health care system. This section (Congressional Budget Office 2018b). A temporary examines the effect of health care spending on state spending bill waived the 2010 “pay-as-you-go” law budgets and the budgets of individuals and families. States or PAYGO requirement that would have triggered an bear a significant share of Medicaid and other health care automatic spending cut to Medicare. However, reduced costs, so rising health care spending also has implications revenues and an increased deficit will intensify pressure for state budgets. For individuals and families, increases on policymakers to slow the growth of Medicare and other in premiums and cost sharing have negated real income federal spending. growth in the past decade. Likewise, premiums and cost sharing for Medicare beneficiaries are projected to grow The Tax Cuts and Jobs Act temporarily reduced individual faster than Social Security benefits, which make up a income taxes beginning in 2018. Thus, as Medicare significant share of many beneficiaries’ income. (and other federal) spending continues to grow, federal revenues are projected to be roughly flat over the next few Health care spending and state budgets years relative to GDP, averaging 16.9 percent from 2018 to States and the federal government jointly finance 2025. Revenues are projected to increase in 2026 because Medicaid, a program that pays for health care services most of the provisions of the Tax Cuts and Jobs Act that provided to people with low incomes. In fiscal year directly affect the individual income tax rate are set to 2013, before the coverage expansions made by PPACA, expire at the end of calendar year 2025. Subsequently, monthly enrollment in Medicaid averaged almost 60 revenues are projected to continue to rise relative to million people, and total spending was $455.6 billion, with GDP, although still at a lower rate than spending growth the states paying 42 percent on average and the federal (Congressional Budget Office 2018a). government paying the remainder (Centers for Medicare With their reliance on general tax dollars and federal & Medicaid Services 2016). Medicaid spending accounted deficit spending, Medicare and the other major health for an estimated 19.3 percent of state expenditures in that care programs have a substantial effect on the federal year (Centers for Medicare & Medicaid Services 2014). debt. Debt equaled 35 percent of GDP at the end of 2007, PPACA gave states the option to expand Medicaid when the economy entered the last recession (Figure 1-14, coverage—beginning in 2014—to non-elderly individuals p. 24). In part because of the recession, the debt soared, with total family income of less than 138 percent of the reaching 78 percent of GDP in 2018—a higher share than federal poverty threshold. States received full federal at any point in U.S. history, except briefly around World financing to cover this expansion population in 2014, War II. phasing down to 90 percent federal financing by 2020. March 2019 | Report to the Congress: Medicare Payment Policy 23

54 FIGURE Health care spending... 1-XX FIGURE Health care spending growth impacts future debt levels 1–14 250 Historical Projected 200 150 100 Higher growth rate Debt as a share of GDP (in percent) 50 Baseline Lower growth rate 0 2045 2000 2005 2010 2015 2020 2025 2030 2035 2040 GDP (gross domestic product). The higher growth rate of per beneficiary spending on Medicare and Medicaid is 0.75 percentage point per year higher than under Note: the baseline assumptions; the lower growth rate is 0.75 percentage point per year lower than under the baseline assumptions. Note: Note and Source are in InDesign. The 2018 Long-Term Budget Outlook Source: (published June 2018) from the Congressional Budget Office. Source: Notes about this graph: • Data is in the datasheet. Make updates in the datasheet. CMS actuaries estimate that, in fiscal year 2015, monthly policy represented a significant increase in payments • I deleted the years from the x-axis and put in my own. to providers since Medicaid primary care FFS payment enrollment in Medicaid increased to cover about 70 rates averaged 59 percent of Medicare fee levels in 2012. million people, and total spending increased to reach • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. $552.3 billion (Centers for Medicare & Medicaid The federal government incurred 100 percent of the cost • The dashed line looked ok here, so I didn’t hand draw it. of the payment increase. Federal spending is expected Services 2016). Because the federal government paid • I can’t delete the legend, so I’ll just have to crop it out in InDesign. to reach about $12 billion. (The actual amount is not yet for 100 percent of the costs of newly eligible enrollees, known because states have up to two years to submit the states’ share of all Medicaid expenditures in 2015 • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph claims for federal reimbursement.) Even though the decreased to 37 percent. Government actuaries project default when you change the data. federal subsidies expired at the end of 2014, 16 states and that the states’ share will remain lower than 40 percent • Use paragraph styles (and object styles) to format. the District of Columbia are continuing to pay enhanced over the next 10 years as more states expand coverage (the states’ share is projected to range between 37 percent rates (Tollen 2015). and 39 percent from 2016 to 2025). A provision also established under PPACA authority PPACA also increased the payment amount primary care allows state demonstrations for beneficiaries dually providers received for seeing Medicaid patients in 2013 eligible for Medicare and Medicaid. Under a financial and 2014 so that it equaled Medicare’s payment. This alignment initiative, CMS has approved 14 demonstrations Context for Medicare payment policy 24

55 FIGURE Medicare margins... FIGURE X-X Growth in health care spending and premiums outpaced 1–15 growth in household income, 2007–2017 70,000 Median household income 60,000 Per capita personal health care expenditures Average premium 50,000 for individual coverage Average premium for family coverage 40,000 $61,372 Dollars 30,000 $50,233 20,000 10,000 $18,764 $12,106 $9,106 $6,690 $6,375 $4,479 0 2017 2007 Note: Household income, health expenditures, and premiums are all measured in nominal dollars. Average premiums for individual and family coverage are for employer- sponsored health insurance and include contributions from workers and employers. Source: MedPAC analysis of Census Bureau, Current Population Survey, Annual Social and Economic Supplements 2018; National Health Expenditure Accounts from CMS 2018; and Kaiser Family Foundation and Health Research & Educational Trust 2018 survey of employer health benefits. Note: Note and Source are in InDesign. Source: in 13 states, and 12 demonstrations are still in operation. those covered by employer-sponsored health insurance, Most demonstrations are scheduled to last for five to seven an increase in premiums results in lower wage growth years, but some could be extended to last longer. About because, through wage reductions, employers offset their 440,000 dual eligibles are currently enrolled in what is increased costs of providing health insurance to their Notes about this graph: one of the largest demonstration projects that CMS has employees (Baicker and Chandra 2006, Gruber 2000). • Data is in the datasheet. Make updates in the datasheet. ever conducted related to dual-eligible beneficiaries. Most As health care spending increases, an increasing share • WATCH FOR GLITCHY RESETS WHEN YOU UPDATE DATA!!!! demonstrations (11 of 14) are testing a “capitated” model, of income from individuals and families is transferred which use health plans known as Medicare-Medicaid to insurers, hospitals, physicians, and other providers of • The column totals were added manually. Plans to provide all Medicare benefits and all or most health care services. • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. Medicaid benefits to dual-eligible individuals (Medicare In the past decade, per capita health care spending and Payment Advisory Commission 2018). • I can’t delete the legend, so I’ll just have to crop it out in InDesign. premiums have grown much more rapidly than median • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph household incomes and thus account for a greater share Health care spending and individual and default when you change the data. of income (Figure 1-15). In 2007, per capita personal family budgets health care spending was $6,375, accounting for 13 • Use paragraph styles (and object styles) to format. For individuals and families, growth in health care percent of median household income, which was $50,233. spending has meant higher health insurance premiums • Data was from: R:\Groups\MGA\data book 2007\data book 2007 chp1 Insurance premiums for individuals and families were and a larger proportion of tax revenue devoted to health $4,479 and $12,106, respectively; family premiums care (Auerbach and Kellermann 2011). Additionally, for March 2019 | Report to the Congress: Medicare Payment Policy 25

56 Health care occupations employment and salaries technologists and technicians, dental hygienists, and ealth care occupations represent a large (9 emergency medical technicians and paramedics) percent) and growing (20 percent growth rate are similar to the average for the non–health care from 2007 to 2017) share of the country’s H workforce. However, health care support occupations’ workforce (Table 1-2). According to data from the salaries (e.g., home health aides, orderlies, medical Bureau of Labor Statistics (BLS), mean salaries for assistants, and medical transcriptionists) are less than clinicians—health care practitioners who diagnose and average salaries. BLS data also indicate that wages treat conditions—are more than twice the average of all for health care professionals may have grown more other occupations (Boards of Trustees 2018, Bureau of rapidly (31 percent), in nominal dollars, than for other Labor Statistics 2018, Bureau of Labor Statistics 2008). occupations (27 percent). Salaries for health care technicians (e.g., radiologic ■ TABLE Employment and salary for health care and all other occupation categories, 2017 1–2 Increase Increase Employees Mean Share of all from from Occupation categories (in millions) salary occupations 2007 2007 N/A 6% 143 All occupations $50,620 % 28 91% 130 5 $49,258 27 All but health care total 27 $48,695 96 All but clinicians 137 5 Health care total 13 31 $64,642 9 20 Health care practitioners and 31 $80,760 6 24 technical occupations 9 32 5 $100,780 4 Clinicians 26 $47,310 2 19 3 Technicians 25 4 Health care support occupations 24 13 $31,310 3 Note: N/A (not applicable). “Clinicians” includes health care practitioners who diagnose or treat conditions, such as physicians, dentists, physician assistants, registered nurses, and physical therapists. “Technicians” includes health care technical occupations such as radiologic technologists and technicians, dental hygienists, emergency medical technicians and paramedics, and pharmacy technicians. “Health care support occupations” includes occupations such as home health aides, orderlies, medical assistants, and medical transcriptionists. Data from self-employed persons are not collected and are not included in the estimates. Salary increases from 2006 are measured in nominal dollars. The Bureau of Labor Statistics cautions against using Occupational Employment Statistics (OES) data to compare two points in time because the survey methodology is designed to create detailed cross-sectional employment and wage estimates but presents challenges in using OES data as a time series. These challenges include changes in the occupational, industrial, and geographical classification systems; changes in the way data are collected; changes in the survey reference period; and changes in mean wage estimation methodology, as well as permanent features of the methodology. Categories may not sum due to rounding. Source: MedPAC analysis of Bureau of Labor Statistics May 2017 National Occupational Employment and Wage Estimates United States and Bureau of Labor Statistics May 2007 National Occupational Employment and Wage Estimates United States. accounted for 24 percent of median household income accounting for 15 percent of median household income, (Census Bureau 2018, Centers for Medicare & Medicaid which was $61,372. The premiums for typical individual Services 2018a, Kaiser Family Foundation and Health and family health insurance were $6,690 and $18,764, 11 Research & Educational Trust 2018). respectively; family premiums accounted for 31 percent By 2017, per of median household income. From 2007 to 2014, middle- capita personal health care spending had grown to $9,106, Context for Medicare payment policy 26

57 income households’ health care spending grew by 25 Life expectancy by sex, race, and Hispanic percent, while their spending fell for categories such as origin food, housing, clothing, and transportation (Baily and In general, life expectancy in the United States has been Holmes 2015). While health care is a growing expense increasing over the past century (although more slowly for households, it is a source of income for health care than in other Organisation for Economic Co-operation 13 providers. A greater share of the nominal-dollar income and Development (OECD) countries). These increases increase may have gone to health care providers rather in longevity are influenced by a range of factors, than other occupation categories (see text box on health including health behavior changes, increased disease care occupations). prevention efforts, and advances in medical treatments. In 2016, average life expectancy at birth for an individual Many Medicare beneficiaries are not exempt from the living in the United States was 78.6 years (Table 1-3, financial challenges of the program’s ever-growing cost- p. 28). However, an individual’s life expectancy can 12 sharing liabilities. In 2018, SMI (Medicare Part B and vary significantly from this average based on certain Part D) premiums and cost sharing likely consumed 24 characteristics, including race, sex, socioeconomic status, percent of the average Social Security benefit, up from and geographic location. Variations have existed ever since 7 percent in 1980 (Boards of Trustees 2018). (Those official data have been collected. One example is that, percentages do not include beneficiary spending on in 2016, women on average had a longer life expectancy premiums for Medicare supplemental insurance.) The than men (81.1 years vs. 76.1 years, respectively) (Table Medicare Trustees estimate that those costs will consume 1-3). Though this longevity gap has lessened in recent 30 percent of the average Social Security benefit by years (data not shown), researchers speculate that these 2035. On average, Social Security benefits account differences are caused by a combination of genetics, for more than 60 percent of income for seniors. For reductions in infections, and behavioral and lifestyle more than one-fifth of seniors, Social Security benefits factors (Beltran-Sanchez et al. 2015). account for 100 percent of income (Social Security Administration 2016). However, some seniors also rely Race and ethnicity are also associated with variations in on accumulated assets to supplement their income in life expectancy. The Hispanic population in the United retirement. Additionally, despite the increasing cost- States in 2016 had a higher life expectancy at birth sharing burden, the availability of SMI Part B and Part (81.8 years) than the non-Hispanic White and African D benefits greatly reduces the costs that beneficiaries American populations, at 78.5 and 74.8 years, respectively would otherwise pay for health care services without (Table 1-3, p. 28). Though these differences have shifted those benefits since general revenues cover a large share somewhat over time, the general trend has persisted, that of those costs. the Hispanic population has the longest life expectancy and non-Hispanic African Americans have the shortest (Arias 2016). Recent trends in life expectancy, Life expectancy by geographic areas morbidity, and mortality Life expectancy in the U.S. varies based on an array of geographic characteristics, including urban and rural Several recent studies and news reports have highlighted location and among states. A 2017 study by Zolot found aspects of decreasing life expectancy and increasing a greater than 20-year difference in life expectancy by mortality and morbidity among some Americans (see county and a trend that these geographic disparities have text box on recent mortality and morbidity trends, p. 29). been increasing over the past few decades (Zolot 2017). These aspects include—for specific groups—decreases A 2014 study by Singh and Siahpush found that life in life expectancy; increasing rates of suicide and deaths expectancy was inversely related to levels of rurality and from drug poisonings; and troubling health indicators that rural African Americans and Whites had lower life and behaviors, such as increased alcohol consumption, expectancies than their urban counterparts (Singh and smoking, and obesity. These trends interact with 14 Siahpush 2014). From 2005 through 2009, those in large longstanding underlying variations in life expectancy, metropolitan areas had a life expectancy of 79.1 years mortality, and morbidity by sex, income, race and compared with 76.9 years in small towns and 76.7 years ethnicity, and geographic location. in rural areas. Compared with their urban peers, people March 2019 | Report to the Congress: Medicare Payment Policy 27

58 TABLE Life expectancy at birth by race/ethnicity and sex, 2007 to 2016 1–3 Change Change 2007 2015 2016 2015–2016 2007–2016 78.1 78.6 –0.1 0.5 78.7 All races and ethnicities, both sexes –0.2 White, not Hispanic, both sexes 78.4 78.7 78.5 0.1 African American, not Hispanic, both sexes 1.3 74.8 75.1 –0.3 73.5 Hispanic, both sexes 80.7 81.9 81.8 1.1 –0.1 0 All races and ethnicities, female 80.6 81.1 81.1 0.5 80.8 81.0 81.0 0.2 0 White, not Hispanic, female –0.2 76.7 78.1 77.9 1.2 African American, not Hispanic, female Hispanic, female –0.1 84.3 83.2 1.0 84.2 75.5 76.3 76.1 0.6 –0.2 All races and ethnicities, male White, not Hispanic, male 75.9 76.3 76.1 0.2 –0.2 African American, not Hispanic, male 71.9 71.5 1.6 –0.4 69.9 –0.2 77.8 79.3 79.1 1.3 Hispanic, male Source: National Center for Health Statistics 2018. 15 in rural areas had higher rates of both smoking and lung et al. 2016). The study found that a state’s economic cancer, along with obesity. Additionally, rural residents and social environment (e.g., welfare policies, tobacco on average had a lower median family income and higher tax rate, level of economic inequality) had a significant poverty rate, and fewer had college degrees, which may effect on women’s mortality rate. The researchers found contribute to the difference in life expectancy. Another that many of the states with the best economic and social study by Chetty and colleagues exploring the association indicators had some of the lowest mortality rates among between life expectancy and income found that low- women. The same correlation was not seen among income individuals’ life expectancy varied substantially males. These findings imply that geographic inequities in based on where they lived (Chetty et al. 2016). The study women’s mortality rates may not be fully explained just found that individuals in the lowest income quartile often by women’s personal characteristics; rather, the influence lived longer and had more healthful behaviors if they of socioeconomic and political contexts must also be resided in urban areas with highly educated populations, considered. high incomes, and high levels of government expenditures. Numerous researchers and media stories have highlighted Some potential explanations for these findings are the growing opioid abuse and mortality trend (Case that these areas may have public policies that improve and Deaton 2017, Case and Deaton 2015, Rudd et al. health (e.g., smoking bans) or they may have greater 2016, Zolot 2017). Case and Deaton note, “In 2000, the funding for public services. However, the Commission’s epidemic was centered in the southwest. By the mid-2000s research has found little difference between rural and it had spread to Appalachia, Florida, and the west coast. urban beneficiaries’ experience with access to care and Today, it’s country-wide” (Case and Deaton 2017). Figure amount of service use. With respect to quality of care, 1-16 (p. 30) shows the age-adjusted drug overdose–related quality is similar for most types of providers in rural and death rate per 100,000 population in 2016. In 2016, the urban areas; however, rural hospitals tend to have below- five states with the highest rates of death due to drug average rankings on mortality and some process measures overdose were West Virginia (52.0 per 100,000), Ohio (Medicare Payment Advisory Commission 2012). (39.1 per 100,000), New Hampshire (39.0 per 100,000), A recent study by Montez and colleagues examined Pennsylvania (37.9 per 100,000), and Kentucky (33.5 per variation in women’s mortality rates across states (Montez 100,000) (data not shown). Context for Medicare payment policy 28

59 Recent mortality and morbidity trends As with any population-level trend, the causes of everal recent studies and news reports have increased midlife morbidity and mortality among highlighted aspects of increasing mortality non-Hispanic Whites are difficult to identify. A recent and morbidity among some Americans (Arias S study found that varying inequalities in women’s 2016, Case and Deaton 2017, Case and Deaton 2015, mortality across states may be partially explained by Montez et al. 2016, Zolot 2017). While researchers macro-level socioeconomic and political factors—for have applied diverse methods and reported various example, policies that shape access to health care, aspects of the trend, two key findings are (1) increases use of tobacco, availability of affordable housing, in mortality in groups of Whites, especially those with children’s health care, and financial safety nets (Montez only a high school diploma or less, and (2) lower and et al. 2016). Some researchers point to the availability decreasing life expectancy for residents of certain of opioid drugs as a possible source of rising mortality geographic areas. rates. Increased reports of pain combined with the Over the past century, the U.S. has experienced increased availability of opioid prescriptions for pain generally consistent declines in the mortality rate. that began in the late 1990s have been widely noted, However, there has recently been an increase in as well as the associated mortality (Rudd et al. 2016). mortality among the middle-aged (45 to 54 years old) Studies have also found that recent restrictions of non-Hispanic White population (Case and Deaton opioid prescriptions may lead to unintended negative 2015, Kochanek et al. 2015). The analysis by Case consequences such as increased use of heroin and Deaton found no similar mortality rate increase in (Compton et al. 2016). There is concern that those other industrialized countries or in the non-Hispanic affected by opioid and substance use in midlife include African American or Hispanic population of this age current Medicare beneficiaries under 65 and others who group (Case and Deaton 2015). Case and Deaton will age into Medicare in worse health than current note that three causes of death have dramatically beneficiaries. Researchers have found that patients increased among this group in the past decade: with a diagnosed opioid dependency are high users of suicides, intentional and unintentional poisonings, and health care services, including office visits, lab tests, chronic liver disease. Additionally, increases in midlife and related treatments (FAIR Health 2016). However, mortality in this group are paralleled by increases in this use may be related to the underlying conditions for self-reported midlife morbidity and troubling health which opioids were used as much as the consequences indicators and behaviors such as increased alcohol of opioid abuse or related effects. Addiction is hard to consumption, smoking, and obesity. Case and Deaton’s treat, chronic pain is challenging to control, and these findings indicate that the increase in reports of poor conditions appear to be potential problems among the health by this group has been matched by increasing next generation of Medicare beneficiaries. ■ reports of physical pain and psychological distress. Significant increases in drug overdose death rates from Pennsylvania, South Carolina, Tennessee, Texas, Vermont, 2015 to 2016 were seen primarily in the Northeast, Virginia, West Virginia, and Wisconsin. Midwest, and South census regions. States with Life expectancy at age 65 statistically significant increases in drug overdose death rates from 2015 to 2016 included Connecticut, Delaware, Recent decreases in life expectancy and increases in Florida, Illinois, Indiana, Kentucky, Louisiana, Maine, mortality are mostly isolated to the under-65 population. Maryland, Massachusetts, Michigan, Minnesota, Missouri, Between 2007 and 2016, life expectancy at 65 (i.e., New Jersey, New York, North Carolina, Ohio, Oklahoma, remaining years of life) increased for all groups (Table 1-4, p. 31). March 2019 | Report to the Congress: Medicare Payment Policy 29

60 FIGURE FIGURE xxxxxxxx X-X Age-adjusted opioid-related death rate per 100,000 population, 2016 1–16 6.9 to 11.0 11.1 to 13.5 13.6 to 16.0 16.1 to 18.5 18.6 to 21.0 21.1 to 52.0 MedPAC analysis of data from the Centers for Disease Control and Prevention, National Center for Health Statistics, Underlying Cause of Death 1999– Source: 2016 on CDC WONDER Online Database, released December 2017. Source: Note and Source in InDesign. leading causes. Suicide was the 10th leading cause of Leading causes of death death among all Americans in both 1980 and 2016. Over the past few decades, there has been little change in the leading causes of death in the U.S., both for all Some of the leading causes of death overlap with the Americans and those 65 and older (Table 1-5, opposite most prevalent and most expensive chronic conditions page, and Table 1-6, p. 32). Heart disease and cancer have among Medicare FFS beneficiaries (Table 1-7, p. 33). In remained the first and second leading causes of death, Table 1-7, the Medicare total per capita spending amounts respectively, for both age groups for more than 75 years represent all Medicare spending for FFS beneficiaries (Hoyert 2012, National Center for Health Statistics 2018). with the specified condition (i.e., the spending cannot In each year between 1935 and 2016, three causes—heart be attributed strictly to the specified condition because disease, cancer, and stroke—remained among the five Context for Medicare payment policy 30

61 TABLE Life expectancy at age 65 by race/ethnicity and sex, 2007 to 2016 1–4 Change Change 2007 2015 2016 2007–2016 2015–2016 All races and ethnicities, both sexes 18.8 19.3 19.4 0.6 0.1 White, not Hispanic, both sexes 18.8 19.3 19.3 0.5 0 18.0 African American, not Hispanic, both sexes 17.2 18.1 –0.1 0.8 Hispanic, both sexes 20.5 21.4 0.9 0 21.4 All races and ethnicities, female 20.5 20.6 0.6 0.1 20.0 White, not Hispanic, female 20.0 20.4 20.5 0.5 0.1 African American, not Hispanic, female 18.7 19.5 19.5 0.8 0 22.7 0.1 Hispanic, female 21.7 22.6 1.0 All races and ethnicities, male 18.0 0 0.6 18.0 17.4 0 White, not Hispanic, male 17.4 18.0 18.0 0.6 16.2 16.2 15.3 African American, not Hispanic, male 0 0.9 0 1.0 19.7 19.7 18.7 Hispanic, male National Center for Health Statistics 2018. Source: beneficiaries may have other health conditions that trends in part because treatments for conditions are contribute to their total Medicare use and spending influenced by changes in technology and definitions of amounts). what constitutes disease shift over time. The Commission explored this question in 2007 and found upward pressure It is unclear how the prevalence of these and other acute on Medicare costs because of a greater proportion of and chronic conditions contributes to Medicare spending beneficiaries being treated for multiple chronic conditions TABLE Leading causes of death, 1980 and 2016 1–5 Table 1-5a. Leading causes of death, 1980 able 1-5b. Leading causes of death, 2016 T Share of Share of Cause of death Cause of death deaths deaths 1. 38.2% 1. Heart disease 23.1% Heart disease 2. Cancer 20.9 2. Cancer 21.8 3. 3. Stroke Unintentional injuries 5.9 8.6 4. 4. Chronic lower respiratory disease 5.6 5.3 Unintentional injuries Chronic obstructive pulmonary diseases 2.8 5. Stroke 5.2 5. Pneumonia and influenza 2.7 6. Alzheimer’s disease 4.2 6. 7. Diabetes mellitus 1.8 7. Diabetes mellitus 2.9 1.9 8. 1.5 8. Pneumonia and influenza Chronic liver disease and cirrhosis 9. Atherosclerosis 1.5 9. Nephritis, nephrotic syndrome, and nephrosis 1.8 10. Suicide 10. Suicide 1.4 1.6 Note: Starting with 2011 data, the rules for selecting renal failure as the underlying cause of death were changed, affecting the number of deaths in the “nephritis, nephrotic syndrome, and nephrosis” and “diabetes mellitus” categories. These changes directly affect the cases of death with mention of renal failure and other associated conditions such as diabetes mellitus with renal complications. The result is a decrease in the number of deaths attributed to nephritis, nephrotic syndrome, and nephrosis and an increase in the number of deaths attributed to diabetes mellitus. Therefore, trend data for these two causes of death should be intrepreted with caution. 2018 data on mortality from the National Center for Health Statistics. Source: March 2019 | Report to the Congress: Medicare Payment Policy 31

62 TABLE Leading causes of death at age 65 and older, 1980 and 2016 1–6 Table 1-6a. Leading causes of death at age 65 T able 1-6b. Leading causes of death at age 65 and older, 2016 and older, 1980 Share of Share of deaths Cause of death deaths Cause of death 1. 44.4% Heart disease 1. Heart disease 25.3% 21.1 2. Cancer 19.3 2. Cancer 6.5 3. Stroke 10.9 3. Chronic lower respiratory diseases 3.4 4. Pneumonia and influenza 4. Stroke 6.1 5. Chronic obstructive pulmonary diseases 5. Alzheimer’s disease 5.7 3.2 2.8 2.1 6. Atherosclerosis Diabetes mellitus 6. 7. 1.9 7. Unintentional injuries 2.7 Diabetes mellitus 2.1 Unintentional injuries 1.9 8. Pneumonia and influenza 8. Nephritis, nephrotic syndrome, and nephrosis 1.0 9. Nephritis, nephrotic syndrome and nephrosis 2.1 9. 10. Chronic liver disease and cirrhosis 0.7 10. Septicemia 1.5 Note: Starting with 2011 data, the rules for selecting renal failure as the underlying cause of death were changed, affecting the number of deaths in the “nephritis, nephrotic syndrome, and nephrosis” and “diabetes mellitus” categories. These changes directly affect affect the number of deaths attributed to renal failure and other associated conditions such as diabetes mellitus with renal complications. The result is a decrease in the number of deaths attributed to nephritis, nephrotic syndrome, and nephrosis and an increase in the number of deaths attributed to diabetes mellitus. Therefore, trend data for these two causes of death should be intrepreted with caution. Source: 2018 data on mortality from the National Center for Health Statistics. small set of outcomes, patient experience, and resource use (Medicare Payment Advisory Commission 2007). This measures that are not unduly burdensome for providers increase reflected growth in the prevalence of obese to report. Further, these population-based measures beneficiaries, advances in technology for diagnosing and can be used to assess and compare the quality of care treating conditions, and changes in disease definitions. across different populations, such as MA beneficiaries, More recently, CBO found that, while ample evidence beneficiaries under accountable care organizations, or exists of increased health care spending associated with FFS beneficiaries. The measures can also be applied to obesity, evidence about the effects of weight loss on the health and health care spending of obese people is populations in defined market areas or populations served inconclusive at best (Congressional Budget Office 2015). by distinct provider types. Currently, Medicare does not consistently measure quality across MA plans, FFS populations, and providers, so we The relationship between Medicare cannot report trends about the entire Medicare program’s quality of care. Where feasible to measure, we report spending and quality whether the quality of care delivered in certain provider settings has improved or has been maintained over the past The Commission contends that Medicare payments should few years. For example, in the FFS population, hospital- not be made without consideration of the quality of care level readmission rates, readmission rates within 30 days delivered to beneficiaries (Medicare Payment Advisory after discharge from a skilled nursing facility, and dialysis Commission 2018). The Commission has supported the facility readmission rates have improved over the past few implementation of quality incentive programs across years. the Medicare program—for example, the Hospital Readmissions Reduction Program (Medicare Payment As Medicare per beneficiary spending has increased over the Advisory Commission 2008). The Commission asserts life of the program, has the quality of health care received that Medicare quality incentive programs should use a Context for Medicare payment policy 32

63 TABLE Selected chronic conditions by prevalence and total 1–7 per capita spending among Medicare FFS beneficiaries, 2015 Total per capita spending Prevalence among for beneficiaries with Medicare FFS the specified condition beneficiaries Chronic condition Five chronic conditions most prevalent among Medicare FFS beneficiaries: Hypertension 58.3% $13,718.10 47.3 Hyperlipidemia 13,053.20 32.1 Rheumatoid arthritis/osteoarthritis 15,231.10 Diabetes mellitus 28.2 15,067.40 18,214.30 28.2 Ischemic heart disease Five chronic conditions with highest total per capita spending among Medicare FFS beneficiaries: Stroke 3.9 29,852.60 Heart failure 14.5 27,078.20 COPD 12.0 24,332.90 24,270.90 N/A Schizophrenia/other psychotic disorders Chronic kidney disease 24,027.90 19.3 Note: FFS (fee-for-service), COPD (chronic obstructive pulmonary disease), N/A (not available). Data include all Medicare beneficiaries who were eligible for or enrolled in Medicare on or after January 1, 2015. Period prevalence is calculated for these rates: beneficiaries with full or nearly full FFS coverage (i.e., 11 or 12 months of Medicare Part A and Part B (or coverage until time of death) and 1 month or less of HMO coverage) during the year who received treatment for the condition within the condition-specified look-back period (chronic conditions have a 1- to 3-year look-back period). Beneficiaries may be counted in more than one chronic condition category. The Medicare utilization and spending information presented above represents total Medicare FFS spending for beneficiaries with the condition. The information should not be used to attribute utilization or payments strictly to the specific condition selected because beneficiaries with any of the specific conditions presented may have other health conditions that contribute to their Medicare utilization and spending amounts. Source: 2017 data from the Chronic Conditions Warehouse from the Centers for Medicare & Medicaid Services. • Between 1991 and 2016, the share of people ages 65 by Medicare beneficiaries improved? From the perspective to 74 reporting fair or poor health status declined from of beneficiary health and longevity, indicators show 26 percent to 19 percent (Figure 1-18, p. 35); the share improvements, primarily for beneficiaries ages 65 and older; the limited data available for younger Medicare beneficiaries of people ages 75 and older reporting fair or poor health status declined from 34 percent to 26 percent; include one indication of potentially poorer quality: between 2010 and 2016, the share of adults who report Life expectancy at age 65 has steadily increased since • some difficulty in functional domains reporting fair the introduction of Medicare. Individuals who reached or poor health status declined from 17 percent (the age 65 in 2015 had a remaining life expectancy of first year the measure was reported) to 16; but, for 19.3 years, compared with 15.1 years for this age that same period, the share of adults who report a lot group in 1970. However, these beneficiaries’ gains in of difficulty in functional domains or cannot perform longevity are outpaced by their peers in other OECD them at all who report fair or poor health status countries. From 1970 to 2015, U.S. life expectancy increased from 47 percent in to 52 percent. at age 65 improved by 4.2 years (Figure 1-17, p. 34), While the share of people ages 65 and older with • compared with an average gain of 5.3 years for the 16 chronic conditions, such as diabetes, hypertension, 35 OECD countries. (Comparable information for and high cholesterol, has increased over time, the the Medicare population under age 65 is not readily share of people who have those conditions under available.) control has also increased (Federal Interagency Forum March 2019 | Report to the Congress: Medicare Payment Policy 33

64 FIGURE Medicare margins... FIGURE X-X Life expectancy at age 65 is lower and increased less in 1–17 the United States than in other OECD countries, 1970–2015 25 21.9 21.5 21.0 20.9 20.9 20.6 20.3 20.3 20.2 20.2 20.2 20.1 19.7 19.7 19.5 19.5 19.3 20 18.0 17.9 17.8 17.7 15.5 15 Years 10 5 2015 1970 0 Italy Israel Japan Spain Korea Turkey France Brazil Austria Mexico Finland Poland Sweden Norway Australia Canada Germany OECD35 Switzerland United States United Kingdom Russian Federation Note: OECD (Organisation for Economic Co-operation and Development). “OECD35” refers to the average of all 35 OECD countries. Selected OECD countries are shown. Early life expectancy figures for Italy, Canada, and Finland are as of 1971 rather than 1970. For Canada, the recent life expectancy figure is as of 2012; for Brazil, 2013. Data are not available for 1970 for Brazil, Israel, and the Russian Federation. 2017 data on life expectancy at age 65 from the Organisation for Economic Co-operation and Development. Source: Note: Note and Source are in InDesign. Source: on Aging-Related Statistics 2016, National Center for disabilities has shifted over time, decreasing overall Health Statistics 2015). (Comparable information for from 36 percent to 25 percent. the Medicare population under age 65 is not readily Notes about this graph: available.) • Data is in the datasheet. Make updates in the datasheet. Baby boomers will make up the next However, many factors other than health care also • WATCH FOR GLITCHY RESETS WHEN YOU UPDATE DATA!!!! generation of Medicare beneficiaries impact individual and population health, including • The column totals were added manually. poverty, income levels, and health-related behaviors As the baby-boom generation ages, enrollment in the such as smoking and alcohol consumption. For • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. Medicare program will surge. In 15 years, Medicare is example, between 1970 and 2017, the poverty rate • I can’t delete the legend, so I’ll just have to crop it out in InDesign. projected to have more than 80 million beneficiaries— among people ages 65 years and older fell, with the up from 57 million beneficiaries today—almost 90 • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph support of the Social Security program, from almost 17 percent of whom will be of the baby-boom generation. default when you change the data. 25 percent to about 9 percent, potentially having a These individuals will define the upcoming Medicare substantial effect on individual and population health • Use paragraph styles (and object styles) to format. population in terms of age distribution, health status, for that age group (Figure 1-19, p. 36). Between 1997 • Data was from: R:\Groups\MGA\data book 2007\data book 2007 chp1 health insurance experiences before Medicare enrollment, and 2017, the poverty rate for younger adults with and financial security. Context for Medicare payment policy 34

65 FIGURE FIGURE Title here... The share of Medicare eligibles reporting fair or poor 1–18 1-XX health status changed over time, available years 1991–2016 60 Age 75 years or older Age 65 to 74 years 50 A lot of difficulty in functional domains or cannot do at all Some difficulty in functional domains 40 30 20 health status (in percent) People reporting fair or poor 10 0 2003 2013 1991 1995 1997 1998 1999 2000 2001 2002 2015 2004 2005 2012 2006 2007 2008 2011 2014 2016 2010 2009 Note: “A lot of difficulty in functional domains or cannot do at all” and “some difficulty in functional domains” include people 18 and older who report one or more of the following six functional limitations: seeing (even if wearing glasses), hearing (even if wearing hearing aids), mobility (walking or climbing stairs), communication (understanding or being understood by others), cognition (remembering or concentrating), and self-care (such as washing all over or dressing). These measures of functional limitations among adults 18 years and older did not begin being reported until 2010. 2018 data on health status from the National Center for Health Statistics. Source: Note: Note and Source are in InDesign. grow as baby boomers continue to age (Boards of Trustees The Medicare population becomes younger Source: 2014, Census Bureau 2014). In 2013, per beneficiary as it expands and then grows older as the baby-boom generation ages spending for those ages 85 and older was about twice that of those ages 65 to 74. So, the changing age structure of Enrollment in the Medicare program is projected to grow Notes about this graph: the Medicare population will exert somewhat less pressure rapidly as members of the baby-boom generation age into • Data is in the datasheet. Make updates in the datasheet. on spending in the very near term, at least on a per capita the program (see Figure 1-10a, p. 19). These individuals basis, and then pressure will increase again over the longer • I deleted the years from the x-axis and put in my own. began aging into Medicare in 2011 at an average rate of 18 term. 10,000 people per day. Medicare enrollment is projected • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. to grow by nearly 50 percent by 2030, and this growth will • The dashed line looked ok here, so I didn’t hand draw it. be made up almost entirely of baby boomers (Figure 1-20, p. 37) (Census Bureau 2014). • I can’t delete the legend, so I’ll just have to crop it out in InDesign. Inefficient spending suggests Medicare • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph could spend less without compromising The Medicare population over the next 15 years will default when you change the data. be relatively younger, as members of the baby-boom care, but not without challenges generation join and increase the number of beneficiaries in • Use paragraph styles (and object styles) to format. younger age categories (Figure 1-21, p. 38). With few exceptions throughout modern history, health care spending in the U.S. has grown robustly, outpacing The share of the Medicare population ages 85 years or the growth in the economy. Even if Medicare’s recent older is projected to decline slightly through 2025 and then low growth in per beneficiary spending is sustained, March 2019 | Report to the Congress: Medicare Payment Policy 35

66 FIGURE Title here... FIGURE 1-XX The poverty rate has fallen over time among people ages 65 years and older 1–19 and adults with disabilities, available years 1970–2017 40 35 30 25 20 15 10 Poverty rate (in percent) Disabled, 18 to 64 years 5 65 years and older 0 1980 1975 1970 1985 1995 2000 1990 2010 2015 2005 Data on the poverty rate among people with disabilities have been reported for only seven years: 1997, 2000, 2010, 2014, 2015, 2016, and 2017. Note: Source: Data on income and poverty from the Census Bureau. Note: Note and Source are in InDesign. enrollment growth from the aging of the baby boomers Services that have been widely recognized as low value Source: will contribute to growth in total spending regardless. continue to be performed regularly (Schwartz et al. 2014). However, the Commission does not believe that ever- The U.S. spends more on health care than any other increasing health care spending is inevitable. There is Notes about this graph: country in the world (both on a per capita basis and as strong evidence that a sizeable share of current health care a share of GDP), but studies consistently show it ranks • Data is in the datasheet. Make updates in the datasheet. spending—both overall and by Medicare—is inefficient poorly on indicators of efficiency, equity, and outcomes. or unnecessary, providing an opportunity for policymakers • I deleted the years from the x-axis and put in my own. According to a 2014 study by the Commonwealth Fund, to reduce spending, extend the life of the program, and • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. the United States ranks last of 11 nations on 2 indicators reduce pressure on the federal budget. of healthy lives—mortality amenable to medical care and • The dashed line looked ok here, so I didn’t hand draw it. healthy life expectancy at age 60 (Davis et al. 2014). Geographic variation within and outside the • I can’t delete the legend, so I’ll just have to crop it out in InDesign. U.S. indicates that some share of spending is • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph Medicare’s challenges to increasing inefficient efficiency default when you change the data. Research on Medicare spending shows that areas with The Medicare program is a complex and fragmented higher spending or more intensive use of services do • Use paragraph styles (and object styles) to format. system, consisting of multiple paths to entitlement; not necessarily have higher quality of care or improved multiple types of coverage (Part A, Part B, Part C patient outcomes (Fisher et al. 2003a, Fisher et al. 2003b). (Medicare Advantage), and Part D); multiple payment Measures of service use, adjusted for health status and systems; and different rules for each setting. The Medicare standardized prices, also show considerable variation program must set prices for thousands of discrete services (Medicare Payment Advisory Commission 2011b). Context for Medicare payment policy 36

67 FIGURE FIGURE Title here... X-X By 2030, the entire baby-boom generation will be eligible for Medicare 1–20 Figure 1-20b: Population by age and sex: 2030 Figure 1-20a: Population by age and sex: 2010 Baby-boom 3 2 5 3 85+ 85+ population Baby-boom 15 19 29 24 65–84 65–84 population 41 39 31 32 45–64 45–64 Age Age 41 41 38 39 25–44 25–44 42 51 41 39 5–24 5–24 11 10 10 11 0–5 Female Female Male Male 0–5 100 80 60 20 20 40 60 0 80 0 100 40 Population (in millions) Population (in millions) Source: Census Bureau, 2010 Census; 2017 National Population Projections, middle series. Note: Note and Source are in InDesign. at different levels of aggregation (e.g., inpatient hospital • Fragmented payment system across multiple settings. Source: payments are paid based on the stay, while physician The program sets payment rates each year for at least payments are based on the service) and in different nine health care settings or provider types: acute care Notes about this graph: labor markets across the country. The Medicare program hospitals, physician and other health professional statute and rulemaking include a substantial number of services, home health agencies, skilled nursing • Data is in the datasheet. Make updates in the datasheet. exceptions, adjustments, and modifications to its general facilities, long-term care facilities, hospice, inpatient • I deleted the years from the x-axis and put in my own. policies. Several of Medicare’s structural features (and rehabilitation facilities, ambulatory surgical centers, • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. some shared across the health care system) complicate and end-stage renal disease dialysis facilities. In • The dashed line looked ok here, so I didn’t hand draw it. efforts to achieve spending efficiencies: addition to the yearly rule-making process involved in setting these rates, administrators oversee other • I can’t delete the legend, so I’ll just have to crop it out in InDesign. • Medicare being just one payer in the overall, parts of the program that operate on fee schedules • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph While Medicare is the multipayer health care system. (ambulances, outpatient lab facilities) or on cost- default when you change the data. single largest payer in the health care sector, the policy based payment (rural health centers, critical access signals from multiple payers can interact in ways • Use paragraph styles (and object styles) to format. hospitals). Payment rates for Part C (Medicare that sometimes result in unintended consequences. Advantage) are set using administrative pricing For example, if a dual-eligible nursing home resident based on a competitive process, and Part D payments is hospitalized for three days, he or she would then (prescription drugs) are generally set by market rates. potentially qualify for a Medicare-covered skilled The fragmented payment system across multiple nursing facility stay, shifting the cost burden from health care settings reduces incentives to provide the state Medicaid program to the federal Medicare patient-centered, coordinated care. program. Other care for beneficiaries dually eligible for Medicare and Medicaid can be fragmented. March 2019 | Report to the Congress: Medicare Payment Policy 37

68 FIGURE Title here... FIGURE 1-XX The Medicare population will become younger and then older 1–21 100 90 Ages 85 years and older 80 70 60 50 Percent 40 Ages 65 to 84 years 30 20 10 0 2020 2030 2060 2055 2050 2045 2040 2035 2025 2016 Source: Census Bureau, 2017 National Population Projections. • Coverage of services delivered by any willing a cap on out-of-pocket (OOP) costs (a feature that Note: Note and Source are in InDesign. Under Medicare’s statute, the program provider. exists in nearly all private insurance policies). In Source: generally covers all medically necessary (a criterion response, many beneficiaries purchase supplemental that is open to interpretation) services that are coverage that includes an OOP maximum. Most delivered by any willing provider (any provider that supplemental policies also substantially reduce Notes about this graph: is willing to meet Medicare’s criteria). As a result, or eliminate most of the beneficiary liability for • Data is in the datasheet. Make updates in the datasheet. Medicare does not have the authority to develop coinsurance and deductibles, thereby blunting the provider networks or to credential providers, tools that impact of cost sharing. As a result, there is little • I deleted the years from the x-axis and put in my own. private payers often use to reduce the potential for incentive for beneficiaries to be cost conscious—that • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. fraud and abuse. In some cases, the Medicare program is, to select only those services that are necessary and even has difficulty removing providers or suppliers choose providers who use efficient clinical practices • The dashed line looked ok here, so I didn’t hand draw it. whose claims histories clearly demonstrate aberrant (Medicare Payment Advisory Commission 2012). • I can’t delete the legend, so I’ll just have to crop it out in InDesign. patterns of billing, care, or both. • Different prices for the same or similar services. • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph Beneficiaries face • The program’s benefit design. Because of the different settings in which services default when you change the data. differential cost sharing by service (for example, are delivered, the Medicare program in some cases • Use paragraph styles (and object styles) to format. coinsurance for physician services is 20 percent, while has different payment rates for the same or similar home health has no coinsurance); in addition, the services. Under these circumstances, providers have cost-sharing amounts, percentages, and deductibles an incentive to shift care to the higher paid setting, vary by setting, and some services are not covered which leads to increased program spending and higher (for example, Medicare does not generally cover beneficiary cost sharing. long-term care). Medicare Part A and Part B lack Context for Medicare payment policy 38

69 Undervalued and overvalued services. • In the process developed its ability to identify potentially fraudulent of setting rates for thousands of services, certain billing patterns. However, all of CMS’s activities in services are undervalued relative to others, providing this area are constrained by resources and are subject to incorrect incentives for their use. For example, the statutory requirements that limit its ability to use the same Commission has raised concerns that the Medicare fee tools as private insurers to reduce fraud (Government schedule overpays for services provided by clinicians Accountability Office 2013). in procedural specialties and underpays for services The Congress has recognized the need for CMS to provided by clinicians in primary care specialties pursue value-based purchasing policies. For example, the (Medicare Payment Advisory Commission 2011a). Improving Medicare Post-Acute Care Transformation This imbalance results in significantly higher income Act of 2014 required post-acute care providers to report for clinicians in procedural specialties relative to standardized performance data and linked these measures those in primary care specialties, contributing to a to payment. Earlier, in 2010, PPACA emphasized tying corresponding imbalance in clinician supply. payment to quality in the Medicare program (e.g., by The Medicare program • Prompt payment standards. allowing accountable care organizations that meet quality also follows prompt payment requirements, paying thresholds to share in cost savings and by reducing claims within 30 days of receipt. Otherwise, Medicare payments to hospitals with excessive readmissions and is liable for interest. This emphasis on timely payment hospital-acquired conditions). PPACA also included means that, in many cases, the claim may be paid and new CMS authorities through the establishment of an only thereafter identified as potentially fraudulent or innovation center to test different payment structures erroneous. and methodologies; the intention is to reduce program expenditures while maintaining or improving quality • Vulnerability to patient selection, steering, and of care, which, if successful, could be extended across overuse. Another consequence of Medicare’s payment Medicare. structure is its vulnerability to patient selection, steering, and overuse. For example, with some The Commission’s approach to addressing payment systems, it is financially advantageous for these challenges providers to treat certain kinds of beneficiaries and Medicare’s goal should be to obtain the greatest possible avoid others, provide certain types of services over value for the program’s expenditures, which means others, or treat beneficiaries in a higher paid setting. In maintaining beneficiaries’ access to high-quality services addition, in Medicare’s FFS system, providers may be while encouraging efficient use. However, managing able to increase their revenue by increasing the volume payment rates alone will not address the Medicare FFS of services they provide without commensurate value system’s key challenge—that providers are usually paid to the beneficiary. Further, clinicians can prescribe more for doing more services but are usually not held pharmaceutical drugs and medical devices while accountable for outcomes. Resolving this conundrum will receiving payment from manufacturers. require further reform of both the payment and delivery systems. These features make the program vulnerable to inappropriate care, waste, and fraud. GAO annually In pursuit of this goal, the Commission has made multiple designates Medicare as a high-risk program because of recommendations to the Congress and the Secretary that, if its size, complexity, and susceptibility to mismanagement implemented, have the potential to improve the quality of and improper payments, which include fraud and errors care and move the Medicare program beyond just blindly but not overuse. For fiscal year 2014, the agency found paying FFS rates. For example, the Commission has made improper payments of 12.7 percent for FFS Medicare, 9 the following recommendations: percent for Part C, and 3.3 percent for Part D (Government Accountability Office 2013). Payments should be based on Site-neutral payments. • patient characteristics rather than the site of service. In recent years, CMS has gained new authorities to exclude potentially fraudulent providers from the program —reduce payment rates for evaluation • March 2012 and apply different levels of scrutiny to new providers and management office visits provided in hospital based on their fraud potential. CMS has also further outpatient departments so that total payment rates March 2019 | Report to the Congress: Medicare Payment Policy 39

70 for these visits are the same whether the service • determine star ratings as though the is provided in an outpatient department or a consolidations had not occurred and maintain physician office. the preconsolidation reporting units until new geographic reporting units are implemented. —eliminate the differences in March 2015 • payment rates between inpatient rehabilitation —for physicians: March 2018 • facilities and skilled nursing facilities for selected eliminate the current Merit-based Incentive • conditions. Payment System; and • Readmissions measures. Providers should be • establish a new voluntary value program in measured and held accountable for the share of their FFS Medicare in which: patients who are readmitted to the hospital. • clinicians can elect to be measured as part • June 2008 —confidentially report readmission of a voluntary group and rates and resource use around hospitalization episodes to hospitals and physicians. Beginning • clinicians in voluntary groups can in the third year, providers’ relative resource use qualify for a value payment based on should be publicly disclosed. their group’s performance on a set of population-based measures. • —reduce payments to hospitals with June 2008 relatively high readmission rates for select • The Medicare program should Value-based payment. conditions and allow shared accountability pay for value rather than quantity. between physicians and hospitals. • March 2005 —establish a quality incentive • March 2012 —reduce payments to skilled nursing payment policy for hospitals in Medicare. facilities with relatively high risk-adjusted rates of —establish a quality incentive • March 2005 rehospitalization during Medicare-covered stays payment policy for physicians in Medicare. and be expanded to include a time period after discharge from the facility. —establish a quality incentive March 2005 • payment policy for home health agencies in —reduce payments to home health • March 2014 Medicare. agencies with relatively high risk-adjusted rates of hospital readmission. • March 2012 —implement a value-based purchasing program for ambulatory surgical The results of quality measurement Quality measures. • center services no later than 2016. programs should be meaningful for providers and patients. June 2017 • —no later than 2022, create and phase in a voluntary Drug Value Program (DVP) that —for Medicare Advantage: March 2018 • must have the following elements: establish geographic areas for Medicare • Medicare contracts with a small number of • Advantage quality reporting that accurately private vendors to negotiate prices for Part B reflect health care market areas; products. • calculate star ratings for each contract at the Providers purchase all DVP products at • geographic level for public reporting and for the price negotiated by their selected DVP the determination of quality bonuses; vendor. for any consolidations effective on or after • Medicare pays providers the DVP-negotiated • January 1, 2018, require companies to report price and pays vendors an administrative fee, quality measures using the geographic with opportunities for shared savings. reporting units and definitions as they existed before consolidation; and Beneficiaries pay lower cost sharing. • Context for Medicare payment policy 40

71 • Medicare payments under the DVP cannot Because of its size and because other payers use its exceed 100 percent of average sales price. payment methods, Medicare is an important influence on the nation’s health care delivery system and its evolution. Vendors use tools including a formulary and, • Reciprocally, trends in the private health insurance market for products meeting selected criteria, binding can influence whether Medicare’s payment reforms are arbitration. ultimately successful. Because of this interaction between public and private payers, the alignment of incentives across payers is an important consideration for delivery system reforms. Conclusion Despite the relatively lower rates of spending growth The high and growing level of health care spending as recently experienced by Medicare, the program is a share of the economy means that—absent substantial projected to continue to absorb increasing amounts of changes in spending or the economy—an ever-increasing federal revenue. Absent changes to current policy, other amount of the country’s economic activity and gain will be public investments such as education and infrastructure dedicated to purchasing health care. Medicare is the single will be crowded out by high and growing levels of health largest payer in the health care sector and will expand with care spending. State and federal budgets face continued the aging of the baby-boom generation, greatly increasing fiscal pressure, effects intensified by health care spending program spending. Significant cross-sectional variation trends. In light of strained federal, family, and individual in use and spending, which does not correspond to better budgets, the Medicare program must urgently pursue quality, raises concern that higher health care use and reforms that decrease spending and improve quality. ■ spending are not improving overall health and are putting beneficiaries at risk, both medically and financially. March 2019 | Report to the Congress: Medicare Payment Policy 41

72 Endnotes 8 For Part D, the beneficiary premium share is based on 25.5 Going forward, the Medicare Trustees project that 1 percent of the average cost of the basic benefit. opportunities for further generic use may diminish. Growth in the use and development of high-cost specialty drugs is 9 Among a range of options for addressing Medicare spending beginning to overtake the moderating price influence of is raising the eligibility age for Medicare. In December generics (Medicare Payment Advisory Commission 2016). 2016, CBO scored the option of gradually increasing the Medicare eligibility age from 65 to 67, beginning in 2020 Figure 1-2 (p. 9) shows that the share of spending accounted 2 (Congressional Budget Office 2016b). Implementing this for by private health insurance (35 percent in 2017) is greater option would reduce federal budget deficits between 2020 than Medicare’s share (22 percent in 2017). However, in and 2026 by $18 billion. All told, CBO estimates that, by contrast to Medicare, private health insurance is not a single 2046, spending on Medicare (net of offsetting receipts) purchaser of health care; rather, it includes many payers, such would be about 2 percent less under this option than it would as traditional managed care, self-insured health plans, and be under current law, amounting to 5.6 percent of gross indemnity plans. domestic product rather than 5.7 percent. On the basis of its 3 The Commission’s calculations are based on aggregate Part estimates for 2020 through 2026, CBO projects that roughly D reimbursements to plans and employers on an incurred three-fifths of the long-term savings from Medicare under basis as shown in Table IV.B10 of the 2018 annual report this option would be offset by changes in federal outlays of the Boards of Trustees of the Medicare trust funds. Per for Social Security, Medicaid, and subsidies for coverage beneficiary spending excludes premium payments. through the marketplaces as well as by reductions in revenues. Supporters of this option point to the increase in overall life Outpatient hospital services and outpatient lab services are 4 expectancy since the introduction of the Medicare program. combined in Figure 1-6 (p. 14) because a large portion of However, these gains in longevity have not been shared by outpatient laboratory services were bundled into the outpatient all Americans. People who have lower socioeconomic status, prospective payment system effective January 1, 2014. are racial or ethnic minorities, or live in rural areas all tend to have lower life expectancy. For example, within 5 miles of The Medicare Trustees project enrollment and costs for each 5 Washington, DC, residents of Friendship Heights, MD, have a of the three categories of Medicare enrollees: aged, disabled, life expectancy of 96.1 years, while those in Anacostia’s Barry and end-stage renal disease (ESRD). While the numbers of Farm average 63.2 years (National Center for Health Statistics under-65 and ESRD beneficiaries are projected to increase, 2018). this growth is outpaced by the influx of baby boomers turning 65. Aged beneficiaries accounted for about 83 percent of FFS Other major health programs include Medicaid, the Children’s 10 enrollees in 2007, and their number is projected to grow to Health Insurance Program, and federal subsidies for the about 88 percent by 2026. federal and state exchanges legislated under PPACA. In addition to payroll taxes, the HI Trust Fund’s income 6 Household income, health expenditures, and premiums are all 11 sources include taxation of Social Security benefits (8 percent measured in nominal dollars. in 2017), premiums from people who are not eligible for premium-free Part A (1 percent in 2017), general revenue Medicare beneficiaries with low income and assets have their 12 transfers for certain uninsured beneficiaries who are not premiums and, in some cases, their cost sharing paid for by entitled to HI coverage based on their work history but Medicaid, and some others have retiree coverage or medigap are eligible through special statutes (less than 1 percent in policies that cover cost sharing. 2017), monies from fraud and abuse control activities (less The National Center for Health Statistics defines life 13 than 1 percent in 2017), and interest earned on the trust fund expectancy as the average number of years that a hypothetical investments (2 percent in 2017). group of infants would live at each attained age if the group The standard HI payroll tax rate is scheduled to remain 7 were subject, throughout its lifetime, to the age-specific death constant at 2.9 percent (for employees and employers, rates prevailing in the actual population in a given year (Arias combined). In addition, starting in 2013, high-income workers 2016). pay an additional 0.9 percent of their earnings above $200,000 14 The authors noted limitations to their study: “Life expectancy for single workers or $250,000 for married couples filing joint estimates for Hispanics, Asian/Pacific Islanders, and income tax returns. American Indians/Alaska Natives should be interpreted with Context for Medicare payment policy 42

73 caution as vital statistics–based mortality rates for these Baby boomers are people born during the demographic post– 17 groups tend to be underestimated by 5 percent, 7 percent, and World War II baby boom between the years 1946 and 1964. 30 percent, respectively.” 18 For example, the Medicare Trustees estimate hospital The measures of life expectancy and mortality rate are not 15 inpatient admissions per beneficiary will decline through interchangeable. However, the two measures are closely 2022 and begin increasing later in the projection period with related. The National Center for Health Statistics life the aging of the baby-boom population (Boards of Trustees expectancy estimate represents the average number of years 2014). CBO also projects comparatively slow growth in per of life remaining if a group of persons were to experience the beneficiary spending for the next decade (2015 to 2025) mortality rates for that specific year of calculation over the in part because of the influx of younger beneficiaries, who course of their remaining life. tend to use fewer health care services and therefore lower Medicare’s average spending per beneficiary (Congressional 16 Researchers at the Commonwealth Fund attribute this Budget Office 2015). difference to the effects of the U.S.’s poorer performance on access to care (measured in terms of timeliness and affordability), administrative efficiency (as reported by patients and doctors), and income-related disparities in access to care and quality (Schneider and Squires 2017). March 2019 | Report to the Congress: Medicare Payment Policy 43

74 References Boards of Trustees, Federal Hospital Insurance and Federal Aaron, H. J. 2015. The myth behind America’s deficit. 2014 Supplementary Medical Insurance Trust Funds. 2014. Washington, DC: The Brookings Institution. annual report of the Boards of Trustees of the Federal Hospital Changes in life expectancy by race and hispanic Arias, E. 2016. Insurance and Federal Supplementary Insurance Trust Funds. origin in the United States, 2013–2014. NCHS data brief, no. 244. Washington, DC: Boards of Trustees. Hyattsville, MD: National Center for Health Statistics. Bureau of Labor Statistics, Department of Labor. 2018. May 2017 Auerbach, D. I., and A. L. Kellermann. 2011. A decade of health national occupational employment and wage estimates United care cost growth has wiped out real income gains for an average States. http://www.bls.gov/oes/. Health Affairs 30, no. 9 (September): 1630–1636. US family. Bureau of Labor Statistics, Department of Labor. 2008. May 2007 Baicker, K., and A. Chandra. 2006. The labor market effects of national occupational employment and wage estimates United rising health insurance premiums. Journal of Labor Economics States. http://www.bls.gov/oes/. 24, no. 3: 609–634. Case, A., and A. Deaton. 2017. Mortality and morbidity in the Baily, M. N., and S. E. Holmes. 2015. Serving the best interests Washington, DC: Brookings Institution. 21st century. . Washington, DC: The of retirement savers: Framing the issues Case, A., and A. Deaton. 2015. Rising morbidity and mortality in Brookings Institution. midlife among white non-Hispanic Americans in the 21st century. Baker, L. C., M. K. Bundorf, and D. P. Kessler. 2014a. Vertical Proceedings of the National Academy of Sciences of the United integration: Hospital ownership of physician practices is States of America 112, no. 49 (December 8): 15078–15083. associated with higher prices and spending. Health Affairs 33, no. Census Bureau. 2018. Historical income tables. https://www. 5 (May): 756–763. census.gov/data/tables/time-series/demo/income-poverty/ Baker, L. C., M. K. Bundorf, A. B. Royalty, et al. 2014b. historical-income-households.html. Physician practice competition and prices paid by private insurers Census Bureau. 2014. National population projections: Journal of the American Medical Association for office visits. Downloadable files: Table 1. Projected population by single year 312, no. 16 (October 22–29): 1653–1662. of age, sex, race, and Hispanic origin for the United States, 2012 Beltran-Sanchez, H., C. E. Finch, and E. M. Crimmins. 2015. to 2060: Middle series. https://www.census.gov/data/tables/2014/ Twentieth century surge of excess adult male mortality. demo/popproj/2014-summary-tables.html. Proceedings of the National Academy of Sciences of the United Centers for Medicare & Medicaid Services, Department of States of America 112, no. 29 (July 21): 8993–8998. Health and Human Services. 2018a. National health expenditures, Boards of Trustees, Federal Hospital Insurance and Federal 1960–2017 historical data. https://www.cms.gov/Research- 2018 Supplementary Medical Insurance Trust Funds. 2018. Statistics-Data-and-Systems/Statistics-Trends-and-Reports/ annual report of the Boards of Trustees of the Federal Hospital NationalHealthExpendData/NationalHealthAccountsHistorical. Insurance and Federal Supplementary Insurance Trust Funds. html. Washington, DC: Boards of Trustees. Centers for Medicare & Medicaid Services, Department of Boards of Trustees, Federal Hospital Insurance and Federal Health and Human Services. 2018b. National health expenditures, Supplementary Medical Insurance Trust Funds. 2017. 2017 projections 2017–2026. https://www.cms.gov/Research- annual report of the Boards of Trustees of the Federal Hospital Statistics-Data-and-Systems/Statistics-Trends-and-Reports/ Insurance and Federal Supplementary Insurance Trust Funds. NationalHealthExpendData/NationalHealthAccountsProjected. Washington, DC: Boards of Trustees. html. Boards of Trustees, Federal Hospital Insurance and Federal Centers for Medicare & Medicaid Services, Department of 2016 Supplementary Medical Insurance Trust Funds. 2016. Health and Human Services. 2017. National health expenditures, annual report of the Boards of Trustees of the Federal Hospital 1960–2016 historical data. http://www.cms.gov/Research- Insurance and Federal Supplementary Insurance Trust Funds. Statistics-Data-and-Systems/Statistics-Trends-and-Reports/ Washington, DC: Boards of Trustees. NationalHealthExpendData/NationalHealthAccountsHistorical. html. Context for Medicare payment policy 44

75 Centers for Medicare & Medicaid Services, Department of Cutler, D. M., and N. R. Sahni. 2013. If slow rate of health care Health and Human Services. 2016. 2016 actuarial report on the spending growth persists, projections may be off by $770 billion. financial outlook for Medicaid. Baltimore, MD: CMS. https:// Health Affairs 32, no. 5 (May): 841–850. www.cms.gov/Research-Statistics-Data-and-Systems/Research/ Mirror, mirror on Davis, K., K. Stremikis, D. Squires, et al. 2014. ActuarialStudies/Downloads/MedicaidReport2016.pdf. the wall, 2014 update: How the U.S. health care system compares Centers for Medicare & Medicaid Services, Department of internationally. New York: Commonwealth Fund. http://www. Health and Human Services. 2015. State level chronic conditions commonwealthfund.org/publications/fund-reports/2014/jun/ report: Medicare utilization and spending. https://www.cms.gov/ mirror-mirror. Research-Statistics-Data-and-Systems/Statistics-Trends-and- Ellison, A. 2016. Hospital M&A activity jumps 70% in 5 years. Reports/Chronic-Conditions/MCC_Main.html. , January 21. Becker’s Hospital Review Centers for Medicare & Medicaid Services, Department of Health FAIR Health. 2016. The opioid crisis among the privately 2014 actuarial report on the financial and Human Services. 2014. insured. New York, NY: FAIR Health. Baltimore, MD: CMS. outlook for Medicaid. Federal Interagency Forum on Aging-Related Statistics. 2016. Chetty, R., M. Stepner, S. Abraham, et al. 2016. The association Washington, Older Americans 2016: Key indicators of well-being. between income and life expectancy in the United States, 2001– DC: Government Printing Office. 2014. Journal of the American Medical Association 315, no. 16 (April 26): 1750–1766. Federal Trade Commission. 2016. Keynote address of FTC Chairwoman Edith Ramirez at the Antitrust in Healthcare Compton, W. M., C. M. Jones, and G. T. Baldwin. 2016. Conference, May 12. https://www.ftc.gov/system/files/ Relationship between nonmedical prescription-opioid use and documents/public_statements/950143/160519antitrusthealthcarek heroin use. 374, no. 2 (January New England Journal of Medicine eynote.pdf. 14): 154–163. Fisher, E. S., D. E. Wennberg, T. A. Stukel, et al. 2003a. The The 2018 long-term budget Congressional Budget Office. 2018a. implications of regional variations in Medicare spending. Part 1: outlook. Washington, DC: CBO. The content, quality, and accessibility of care. Annals of Internal The budget and economic Congressional Budget Office. 2018b. Medicine 138, no. 4 (February 18): 273–287. outlook: 2018 to 2028. Washington, DC: CBO. Fisher, E. S., D. E. Wennberg, T. A. Stukel, et al. 2003b. The Congressional Budget Office. 2016a. Federal subsidies for health implications of regional variations in Medicare spending. Part 2: insurance coverage for people under age 65: 2016 to 2026. Health outcomes and satisfaction with care. Annals of Internal Washington, DC: CBO. Medicine 138, no. 4 (February 18): 288–298. Congressional Budget Office. 2016b. Raise the age of eligibility Gaynor, M., K. Ho, and R. Town. 2014. The industrial Washington, DC: CBO. for Medicare to 67. organization of health care markets. NBER working paper no. 19800. Cambridge, MA: National Bureau of Economic Research. The budget and economic Congressional Budget Office. 2015. outlook: 2015 to 2025. Washington, DC: CBO. The impact of hospital Gaynor, M., and R. Town. 2012. The Synthesis Project, policy brief no. consolidation — Update. That price ain’t Cooper, Z., S. V. Craig, M. Gaynor, et al. 2018. 9. Princeton, NJ: Robert Wood Johnson Foundation. right? Hospital prices and health spending on the privately NBER working paper no. 21815. Cambridge, MA: insured. Medicare: Increasing Government Accountability Office. 2015. National Bureau of Economic Research. hospital physician consolidation highlights need for payment reform. GAO–16–189. Washington, DC: GAO. Cooper, Z., S. V. Craig, M. Gaynor, et al. 2015. That price ain’t right? Hospital prices and health spending on the privately Government Accountability Office. 2013. High-risk series: An insured. NBER working paper no. 21815. Cambridge, MA: update. Washington, DC: GAO. National Bureau of Economic Research. Gruber, J. 2000. Health insurance and the labor market. In Cuckler, G. A., A. M. Sisko, J. A. Poisal, et al. 2018. National Handbook of Health Economics , vol. 1, edited by A. J. Culyer and Health Expenditure projections, 2017–26: Despite uncertainty, J. P. Newhouse. New York, NY: Elsevier Science. 37, fundamentals primarily drive spending growth. Health Affairs no. 3 (March): 482–492. March 2019 | Report to the Congress: Medicare Payment Policy 45

76 Prescription drug spending Hanna, C., and C. Uccello. 2018. Medicare Payment Advisory Commission. 2016. Report to . in the U.S. health care system: An actuarial perspective the Congress: Medicare and the health care delivery system. Washington, DC: American Academy of Actuaries. Washington, DC: MedPAC. 2016 health care cost and Health Care Cost Institute. 2018. Medicare Payment Advisory Commission. 2015. Report to the utilization report. Washington, DC: HCCI. Congress: Medicare payment policy. Washington, DC: MedPAC. 2015 health care cost and Health Care Cost Institute. 2016. Medicare Payment Advisory Commission. 2014a. A data book: Washington, DC: HCCI. utilization report. Washington, Health care spending and the Medicare program. DC: MedPAC. 2014 health care cost and Health Care Cost Institute. 2015. Washington, DC: HCCI. utilization report. Medicare Payment Advisory Commission. 2014b. Report to the Congress: Medicare and the health care delivery system. 2013 health care cost and Health Care Cost Institute. 2014. Washington, DC: MedPAC. . Washington, DC: HCCI. utilization report Medicare Payment Advisory Commission. 2013. Report to the The Holahan, J., L. Blumberg, and L. Clemans-Cope. 2017. Congress: Medicare payment policy. Washington, DC: MedPAC. evidence on recent health care spending growth and the impact of the Affordable Care Act. Washington, DC: Urban Institute. Medicare Payment Advisory Commission. 2012. Report to the Congress: Medicare and the health care delivery system. Hoyert, D. L. 2012. 75 years of mortality in the United States, Washington, DC: MedPAC. 1935–2010. NCHS data brief, no. 88. Hyattsville, MD: National Center for Health Statistics. Medicare Payment Advisory Commission. 2011a. Moving forward from the sustainable growth rate (SGR) system. Letter to The health care services Irving Levin Associates Inc. 2016. the Congress. October 14. acquisition report: 22nd edition. Norwalk, CT: Irving Levin Associates Inc. Medicare Payment Advisory Commission. 2011b. Report to the Congress: Regional variation in Medicare service use. Kaiser Family Foundation and Health Research & Educational Washington, DC: MedPAC. Trust. 2018. Employer health benefits: 2018 annual survey. Menlo Park, CA: Kaiser Family Foundation/HRET. Medicare Payment Advisory Commission. 2009. Report to the . Washington, DC: MedPAC. Congress: Medicare payment policy Kane, C. K. 2015. Updated data on physician practice Chicago, IL: arrangements: Inching toward hospital ownership. Report to the Medicare Payment Advisory Commission. 2008. American Medical Association. Washington, DC: Congress: Reforming the delivery system. MedPAC. Leading Kochanek, K., E. Arias, and R. N. Anderson. 2015. causes of death contributing to decrease in life expectancy gap Medicare Payment Advisory Commission. 2007. Report to the between Black and White populations: United States, 1999–2013. Congress: Promoting greater efficiency in Medicare. Washington, Data brief, no. 218. Hyattsville, MD: National Center for Health DC: MedPAC. Statistics. Milusheva, S., and G. Burtless. 2012. Effects of employer health Martin, A. B., M. Hartman, B. Washington, et al. 2018. National costs on the trend and distribution of Social-Security-taxable health spending in 2017: Growth slows to post-Great Recession Washington, DC: The Brookings Institution. wages. rates; Share of GDP stabilizes. Health Affairs (December 6): Montez, J. K., A. Zajacova, and M. D. Hayward. 2016. Published ahead of print. Explaining inequalities in women’s mortality between U.S. states. Report to Medicare Payment Advisory Commission. 2018. 2 (December): 561–571. SSM Population Health the Congress: Medicare and the health care delivery system. National Center for Health Statistics, Centers for Disease Control Washington, DC: MedPAC. and Prevention, Department of Health and Human Services. 2018. Medicare Payment Advisory Commission. 2017. Report to U.S. Small-area Life Expectancy Estimates Project: USALEEP. the Congress: Medicare and the health care delivery system. Hyattsville, MD: NCHS. Washington, DC: MedPAC. Context for Medicare payment policy 46

77 National Center for Health Statistics, Department of Health and Singh, G. K., and M. Siahpush. 2014. Widening rural–urban Health, United States, 2014: With special Human Services. 2015. disparities in life expectancy, U.S., 1969–2009. American Journal . Hyattsville, MD: NCHS. feature on prescription drugs of Preventative Medicine 46, no. 2 (February): e19–e29. Robinson, J. C., and K. Miller. 2014. Total expenditures per Income of the population Social Security Administration. 2016. patient in hospital-owned and physician-owned physician Baltimore, MD: SSA. http://www.ssa.gov/policy/ 55 or older. Journal of the American Medical organizations in California. docs/statcomps/income_pop55/index.html. Association 312, no. 16 (October 22–29): 1663–1669. Spitalnic, P. 2016. The 2016 Medicare trustees report: One year Rudd, R. A., N. Aleshire, J. E. Zibbell, et al. 2016. Increases in closer to IPAB cuts? Keynote address at Brookings Institution drug and opioid overdose deaths—United States, 2000–2014. event at the Brookings Institution, June 23. https://www. Morbidity and Mortality Weekly Report 64, no. 50–51 (January brookings.edu/events/the-2016-medicare-trustees-report-one- 01): 1378–1382. year-closer-to-ipab-cuts/. Scheffler, R. M., D. R. Arnold, and C. M. Whaley. 2018. Stensland, J., Z. R. Gaumer, and M. E. Miller. 2010. Private-payer Consolidation trends in California’s health care system: Impacts 29, Health Affairs profits can induce negative Medicare margins. on ACA premiums and outpatient visit prices. Health Affairs 37, no. 5 (May): 1045–1051. no. 9 (September): 1409–1416. Health policy brief: Medicaid primary care Tollen, L. 2015. Schneider, E. C., and D. Squires. 2017. From last to first: Could Updated May 15. Washington, DC: Health Affairs/Robert parity. the U.S. health care system become the best in the world? New Wood Johnson Foundation. England Journal of Medicine 377, no. 10 (September 7): 901– White, C., and V. Y. Wu. 2014. How do hospitals cope with 904. Health Services sustained slow growth in Medicare prices? Schwartz, A., B. Landon, A. Elshaug, et al. 2014. Measuring 49, no. 1 (February): 11–31. Research 174, no. 7 low-value care in Medicare. JAMA Internal Medicine Zolot, J. 2017. U.S. life expectancy varies depending on county of (July): 1067–1076. 117, no. 8 (August): 15. birth. American Journal of Nursing Selden, T. M., Z. Karaca, P. Keenan, et al. 2015. The growing difference between public and private payment rates for inpatient hospital care. 34, no. 12 (December 1): 2147–2150. Health Affairs March 2019 | Report to the Congress: Medicare Payment Policy 47

78

79 CHAPTER 2 Assessing payment adequacy and updating payments in fee-for-service Medicare

80

81 CHAPTER 2 Assessing payment adequacy and updating payments in fee-for-service Medicare In this chapter Chapter summary As required by law, the Commission annually makes payment update Are Medicare payments • recommendations for providers paid under fee-for-service (FFS) Medicare. An adequate in 2019? update is the amount (usually expressed as a percentage change) by which the • What cost changes are base payment for all providers in a payment system is changed relative to the expected in 2020? prior year. To determine an update, we first assess the adequacy of Medicare payments for providers in the current year (2019) by considering beneficiaries’ How should Medicare • access to care, the quality of care, providers’ access to capital, and Medicare payments change in 2020? payments and providers’ costs. Next, we assess how those providers’ costs • Payment adequacy in are likely to change in the year the update will take effect (the policy year, context 2020). As part of the process, we examine payments to support the efficient delivery of services, consistent with our statutory mandate. Finally, we make a judgment about what, if any, update is needed. (The Commission also assesses Medicare payment systems for Part C and Part D and makes recommendations as appropriate. But because they are not FFS payment systems, they are not part of the discussion in this chapter.) This year, we consider recommendations in nine FFS sectors: acute care hospitals, physicians and other health professionals, ambulatory surgical centers, outpatient dialysis facilities, skilled nursing facilities, home health care agencies, inpatient rehabilitation facilities, long-term care hospitals, and hospices. Each year, the Commission looks at all available indicators of March 2019 | Report to the Congress: Medicare Payment Policy 51

82 payment adequacy and reevaluates any assumptions from prior years, using the most recent data available to make sure its recommendations accurately reflect current conditions. We may also consider recommending changes that redistribute payments within a payment system to correct any biases that may make treating patients with certain conditions financially undesirable, make particular procedures unusually profitable, or otherwise result in inequity among providers. Finally, we may also make recommendations to improve program integrity. Our recommendations, if enacted, could significantly change the revenues providers receive from Medicare. Rates set to cover the costs of relatively efficient providers help create fiscal pressure on all providers to control their costs. Medicare rates also have broader implications for health care spending. For example, Medicare rates are commonly used to set hospital rates charged to uninsured patients eligible for financial assistance, used by Medicare Advantage plans to set hospital prices, and used by the Department of Veterans Affairs (VA) to pay non-VA providers (Department of Veterans Affairs 2010, Internal Revenue Service 2014, Medicare Payment Advisory Commission 2013). The Commission also examines payment rates for services that can be provided in multiple settings. Medicare often pays different amounts for similar services across settings. Basing the payment on the rate in the most efficient setting would save money for Medicare, reduce cost sharing for beneficiaries, and reduce the financial incentive to provide services in the higher paid setting. However, putting into practice the principle of paying the same rate for the same service across settings can be complex because it requires that the definition of the services and the characteristics of the beneficiaries be sufficiently similar across settings. In March 2012, we recommended equalizing rates for evaluation and management office visits provided in hospital outpatient departments and physicians’ offices (Medicare Payment Advisory Commission 2012). In 2014, we extended that recommendation to additional services provided in those two settings and recommended consistent payment between acute care hospitals and long-term care hospitals for certain classes of patients (Medicare Payment Advisory Commission 2014). In the Bipartisan Budget Act of 2015, the Congress made payment to outpatient departments for certain services equal to the physician fee schedule rates for those same services provided at any new outpatient off-campus location beginning in 2018. In 2016, to make payments across all of the post-acute care payment settings comparable, the Commission recommended elements of a single prospective payment system (PPS) for all post-acute care to replace the four independent PPSs Assessing payment adequacy and updating payments in fee-for-service Medicare 52

83 in use today (the skilled nursing facility, inpatient rehabilitation facility, long-term care hospital, and home health PPSs) (Medicare Payment Advisory Commission 2016). Most recently, in 2018, we recommended blending setting-specific and unified post-acute care PPS relative weights to help transition to a unified system (Medicare Payment Advisory Commission 2018). The Commission will continue to analyze opportunities for applying this principle to other services and settings. ■ March 2019 | Report to the Congress: Medicare Payment Policy 53

84

85 geographic regions or providers. For example, in reaction Background to patterns of unusually long stays in a subset of hospices, we recommended medical review focused on hospices that The goal of Medicare payment policy should be to obtain have many long-stay patients. In 2016, we recommended good value for the program’s expenditures, which means the Secretary closely examine the coding practices of maintaining beneficiaries’ access to high-quality services certain inpatient rehabilitation facilities that appear to while encouraging efficient use of resources. Anything result in very high Medicare margins. less does not serve the interests of the taxpayers and We compare our recommendations for updates and beneficiaries who finance Medicare through their taxes other policy changes for 2020 with the base payment and premiums. Steps toward this goal involve: rates specified in law to understand the implications for • setting the base payment rate (i.e., the payment for beneficiaries, providers, and the Medicare program. As has services of average complexity) at the right level; been the Commission’s policy in the past, we consider our recommendations each year in light of the most current • developing payment adjustments that accurately data and, in general, recommend updates for a single year. reflect market, service, and patient cost differences beyond providers’ control; • adjusting payments for quality; and Are Medicare payments adequate in considering the need for annual payment updates and • 2019? other policy changes. The first part of the Commission’s approach to developing To help determine the appropriate base payment rate for a payment updates is to assess the adequacy of current given payment system in 2020, we first consider whether Medicare payments. For each sector, we make a judgment payments are adequate for relatively efficient providers in by examining information on the following: 2019. To inform the Commission’s judgment, we examine the most recent available data on beneficiaries’ access to beneficiaries’ access to care • care, the quality of care, and providers’ access to capital, • quality of care as well as projected Medicare payments and providers’ costs for 2019. We then consider how providers’ costs • providers’ access to capital will change in 2020. Taking these factors into account, we recommend how Medicare payments for the sector in • Medicare payments and providers’ costs for 2019 aggregate should change in 2020. Some measures focus on beneficiaries (e.g., access to Within a given level of funding for a sector, we may also care) and some focus on providers (e.g., the relationship consider changes in payment policy to improve relative between payments and costs). The direct relevance, payment accuracy across patients and procedures. Such availability, and quality of each type of information changes are intended to improve equity among providers vary among sectors, and no single measure provides all or access to care for beneficiaries and may also affect the information needed for the Commission to judge the distribution of payments among providers in a sector. payment adequacy. Ultimately, the Commission makes its For example, in 2018, the Commission recommended recommendations considering all of these factors. that CMS use a blend of the setting-specific relative Beneficiaries’ access to care weights and the unified post-acute care (PAC)–prospective payment system (PPS) relative weights for each of the four Access to care is an important indicator of the willingness PAC settings to redistribute payments within each setting of providers to serve Medicare beneficiaries and the toward medically complex patients (Medicare Payment adequacy of Medicare payments. For example, poor Advisory Commission 2018). access could indicate that Medicare payments are too low. However, factors unrelated to Medicare’s payment policies We also make recommendations to improve program may also affect access to care. These factors include integrity when needed. In some cases, our data analysis coverage policies, beneficiaries’ preferences, local market reveals problematic variation in service utilization across conditions, and supplemental insurance. March 2019 | Report to the Congress: Medicare Payment Policy 55

86 questions about program integrity or whether the definition The measures we use to assess beneficiaries’ access of the corresponding benefit is too vague. Reductions to care depend on the availability and relevance of in the volume of services can sometimes be a signal information in each sector. We use results from several that revenues are inadequate for providers to continue surveys to assess the willingness of physicians and operating or to provide the same level of service. Finally, other health professionals to serve beneficiaries and rapid changes in volume between sectors whose services beneficiaries’ opinions about their access to physician can be substituted for one another may suggest distortions and other health professional services. For home health in payment and raise questions about provider equity. For services, we examine data on whether communities are example, payment rates for evaluation and management served by providers. (E&M) office visits are much higher in hospital outpatient Access: Capacity and supply of providers departments (HOPDs) than in physicians’ offices, and over the last several years, the volume of those services Rapid growth in the capacity of providers to furnish in HOPDs has increased while the volume in physicians’ care may increase beneficiaries’ access and indicate that offices has decreased. payments are more than adequate to cover providers’ costs. Changes in technology and practice patterns may also affect However, changes in the volume of services are not providers’ capacity. For example, less invasive procedures direct indicators of access; increases and decreases can could be performed in outpatient settings, and lower priced be explained by other factors such as population changes, equipment could be more easily purchased by providers, changes in disease prevalence among beneficiaries, increasing the capacity to provide certain services. technology, practice patterns, deliberate policy interventions, and beneficiaries’ preferences. For example, Substantial increases in the number of providers may the number of Medicare beneficiaries in the traditional suggest that payments are more than adequate and could fee-for-service (FFS) program varies from year to year; raise concerns about the value of the services being therefore, we look at the volume of services per FFS furnished. If Medicare is not the dominant payer for a beneficiary as well as the total volume of services. Explicit given provider type (such as ambulatory surgical centers), policy decisions can also influence volume. For example, changes in the number of providers may be influenced during fiscal year 2016, CMS began phasing in a policy more by other payers and their demand for services and that lowers payments for certain LTCH cases. As a result, thus may be difficult to relate to Medicare payments. When LTCHs—as expected—changed their admitting practices facilities close, we try to distinguish between closures that largely in response to the implementation of the policy, have serious implications for access to care in a community and the number of LTCH discharges decreased markedly. and those that may have resulted from excess capacity. For example, in 2016, Medicare’s payment rates for certain Changes in the volume of physician services must be cases in long-term care hospitals (LTCHs) decreased interpreted particularly cautiously. Evidence suggests significantly, and about 40 LTCHs closed—nearly 10 that for discretionary services, volume may go up when percent of LTCH facilities and beds. However, the closures payment rates go down—the so-called volume offset. primarily occurred in market areas with multiple LTCHs, Whether a volume offset phenomenon exists in other and overall LTCH occupancy declined during the same time sectors depends on how discretionary the services are period—indicating adequate capacity. and on the ability of providers to influence beneficiaries’ demand for them. Access: Volume of services The volume of services can be an indirect indicator of Access: Marginal profit beneficiary access to services. An increase in volume Another factor we consider when evaluating access to shows that beneficiaries are receiving more services care is whether providers have a financial incentive to and suggests sufficient access—although it does not expand the number of Medicare beneficiaries they serve. necessarily demonstrate that the services are appropriate. In considering whether to treat a patient, a provider with Volume is also an indicator of payment adequacy; an excess capacity compares the marginal revenue it will increase in volume beyond what would be expected receive (e.g., the Medicare payment) with its marginal relative to the increase in the number of beneficiaries could costs—that is, the costs that vary with volume. If Medicare suggest that Medicare’s payment rates are too high. Very payments are larger than the marginal costs of treating an rapid increases in the volume of a service might even raise Assessing payment adequacy and updating payments in fee-for-service Medicare 56

87 and cost measures for use in a hospital value incentive additional beneficiary, a provider has a financial incentive program (HVIP). to increase its volume of Medicare patients. In contrast, if payments do not cover the marginal costs, the provider Providers’ access to capital may have a disincentive to care for Medicare beneficiaries. We note, however, that in instances in which a sector does Providers must have access to capital to maintain and not have substantial excess capacity or in which Medicare modernize their facilities and to improve their capability composes a dominant share of a sector’s patients, marginal to deliver patient care. Widespread ability to access capital profit may be a less useful indicator of access to care. throughout a sector may reflect the adequacy of Medicare payments. Some sectors such as hospitals require large Quality of care capital investments, and access to capital can be a useful indicator. Other sectors such as home health care do not The relationship between the quality of care and the need large capital investments, so access to capital is a adequacy of Medicare payment is not direct. Simply more limited indicator. In some cases, a broader measure increasing payments through an update for all providers in such as changes in employment may be a useful indicator a sector, regardless of their individual quality, is unlikely of financial health within a sector. Similarly, in sectors to influence the quality of care because, historically, where providers derive most of their payments from other Medicare payment systems have created little or no payers (such as ambulatory surgical centers) or other lines incentive for providers to spend additional resources on of business, or when conditions in the credit markets are improving quality. extreme, access to capital may be a limited indicator of the The Medicare program has begun to implement quality- adequacy of Medicare payments. based payment policies in a number of sectors; however, One indicator of a sector’s access to capital is its all- some issues have arisen. First, it is very difficult to payer profitability reflecting income from all sources. We differentiate quality performance among providers when refer to this amount as the sector’s total margin, which is the number of cases per provider is low. This issue has calculated as aggregate income, minus costs, divided by been particularly vexing in measuring quality performance income. Total margins can inform our assessment of a for individual clinicians. Second, the Commission has sector’s overall financial condition and hence its access to been increasingly concerned that Medicare’s approach capital. to quality measurement is flawed because it relies on too many clinical process measures. Many current Medicare payments and providers’ costs for process measures are weakly correlated with outcomes 2019 of interest such as mortality and readmissions, and most process measures focus on addressing the underuse of For most payment sectors, we estimate Medicare services, while the Commission believes that overuse payments and providers’ costs for 2019 to inform our and inappropriate use are also of concern. Third, reliance update recommendations for 2020. To maintain Medicare on provider-reported measures can create a burden on beneficiaries’ access to high-quality care while keeping providers and can lead to biased reporting in response to financial pressure on providers to make better use of strong financial incentives. As an example of the latter, taxpayers’ and beneficiaries’ resources, we investigate since 2014, home health agencies reported improvements whether payments are adequate to cover the costs of in provider-reported measures such as transferring and relatively efficient providers, where available data permit walking, even though more objective, claims-based such providers to be defined. outcome measures (such as the use of emergency Relatively efficient providers use fewer inputs to produce department care and hospital admissions) have not quality outputs. Efficiency could be increased by using improved or have worsened. the same inputs to produce a higher quality output or by As an alternative approach, we have begun exploring the using fewer inputs to produce the same quality output. The use of a small set of population-based outcome measures Commission follows two principles when selecting a set of to assess and compare the performance of FFS Medicare, efficient providers. First, the providers must do relatively Medicare Advantage, and Medicare accountable care well on cost and quality metrics. Second, the performance organizations within a local area. For example, in Chapter has to be consistent, meaning that the provider cannot 15, we discuss a small set of outcome, patient experience, have poor performance on any metric over the past three March 2019 | Report to the Congress: Medicare Payment Policy 57

88 inpatient and outpatient payments; the updates for other years. The Commission’s approach is to develop a set of distinct units of the hospital, such as SNFs, are covered in criteria and then examine how many providers meet those separate chapters. criteria. It does not establish a set share of providers to be considered efficient and then define criteria to meet that The adequacy of Medicare payments is assessed relative pool size. to the costs of treating Medicare beneficiaries, and the For providers that submit cost reports to CMS—acute Commission’s recommendations address a sector’s care hospitals, skilled nursing facilities (SNFs), home Medicare payments, not total payments. We calculate health agencies, outpatient dialysis facilities, inpatient a sector’s Medicare margin to determine whether total rehabilitation facilities (IRFs), LTCHs, and hospices—we Medicare payments cover average providers’ costs for estimate total Medicare-allowable costs and assess the treating Medicare patients and to inform our judgment relationship between Medicare’s payments and those costs. about payment adequacy. Margins will always be distributed We typically express the relationship between payments around the average, and aggregate payment adequacy and costs as a payment margin, which is calculated as does not mean that every provider has a positive Medicare aggregate Medicare payments for a sector, minus costs, margin. To assess whether changes are needed in the divided by payments. By this measure, if costs increase distribution of payments, we calculate Medicare margins faster than payments, margins will decrease. for certain subgroups of providers with unique roles in the health care system. For example, because location and In general, to estimate payments, we first apply the annual teaching status enter into the payment formula, we calculate payment updates specified in law for 2018 and 2019 to Medicare margins based on where hospitals are located our base data (2017 for most sectors). We then model the (in urban or rural areas) and their teaching status (major effects of other policy changes that will affect the level of teaching, other teaching, or nonteaching). payments in 2019. To estimate 2019 costs, we consider the rate of input price inflation or historical cost growth, and, Multiple factors can contribute to changes in the Medicare as appropriate, we adjust for changes in the product (such margin, including changes in the efficiency of providers, as fewer visits per episode of home health care) and trends changes in coding that may change case-mix adjustment, in key indicators (such as historical cost growth and the and other changes in the product (e.g., reduced lengths of distribution of cost growth among providers). stay at inpatient hospitals). Knowing whether these factors have contributed to margin changes may inform decisions Use of margins about whether and how much to change payments. In most cases, we assess Medicare margins for the In sectors where the data are available, the Commission services furnished in a single sector and covered by makes a judgment when assessing the adequacy of payments a specific payment system (e.g., SNF or home health relative to costs. No single standard governs this relationship services). However, in the case of hospitals, which often for all sectors, and margins are only one indicator for provide services that are paid for by multiple Medicare determining payment adequacy. Moreover, although payment systems, our measures of payments and costs payments can be ascertained with some accuracy, there for an individual sector could become distorted because may be no “true” value for reported costs, which reflect of the allocation of overhead costs or the presence of accounting choices made by providers (such as allocations complementary services. For example, having a hospital- of costs to different services) and the relationship of service based SNF or IRF may allow a hospital to achieve shorter volume to capacity in a given year. Further, even if costs lengths of stay in its acute care units, thereby decreasing are accurately reported, they reflect strategic investment costs and increasing inpatient margins. For hospitals, we decisions of individual providers, and Medicare—as a assess the adequacy of payments for the whole range of prudent payer—may choose not to recognize some of Medicare services they furnish—inpatient and outpatient these costs or may exert financial pressure on providers to (which together account for more than 90 percent of encourage them to reduce their costs. Medicare payments to hospitals), SNF, home health, psychiatric, and rehabilitation services—and compute Appropriateness of current costs an overall Medicare hospital margin encompassing Our assessment of the relationship between Medicare’s costs and payments for all the sectors. The hospital payments and providers’ costs is complicated by update recommendation in Chapter 3 applies to hospital differences in providers’ efficiency, responses to changes Assessing payment adequacy and updating payments in fee-for-service Medicare 58

89 in payment systems, product changes, and cost reporting payers. In some sectors, Medicare itself could, and should, accuracy. Measuring the appropriateness of costs is exert greater pressure on providers to reduce costs. particularly difficult in new payment systems because Variation in cost growth among a sector’s providers can changes in response to the incentives in the new system give us insight into the range of performance that facilities are to be expected. For example, the number and types can achieve. For example, if some providers’ costs grow of visits in a home health episode changed significantly more rapidly than others in a given sector, we might after the home health PPS was introduced, although question whether those rapid increases are appropriate. the payments were based on the older, higher level of Changes in product can also significantly affect unit costs. use and costs. In other systems, coding may change. Returning to the example of home health services, one As an example, the hospital inpatient PPS introduced would expect that substantial reductions in the number of a new patient classification system in 2008 to improve visits per 60-day home health episode would reduce costs payment accuracy. However, for a number of years after per episode. If costs per episode instead were to increase its implementation, it resulted in higher payments because while the number of visits were to decrease, one would provider coding became more detailed, making patient question the appropriateness of the cost growth and not complexity appear higher—although the underlying increase Medicare payments in response. patient complexity was largely unchanged. Any kind of rapid change in policy, technology, or product can make it In summary, Medicare payment policy should not be difficult to measure costs per unit. designed simply to accommodate whatever level of cost growth a sector demonstrates. Cost growth can oscillate To assess whether reported costs reflect the costs of from year to year depending on factors such as economic efficient providers, we examine recent trends in the conditions and relative market power. Payment policy average cost per unit, variation in standardized costs should accommodate cost growth only after taking into and cost growth, and evidence of change in the product. account a broad set of payment adequacy indicators, One issue Medicare faces is the extent to which private including the current level of Medicare payments. payers exert pressure on providers to constrain costs. If private payers do not exert pressure, providers’ costs will increase and, all other things being equal, margins on Medicare patients will decrease. Providers who are What cost changes are expected in under pressure to constrain costs generally have managed 2020? to slow their growth in costs more than those who face less pressure (Medicare Payment Advisory Commission The second part of the Commission’s approach to 2011, Robinson 2011, White and Wu 2014). Some have developing payment update recommendations is to suggested that, in the hospital sector, costs are largely consider anticipated policy and cost changes in the next outside the control of hospitals and that hospitals shift payment year. For each sector, we review evidence about costs onto private insurers to offset Medicare losses. This the factors that are expected to affect providers’ costs. belief assumes that costs are immutable and not influenced One factor is the change in input prices, as measured by whether the hospital is under financial pressure. We by the price index that CMS uses for that sector. (These find that costs do vary in response to financial pressure indexes are estimated quarterly; we use the most recent and that low margins on Medicare patients can result from estimate available when we do our analyses.) For facility a high cost structure that has developed in reaction to providers, we start with the forecasted increase in an high private-payer rates. In other words, when providers industry-specific index of national input prices, called a (particularly not-for-profit providers) receive high “market basket index.” For physician services, we start payment rates from insurers, they face less pressure to with a CMS-derived weighted average of price changes keep their costs low, and so, all other things being equal, for inputs used to provide physician services. Forecasts their Medicare margins are low because their costs are of these indexes approximate how much providers’ costs high. (For-profit providers may prefer to keep costs low to would change in the coming year if the quality and mix of maximize returns to stockholders and, indeed, often have inputs they use to furnish care remained constant—that is, higher Medicare margins than similar nonprofit providers.) if there were no change in efficiency. Other factors may Lack of pressure is more common in markets where a few include the trend in actual cost growth, which could be providers dominate and have negotiating leverage over March 2019 | Report to the Congress: Medicare Payment Policy 59

90 used to inform our estimate if it differs significantly from existing incentives to choose a site of care based on the projected market basket. payment considerations. The difficulty of harmonizing payments across sectors to remove inappropriate incentives illustrates one weakness of FFS payment systems specific to each provider type and highlights How should Medicare payments change the importance of moving beyond FFS to more global in 2020? and patient-centric payment systems. As we continue to support moving Medicare payment systems toward those The Commission’s judgments about payment adequacy, approaches, we will also continue to look for opportunities forthcoming policy changes, and expected cost changes to rationalize payments for specific services across sectors result in an update recommendation for each payment to approximate paying the costs of the most efficient system. An update is the amount (usually expressed as sector and lessen financial incentives to prefer one sector a percentage change) by which the base payment for all over another. Our June 2016 report on a unified PAC providers in a payment system is changed relative to the PPS addressed these issues directly (Medicare Payment prior year. In considering updates, the Commission makes Advisory Commission 2016). its recommendations for 2020 relative to the 2019 base payment as defined in Medicare’s authorizing statute— Consistent payment for the same service Title XVIII of the Social Security Act. The Commission’s across settings recommendations may call for an increase, a decrease, or A beneficiary can sometimes receive a similar service no change from the 2019 base payment. For example, if the in different settings. Depending on which setting the statutory base payment for a sector were $100 in 2019, an beneficiary or the treating clinician chooses, Medicare and update recommendation of a 1 percent increase for a sector the beneficiary may pay different amounts. For example, means that we are recommending that the base payment in when leaving the hospital, patients with joint replacements 2020 for that sector be 1 percent greater, or $101. requiring physical therapy might be discharged with home health care or outpatient therapy, or they might be When our recommendations differ from current law, as discharged to a SNF or IRF, and Medicare payments (and they often do, the Congress and the Secretary of Health beneficiary cost sharing) can differ widely as a result. and Human Services would have to take action and change law or regulation to put them into effect. Each year, we A core principle guiding the Commission is that Medicare look at all available indicators of payment adequacy and should pay the same amount for the same service, even reevaluate prior-year assumptions using the most recent when it is provided in different settings. Putting this data available. The Commission does not start with any principle into practice requires that the definition of presumption that an update is needed or that any increase services in the settings and the characteristics of the in costs should be automatically offset by a payment patients be sufficiently similar. Where these conditions update. Instead, an update (which may be positive, zero, are not met, offsetting adjustments would have to be made or negative) is warranted only if it is supported by the to ensure comparability. Because Medicare’s payment empirical data, in the judgment of the Commission. systems were developed independently and have had different update trajectories, payments for similar services In conjunction with the update recommendations, we can vary widely. Such differences create opportunities may also make recommendations to improve payment for Medicare and beneficiary savings if payment is set at accuracy that might in turn affect the distribution the level applicable to the lowest priced setting in which of payments among providers. These distributional the service can be safely performed. For example, under changes are sometimes, but not always, budget neutral. the current payment systems, a beneficiary can receive Our recommendation to shift payment weights from the same physician visit service in a hospital outpatient therapy to medically complex PAC cases is one example clinic or in a physician’s office. In fact, the same physician of a distributional change that would affect providers could see the same patient and provide the same service, differentially based on their patients’ characteristics. but depending on whether the service is provided in an The Commission, as it makes its update recommendations, outpatient clinic or in a physician’s office, Medicare’s may in some cases take into consideration payment payment and the beneficiary’s coinsurance can differ by 80 differentials across sectors and make sure the relative percent or more. update recommendations for the sectors do not exacerbate Assessing payment adequacy and updating payments in fee-for-service Medicare 60

91 In 2012, the Commission recommended that payments for as a whole. The Commission is concerned by any increase in Medicare spending per beneficiary without a E&M office visits in the outpatient and physician office sectors be made equal. This service is comparable across commensurate increase in value such as higher quality of care or improved health status. Growth in spending per the two settings. Our recommendation sets payment rates for E&M office visits both in the outpatient department and beneficiary, combined with the aging of the baby boomers, physician office sectors equal to those in the physician fee will result in the Medicare program absorbing increasing shares of the gross domestic product and federal spending. schedule, lowering both program spending and beneficiary Medicare’s rising costs are projected to exhaust the liability (Medicare Payment Advisory Commission 2012). Hospital Insurance Trust Fund (which funds Medicare Part In 2014, we extended that principle to additional services for which payment rates in the outpatient PPS should be A) and significantly burden taxpayers. Ensuring that the recent moderate growth trends in Medicare spending per lowered to better match payment rates in the physician office setting (Medicare Payment Advisory Commission beneficiary continue will require vigilance. The financial 2014). In the Bipartisan Budget Act of 2015, the Congress future of Medicare prompts us to look at payment policy made payment for outpatient departments for the same and ask what can be done to develop, implement, and refine payment systems to reward quality and efficient use services equal to the physician fee schedule rates for those services at any new outpatient off-campus clinic beginning of resources while improving payment equity. in 2018. We also recommended consistent payment In many past reports, the Commission has stated that between acute care hospitals and long-term care hospitals Medicare should institute policies that improve the for certain categories of patients (Medicare Payment program’s value to beneficiaries and taxpayers. CMS is Advisory Commission 2014). In 2016, we recommended beginning to take such steps, and we discuss them in the elements of a unified PAC PPS that would make payments sector-specific chapters that follow. Ultimately, increasing based on patients’ needs and characteristics, generally Medicare’s value to beneficiaries and taxpayers requires irrespective of the PAC entity that provided their care knowledge about the costs and health outcomes of (Medicare Payment Advisory Commission 2016). The services. Until more information about the comparative Commission will continue to study other services that are effectiveness of new and existing health care treatments provided in multiple sites of care to find additional services and technologies is available, patients, providers, and the for which the principle of the same payment for the same program will have difficulty determining what constitutes service can be applied. high-quality care and effective use of resources. Budgetary consequences As we examine each of the payment systems, we also look The Medicare Prescription Drug, Improvement, and for opportunities to develop policies that create incentives Modernization Act of 2003 requires the Commission for providing high-quality care efficiently across providers to consider the budgetary consequences of our and over time. Some of the current payment systems create recommendations. Therefore, this report documents how strong incentives for increasing volume, and very few of spending for each recommendation would compare with these systems encourage providers to work together toward expected spending under current law. We also assess common goals. Alternative payment models (e.g., the the effects of our recommendations on beneficiaries Next Generation accountable care organization model) are and providers. Although we recognize budgetary meant to stimulate delivery system reform toward more consequences, our recommendations are not driven by any integrated and value-oriented health care systems and may specific budget target but, instead, reflect our assessment address these issues. In the near term, the Commission will of the level of payment needed to provide adequate access continue to closely examine a broad set of indicators, make to appropriate care. sure there is consistent pressure on providers to control their costs, and set a demanding standard for determining which sectors qualify for a payment update each year. In the longer term, pressure on providers may cause them to Payment adequacy in context increase their participation in alternative payment models. We will continue to contribute to the development of those As discussed in Chapter 1, it is essential to look at models and to increase their efficacy. ■ payment adequacy not only within the context of individual payment systems but also in terms of Medicare March 2019 | Report to the Congress: Medicare Payment Policy 61

92 References Medicare Payment Advisory Commission. 2013. Report to Department of Veterans Affairs. 2010. VA announces use of the Congress: Medicare and the health care delivery system. standard payment rates for some non-VA care. Press release. Washington, DC: MedPAC. December 16. http://www.va.gov/opa/pressrel/pressrelease. cfm?id=2021. Report to the Medicare Payment Advisory Commission. 2012. Congress: Medicare payment policy. Washington, DC: MedPAC. Internal Revenue Service, Department of the Treasury. 2014. Additional requirements for charitable hospitals; community Medicare Payment Advisory Commission. 2011. Report to the health needs assessments for charitable hospitals; requirement Washington, DC: MedPAC. Congress: Medicare payment policy. of a Section 4959 excise tax return and time for filing the return. Final rule. Federal Register 79, no. 250 (December 31): 78954– Robinson, J. 2011. Hospitals respond to Medicare payment 79016. shortfalls by both shifting costs and cutting them, based on 30, no. 7 (July): 1265–1271. market concentration. Health Affairs Medicare Payment Advisory Commission. 2018. Report to the Congress: Medicare payment policy. Washington, DC: MedPAC. White, C., and V. Y. Wu. 2014. How do hospitals cope with sustained slow growth in Medicare prices? Health Services Report to Medicare Payment Advisory Commission. 2016. Research 49, no. 1 (February): 11–31. the Congress: Medicare and the health care delivery system. Washington, DC: MedPAC. Medicare Payment Advisory Commission. 2014. Report to the Washington, DC: MedPAC. Congress: Medicare payment policy. Assessing payment adequacy and updating payments in fee-for-service Medicare 62

93 CHAPTER 3 Hospital inpatient and outpatient services

94 RECOMMENDATION 3 The Congress should: • Replace Medicare’s current hospital quality programs with a new hospital value incentive program (HVIP) that: • includes a small set of population-based outcome, patient experience, and value measures; • scores all hospitals based on the same absolute and prospectively set performance targets; • accounts for differences in patients’ social risk factors by distributing payment adjustments through peer grouping, and • For 2020, update the 2019 Medicare base payment rates for acute care hospitals by 2 percent. The difference between the update recommendation and the amount specified in current law should be used to increase payments in a new HVIP. COMMISSIONER VOTES: YES 17 • NO 0 • NOT VOTING 0 • ABSENT 0

95 CHAPTER 3 Hospital inpatient and outpatient services In this chapter Chapter summary In 2017, the Medicare fee-for-service (FFS) program paid 4,700 hospitals • Are Medicare payments $190 billion consisting of $119 billion for about 10 million Medicare inpatient adequate in 2019? admissions, $66 billion for about 200 million outpatient services, and $6 How should Medicare • billion for uncompensated care provided to non-Medicare patients. On net, payment rates change in inpatient payments increased by $2.6 billion (2.2 percent) and outpatient 2020? payments increased by almost $4.9 billion (8.1 percent). Inpatient payments increased primarily due to a 1 percent increase in payment rates, a slight increase in discharges per capita, and an increase in case mix. Outpatient payments increased due to rapid growth in Part B drug spending, a continued shift in the site of service billing from physician offices to hospital outpatient departments, and an increase in outpatient payment rates. In contrast, payments for uncompensated care decreased by about $0.4 billion. Thus, on net, between 2016 and 2017, overall hospital spending increased $7 billion. Over this same period, hospital spending per FFS beneficiary rose 4.3 percent, increasing from $4,992 to $5,208. Assessment of payment adequacy Most payment adequacy indicators (including access to care, quality of care, and access to capital) are positive. Average Medicare margins continue to be negative, although hospitals with excess capacity still have an incentive to see March 2019 | Report to the Congress: Medicare Payment Policy 65

96 Medicare beneficiaries because Medicare payment rates remain about 8 percent higher than the variable costs associated with Medicare patients. Beneficiaries’ access to care— Access measures for hospital services include the capacity of providers and the volume of services. • Capacity and supply of providers— In 2017, the average hospital occupancy rate was 62.5 percent, suggesting that hospitals have excess inpatient capacity in most markets. Because Medicare payments exceed the marginal cost of providing services, hospitals with excess capacity have a financial incentive to increase services provided to Medicare beneficiaries. Marginal profits were approximately 8 percent on average in 2017. • Volume of services— After declining over several years, inpatient use per beneficiary in 2017 increased by 0.7 percent. Outpatient visits per beneficiary also increased by 0.7 percent, a slower pace of outpatient volume growth than in recent years. Quality of care— From 2013 to 2017, hospital mortality and readmission rates improved slowly. Patient satisfaction also improved somewhat: The share of patients who rated their hospital a 9 or 10 on a 10-point scale increased from 71 percent to 73 percent. Providers’ access to capital— Access to bond markets has been strong, with hospital bond offerings in 2015, 2016, and 2017 of $24 billion, $38 billion, and $35 billion, respectively. While some hospitals struggle with low occupancy and limited access to capital, most hospitals have good access to capital because of strong all-payer profit margins. All-payer margins were 7.1 percent in 2017, only 0.1 percentage point below their all-time high of 7.2 percent in 2013. Medicare payments and providers’ costs— In 2017, hospitals’ aggregate Medicare margin was −9.9 percent, down slightly from –9.7 percent in 2016. The profit margin for relatively efficient providers was about –2 percent. The decline in margins from 2016 to 2017 was primarily due to a decline in supplemental payments for uncompensated care and health information technology. Patient care margins, which exclude uncompensated care payments, increased slightly since 2016 due to a large increase in spending on Part B drugs, which have higher profit margins (in part due to the 340B program) than other hospital services. We project that the overall Medicare margin will decline to about –11 percent in 2019. How should payment rates change in 2020? For 2020, the Commission recommends that the Congress update Medicare inpatient and outpatient payment rates by 2 percent. This update recommendation Hospital inpatient and outpatient services: Assessing payment adequacy and updating payments 66

97 is based on indicators of beneficiaries’ access to hospital care, hospitals’ access to capital, hospital quality, and the relationship between Medicare payments and hospital costs. As we discuss in Chapter 15, the Commission is also recommending a new hospital value incentive program (HVIP) that aligns with the Commission’s principles for quality measurement and replaces the current quality incentive programs. The difference between the 2 percent update and the update amount specified in current law (expected to be 2.8 percent) should be used to increase payments in the new HVIP. Together, these recommendations would increase hospital payments by increasing the base payment rate and by increasing the average rewards hospitals receive under the proposed Medicare HVIP. On net, the 2 percent update, the expected increase in the inpatient HVIP rewards (expected to be equal to 0.8 percent of all payments), and the elimination of the inpatient penalties in the current quality programs (equal to 0.5 percent of all payments) would be expected to increase hospital payment rates by an average of 3.3 percent. ■ March 2019 | Report to the Congress: Medicare Payment Policy 67

98

99 TABLE Growth in Medicare inpatient and outpatient spending 3–1 Percent Average annual change percent change 2016–2017 2007–2016 2016 2007 Hospital services 2017 Inpatient services $111.3 $116.0 $118.6 0.5% 2.2% Total FFS payments (in billions) Payments per FFS beneficiary 3,148 3,026 3,102 –0.4 2.5 Outpatient services 65.5 Total FFS payments (in billions) 60.6 30.9 7.8 8.1 8.4 1,950 7.3 1,799 953 Payments per FFS beneficiary Uncompensated care payments Total (in billions) 6.4 6.0 N/A –6.7 N/A N/A 167 156 N/A –6.4 Payments per FFS beneficiary Inpatient, outpatient, and uncompensated care payments 142.2 Total FFS payments (in billions) 183.0 190.1 2.9 3.8 4.3 Payments per FFS beneficiary 4,101 5,208 2.3 4,992 Note: FFS (fee-for-service), N/A (not applicable). Reported hospital FFS spending includes all hospitals covered by Medicare’s inpatient prospective payment system along with critical access hospitals and Maryland hospitals. Fiscal year 2017 payments include partial imputation to account for the hospitals that had not yet submitted cost reports covering fiscal year 2017. The combined amount for inpatient and outpatient services per capita is based on a weighted average of Part A and Part B services. Percent change columns were calculated before rounding and may not be computable from the payment data in the table, which were rounded. MedPAC analysis of CMS Medicare hospital cost reports and Medicare Provider Analysis and Review files. Source: physician practices. Driven largely by outpatient spending, Background overall Medicare spending on inpatient, outpatient, and uncompensated care increased 4.3 percent per FFS 2 Medicare spending on hospitals beneficiary in 2017. In 2017, the Medicare fee-for-service (FFS) program paid Part of the growth in Medicare spending per beneficiary acute care hospitals almost $119 billion for inpatient care, could be due to the shift in beneficiaries toward Medicare about $66 billion for outpatient care, and $6 billion in Advantage (MA) plans. From 2016 to 2017, MA payments for uncompensated care (Table 3-1). From 2016 enrollment increased 1.3 million while FFS enrollment to 2017, inpatient payments increased by 2.2 percent, or declined slightly. In addition, the shift of beneficiaries $2.6 billion. This growth in inpatient payments resulted toward MA may have also altered the average health needs from an increase in payment rates of 1 percent, a 0.7 of the remaining pool of FFS beneficiaries. However, percent increase in the number of inpatient admissions, after examining changes in discharges and adjusting for 1 and a 0.6 percent increase in inpatient case mix. In the changes in age, we still found a slight increase in inpatient same period, outpatient payments per FFS beneficiary use per FFS beneficiary from 2016 to 2017. grew by 8.1 percent, or approximately $5 billion. The increase in outpatient payments reflects a 20 percent Acute inpatient prospective payment system increase in payments for Part B drugs, growing outpatient Medicare’s inpatient prospective payment system (IPPS) visit volume, and an increase in physician services billed pays acute care hospitals a predetermined amount for most as hospital outpatient services after hospitals acquired March 2019 | Report to the Congress: Medicare Payment Policy 69

100 discharges. The payment rate is the product of a base rate those under less financial pressure to constrain costs. and a relative weight that reflects the expected costliness Therefore, while Medicare continues to adjust payment of cases in a particular clinical category compared with the rates for factors outside of hospitals’ control (such as regional wage rates or patient characteristics), Medicare average of all cases. The labor-related portion of the base does not pay hospitals more for having high costs relative payment rate is adjusted by a hospital geographic wage index to account for differences in hospital input prices to neighboring hospitals with similar patients. In addition, Medicare does not pay more to hospitals with low costs among market areas. Payment rates are updated annually. because low costs are their own reward in a prospective To set inpatient payment rates, CMS uses a clinical payment system. categorization system called Medicare severity– diagnosis related groups (MS–DRGs). The MS–DRG Links between Medicare’s hospital payment rates system classifies each patient case into 1 of 761 groups, and other payers’ payment rates which reflect similar principal diagnoses, procedures, Spending under Medicare’s FFS payment system is used and severity levels. The severity levels are determined to set benchmarks for MA plans and for accountable according to whether patients have a complication or care organizations (ACOs). More importantly, it is comorbidity (CC) associated with the base MS–DRG (the also the foundation of MA plans’ payment rates to categories are no CC, a nonmajor CC, or a major CC). hospitals. In 2018, 33 percent of Medicare beneficiaries A more detailed description of the acute IPPS, including were in MA plans, and most MA plans paid hospitals payment adjustments, can be found at http://medpac.gov/ using rates benchmarked to and almost exactly equal to docs/default-source/payment-basics/medpac_payment_ Medicare FFS rates (Berenson et al. 2015, Maeda and basics_18_hospital_final_v2_sec.pdf?sfvrsn=0. Nelson 2017). In addition, the Department of Veterans Affairs began setting hospital rates equal to Medicare Hospital outpatient prospective payment system FFS rates in 2012 and annually pays for about $2 billion The outpatient prospective payment system (OPPS) of inpatient care at community hospitals (Government pays hospitals a predetermined amount per service. Accountability Office 2013). The rates that uninsured CMS assigns each outpatient service to 1 of about 700 individuals pay are also often benchmarked to Medicare ambulatory payment classification (APC) groups. Each due to limits on rates charged to low-income uninsured APC has a cost-based relative weight, and a conversion individuals that were enacted in the Patient Protection and factor translates these relative weights into dollar payment Affordable Care Act of 2010 (PPACA). The Medicaid amounts. In 2014 and 2015, CMS implemented several program also uses Medicare rates when setting maximum policies that expanded the size of the OPPS payment supplemental “upper payment limit” Medicaid payments bundles so that the OPPS has fewer primary services (also to hospitals (Medicaid and CHIP Payment and Access called separately payable services) and more packaged Commission 2016). Furthermore, Medicare rates can affect items and services. The most substantive of these policies rates charged by commercial insurance. Most recently, was the establishment of comprehensive APCs (C–APCs), Montana’s state employee health plan fixed its hospital which combine all of the OPPS-covered services on the payment rates to 234 percent of Medicare (Appleby 2018). same claim into a single payment, including those that The treasurer of North Carolina has proposed a similar plan would otherwise be separately payable. Since introducing for its state employee health plan starting in 2020 (Tosczak C–APCs in 2015, CMS has increased the number of C– 2018). Given the growth in the use of Medicare FFS prices APCs from 25 to 64. as a benchmark, any update to the Medicare base payment amount will affect many other payers. How Medicare sets payment rates Until 1984, Medicare paid hospitals based on their cost of care. Currently, Medicare pays hospitals rates under Are Medicare payments adequate in a prospective payment system (PPS), meaning rates are 2019? set prospectively and largely do not depend on individual hospitals’ costs. One rationale for ending payments based To judge whether payments in 2019 are adequate for on cost was that cost-based payments reduce the incentive relatively efficient hospitals, we examine several indicators for cost control. A second reason is that, as we will show of payment adequacy. We consider beneficiaries’ access later in this chapter, hospitals with higher costs are often Hospital inpatient and outpatient services: Assessing payment adequacy and updating payments 70

101 FIGURE XXXX... 3-X FIGURE Hospital closures and openings declined from 2013 to 2017 3–1 40 Newly opened hospitals Closed hospitals 30 30 28 24 19 18 20 14 13 Number of hospitals 11 9 10 5 0 2017 2013 2014 2016 2015 Source: MedPAC analysis of the CMS Provider of Services file, internet searches, and personal communication with the Department of Health and Human Services Office of Rural Health Policy. Note: Note and Source are in InDesign. Source: hospital services remains good, in part because excess to care, changes in the quality of care, hospitals’ access inpatient capacity persists in most markets. to capital, and the relationship of Medicare’s payments to hospitals’ costs for both average and relatively efficient Hospital closures decreased slightly in 2017 hospitals. Most of our payment adequacy indicators Notes about this graph: for hospitals are positive, but 2017 Medicare margins While closures are still relatively rare events, there have remained negative for most hospitals and were about –2 • Data is in the datasheet. Make updates in the datasheet. been slightly more hospital closures than hospital openings percent for relatively efficient providers. in recent years. In 2017, we identified 18 closures and 5 • WATCH FOR GLITCHY RESETS WHEN YOU UPDATE DATA!!!! openings (Figure 3-1), a slight decrease from 2016 in both • The column totals were added manually. Beneficiaries’ access to care remained good; measures. Among those that closed in 2017, 10 were in excess inpatient capacity persisted • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. urban counties and 8 were in rural counties. The hospitals To evaluate access to care, we examine the availability of that opened in 2017 were all in urban counties. • I can’t delete the legend, so I’ll just have to crop it out in InDesign. hospital services to Medicare beneficiaries by analyzing From 2015 to 2017, 65 hospitals closed and 29 opened. • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph hospital employment growth, hospital closures, occupancy, The hospitals that closed tended to be smaller (81 beds, on default when you change the data. hospitals’ financial incentive to see Medicare patients, average), with low inpatient occupancy rates (22 percent, and other measures. Our framework also includes an • Use paragraph styles (and object styles) to format. on average), and poor profitability (all-payer margin of evaluation of hospitals’ access to capital, which provides • Data was from: R:\Groups\MGA\data book 2007\data book 2007 chp1 –6 percent, on average) compared with average facilities. an outlook on the industry’s ability to sustain or expand Of these closures, 65 percent were in states that did not its existing resources. Medicare beneficiaries’ access to expand their Medicaid program under PPACA, and 52 March 2019 | Report to the Congress: Medicare Payment Policy 71

102 FIGURE FIGURE Title here... 1-XX Medicare inpatient discharges per beneficiary and outpatient 3–2 visits per beneficiary increased from 2016 to 2017 60 Outpatient visits 50 43.5 per FFS Part B beneficiary Inpatient discharges 40 per FFS Part A beneficiary 30 20 10 0 Cumulative percent change –10 –20 –20.4 –30 2009 2012 2013 2014 2007 2016 2017 2008 2010 2011 2015 Fiscal year Note: FFS (fee-for-service). Data include general and surgical, critical access, and children’s hospitals. Source: MedPAC analysis of CMS’s inpatient and outpatient claims and enrollment data. Note: Note and Source are in InDesign. occupancy rates for hospitals increased slightly from 62.1 percent were in rural counties and were, on average, Source: percent to 62.5 percent. However, a significant degree of 21 miles from the nearest hospital. Nine of the rural inpatient capacity was still underutilized in 2017, which hospitals that closed were critical access hospitals and appeared more significant at rural hospitals. That year, the Notes about this graph: 11 were designated as Medicare-dependent hospitals. average occupancy rate of urban hospitals was 65.9 percent, Urban hospitals that closed were an average of nine miles • Data is in the datasheet. Make updates in the datasheet. while the average occupancy rate of rural hospitals was from the nearest hospital. Some hospitals that closed • I deleted the years from the x-axis and put in my own. 40.2 percent. Over the past decade (2006 to 2017), hospital between 2015 and 2017 either converted to outpatient- occupancy rates declined from 63.8 percent to 62.5 percent; only facilities (e.g., stand-alone emergency departments • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. this change occurred as the volume of Medicare inpatient or imaging centers) or became post-acute care facilities; • The dashed line looked ok here, so I didn’t hand draw it. admissions declined. Given excess inpatient capacity, some others closed completely. The 29 hospitals that opened • I can’t delete the legend, so I’ll just have to crop it out in InDesign. of these hospitals have sought to reduce their inpatient over this 3-year period were often small (51 beds, on capacity and replace it with outpatient capacity (Barclays average), and 88 percent were urban. The newly opened • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph 2018, Goldberg 2018, Japsen 2018). hospitals are a mix of small full-service hospitals and default when you change the data. small specialty or microhospitals. • Use paragraph styles (and object styles) to format. Modest increases in inpatient use Despite closures, rural and urban hospitals have Between 2016 and 2017, inpatient discharges and excess inpatient capacity outpatient visits per beneficiary increased by 0.7 percent. Despite some closu res, existing hospitals often still have These small increases reflect a discontinuation of excess capacity. Between 2016 and 2017, aggregate Hospital inpatient and outpatient services: Assessing payment adequacy and updating payments 72

103 FIGURE FIGURE Title here... The number of short (one- and two-day-stay) Medicare inpatient 3–3 1-XX discharges per beneficiary increased from 2016 to 2017 5 0 –5 –10% –10 –15 –20 –18% Admissions of: –25 1 day –32% 2 days –30 Cumulative percent change 3 days –35 4 or more days –40 –38% –45 2014 2011 2015 2007 2012 2013 2008 2009 2016 2017 2010 Note: Data include general and surgical, critical access, and children’s hospitals. MedPAC analysis of CMS’s inpatient claims and enrollment data. Source: Note: Note and Source are in InDesign. beneficiaries’ admissions occurring in urban hospitals long-term trends where inpatient volume declined and increased from 46 percent to 53 percent. outpatient volume increased rapidly. Despite the leveling Source: of these trends, inpatient use is significantly lower and Increase in inpatient discharges reflects growth in outpatient use is significantly higher than each was a one-day and two-day stays Notes about this graph: decade earlier. From 2007 to 2017, inpatient discharges per beneficiary decreased 20.4 percent, while outpatient The slight increase in the volume of inpatient discharges • Data is in the datasheet. Make updates in the datasheet. visits per beneficiary increased 43.5 percent (Figure 3-2). from 2016 to 2017 reflects a 1.9 percent per beneficiary • I deleted the years from the x-axis and put in my own. increase in medical cases and a 1.5 percent per beneficiary The volume of Medicare inpatient discharges increased • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. decrease in surgical cases (data not shown). Both inpatient at urban hospitals and decreased at rural hospitals. medical and surgical cases have declined substantially • The dashed line looked ok here, so I didn’t hand draw it. From 2016 to 2017, Medicare inpatient discharges per between 2007 and 2017 (−19 percent and −23 percent, beneficiary declined 0.4 percent at rural hospitals and 1.1 • I can’t delete the legend, so I’ll just have to crop it out in InDesign. respectively). percent at small rural hospitals (fewer than 100 beds). By • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph contrast, from 2016 to 2017, inpatient discharges increased One reason for the small increase in discharges in 2017 default when you change the data. 1.1 percent at urban hospitals. Over the past decade, from was an increase in inpatient discharges with short stays. • Use paragraph styles (and object styles) to format. 2007 to 2017, inpatient discharges have declined across Over the decade from 2007 to 2016, short inpatient all geographic areas, but almost twice as fast in rural areas discharges of one to four days generally declined (Figure (–36 percent) as in urban areas (–17 percent) (data not 3-3). However, from 2016 to 2017, the volume of inpatient shown). Moreover, from 2013 to 2017, the share of rural discharges classified as a one-day stay increased 6.6 | March 2019 Report to the Congress: Medicare Payment Policy 73

104 percent, and the volume of two-day discharges increased an increase of 99 percent (14.8 percent per year). This 2.5 percent. This increase in short-stay discharges may rise reflects an increase in outpatient spending on drugs be attributable to changes in CMS’s Recovery Audit in general and a shift in the payment for the drugs from Contractor (RAC) program. The RAC program reduced the physician fee schedule (when administered in a audits of short hospital stays as a part of CMS’s RAC freestanding office) to the OPPS (when administered in the 3 program revisions. hospital outpatient department (HOPD)). The increase in inpatient one-day cases in 2017 is in large The growth in combined program spending and cost part attributable to five medical and surgical MS–DRGs. sharing for drugs has accelerated in recent years (2016 Major joint replacements for lower extremities accounted to 2017), increasing 18.2 percent. In that period, growth for 51 percent of the increase in one-day discharges, in spending on pass-through drugs was especially strong, increasing by more than 37,000 discharges since 2016. increasing from $1.3 billion to $2.3 billion. Even though Other MS–DRGs accounting for a share of the one-day- drug spending has increased under the OPPS, drugs stay increase include heart failure and shock (15 percent), are profitable overall in the outpatient setting because major joint procedures of the upper extremities (8 percent), hospitals’ revenues exceed their costs for drugs, largely chronic obstructive pulmonary disease (7 percent), and driven by the substantial margins for drugs obtained septicemia (5 percent). through the 340B Drug Pricing Program, a federal program that requires drug manufacturers to provide Growth in outpatient hospital services reflects outpatient drugs to certain hospitals at significantly growth in drug costs and incentives to shift reduced prices. patients to higher cost sites of care The growth in spending on Part B drugs reflects both price From 2012 to 2017, Medicare spending for hospital increases in existing drugs and the introduction of new, outpatient services grew at an annual rate of 8.6 percent. expensive cancer drugs. From 2012 to 2017, about 79 Accounting for this strong growth rate was growth in: percent of the increase in spending on separately payable 4 drug administration and the cost of drugs, especially • drugs was for those that treat cancer. During that period, for the treatment of cancer; OPPS spending on cancer drugs increased from $4.1 billion to $8.8 billion. While the increased drug spending emergency department visits and observation care; • resulted in an increased burden on taxpayers, it increased hospitals’ profits on average in 2017 because of discounts • clinic visits, likely fueled by hospital acquisition from the 340B Drug Pricing Program. From 2016 to of physician practices and hospital employment of 2017, off-campus provider-based departments (PBDs) physicians; and had an important impact on the increased OPPS spending complex surgical procedures that often involve • on drugs. Drug spending in off-campus PBDs grew 25.5 prosthetics or medical devices and that migrate from percent and accounted for nearly 29 percent of the growth the inpatient setting. in total drug spending in HOPDs. The mix of services provided in off-campus PBDs is somewhat different from Also, from 2013 to 2014, outpatient spending rose the mix of services provided in on-campus HOPDs (see substantially (from $46.5 billion to $52.5 billion) due, in text box on off-campus outpatient departments, pp. 76–77). part, to CMS’s decision to include most clinical laboratory tests in the OPPS packaged payment rates, whereas these Observation and emergency visits increased through tests had previously been paid and categorized under the 2016 but leveled off in 2017 OPPS spending also clinical laboratory fee schedule. has increased substantially for observation care. From 2012 to 2017, OPPS spending for observation care rose Spending on Part B drugs has driven OPPS spending 263 percent, attributable to higher volume, updates to growth The largest source of OPPS spending growth has OPPS payment rates, and a substantial increase in the been Part B drugs, which include drugs that have pass- ancillary items included in the packaged payment rate for through status (drugs that are new to the market) and observation care in 2016. While the greater packaging of those that are not pass through but are separately payable ancillary items increased spending on observation care, under the OPPS. From 2012 to 2017, OPPS spending for it lowered the spending on the ancillary items that were these drugs increased from $6.0 billion to $12.0 billion, formerly paid for separately. From 2012 through 2017, the Hospital inpatient and outpatient services: Assessing payment adequacy and updating payments 74

105 TABLE Hospital outpatient departments had strong spending growth 3–2 for separately payable drugs, observation care, ED visits, clinic visits, and chemotherapy administration, 2012–2017 Spending (in billions) Percent change 2012–2017 2012 2017 Driver of growth Service or item Drugs $6.0 $12.0 99% High-cost drugs, increased volume, shift from physician offices Observation care 3.1 263 Larger payment bundle 0.9 2.4 4.1 72 Larger payment bundle, ED visits coding to higher levels Clinic visits 1.9 3.4 81 Shift from physician offices Chemotherapy administration 0.4 0.7 84 Shift from physician offices 43.2 52 Total 65.5 ED (emergency department). Spending includes both program outlays and beneficiary coinsurance. “Drugs” refers to Part B drugs that are separately payable Note: under the outpatient prospective payment system, which includes pass-through drugs and drugs that are separately payable but do not have pass-through status. Source: MedPAC analysis of 2012 and 2017 hospital outpatient standard analytic claims files and data from the CMS Office of the Actuary. by 5.3 percent. Increased packaging of ancillary items volume of observation care increased spending by 19.7 into ED visits increased OPPS spending by 25.1 percent percent, while updates to OPPS payment rates increased (and decreased spending on separately payable lab tests). spending by 5.3 percent. Inclusion of certain ancillary Finally, we have found that a shift in the coding of ED items in the packaged payment rate for observation care visits from low-acuity levels (which have lower payment was by far the biggest factor in spending on observation rates) to higher acuity levels (which have higher payment care, increasing spending by 188 percent. Growth in the rates) increased ED spending by 20.3 percent from 2012 volume of separately payable observation care has slowed. to 2017. Similar to observation care, growth in ED visits From 2016 to 2017, volume of observation stays fell 1.2 has slowed. From 2016 to 2017, volume of ED visits was percent, and Medicare spending for these stays rose 1.0 5 unchanged and Medicare spending for them increased by percent. 2.0 percent (data not shown). OPPS spending for emergency department (ED) visits also increased, rising by 72 percent from 2012 to 2017 (Table Shift of services from physician offices to HOPDs has 3-2). Similar to observation care, a number of factors increased OPPS spending Another large source of growth contributed to the increase in spending on ED visits, in spending on hospital outpatient services was a shift including increased volume, updates to the ED payment from (relatively lower cost) physician offices to (relatively rates, increased packaging of ancillary items into the ED higher cost) HOPDs. From 2012 to 2017, spending for and payment rates, and a shift of ED visits coded at lower volume of clinic visits and drug administration (especially acuity levels to higher acuity levels. While the increased for chemotherapy drugs) in the hospital outpatient setting packaging of ancillary items increased spending on ED rose substantially, while the volume of these services fell in visits, it decreased the spending on the ancillary items that freestanding physician offices. Over this period, the volume CMS shifted from separately payable to packaged into the of OPPS clinic visits rose 34 percent and chemotherapy payment for ED visits. From 2012 to 2017, growth in the administration rose 45 percent. At the same time, the volume of ED visits increased spending by 8.4 percent, volume of office visits in freestanding offices fell 0.6 and updates to OPPS payment rates increased spending percent and chemotherapy administration fell 15.2 percent. March 2019 | Report to the Congress: Medicare Payment Policy 75

106 Payments for off-campus outpatient departments nonexcepted services provided in off-campus PBDs significant share of hospital outpatient would be paid at 50 percent of OPPS rates in 2017 and prospective payment system (OPPS) services 40 percent of OPPS rates in 2018 and 2019. is provided in off-campus provider-based A departments (PBDs). In 2017, about 16.6 percent of On average, the services provided in off-campus OPPS volume and 11.1 percent of OPPS revenue was PBDs are less complex than the services provided in for services provided in off-campus PBDs. From 2016 on-campus outpatient settings. In 2017, the average to 2017, volume in off-campus PBDs grew 4.8 percent, relative weight of a service (a measure of resources and spending rose 12.4 percent. needed to furnish a service) provided in an off-campus PBD was 2.18, compared with 5.00 (2.3 times higher) Before 2017, CMS paid for all services provided in off- for the average relative weight of a service provided in campus PBDs at the standard OPPS rates. Payments for an on-campus hospital outpatient department (HOPD). some services provided in off-campus PBDs changed The higher the relative weight, the more complex the in 2017. CMS now pays some of these services at the service. standard OPPS rates but pays for others at rates that are a fraction of the OPPS rates. We have found sharp differences between the services provided in off-campus PBDs and on-campus HOPDs. Whether a service provided in an off-campus PBD In 2017, outpatient clinic visits were by far the most is paid at the standard OPPS rate or at a fraction of frequently provided service in off-campus PBDs, the OPPS rate depends on whether the service was constituting 46 percent of total Medicare service provided in an off-campus PBD that is “excepted” volume and 18 percent of total Medicare revenue in that from Section 603 of the Bipartisan Budget Act (BBA) setting. In contrast, outpatient clinic visits were only of 2015. Section 603 of the BBA of 2015 defines an 14 percent of total Medicare volume and 4 percent of excepted off-campus PBD as one that was billing total Medicare revenue in on-campus HOPDs. Also, under the OPPS before November 2, 2015 (the date separately payable drugs were a much larger share the Congress enacted the BBA of 2015). CMS set of Medicare revenue in off-campus PBDs than in payments for nonexcepted services as a fraction of on-campus HOPDs—40 percent of off-campus PBD the services’ OPPS payment rates, where the average revenue compared with 18.5 percent of on-campus of these payment rates for nonexcepted services HOPD revenue. Finally, in 2017, the 10 ambulatory approximates the average of the payment rates in the payment classifications (APCs) that had the highest Medicare physician fee schedule. CMS determined that (continued next page) Spending on chemotherapy administration grew more Increased spending on clinic visits and chemotherapy slowly than volume in HOPDs from 2016 to 2017 administration in HOPDs reflects the growth in volume because CMS restructured the APCs for chemotherapy, in HOPDs. From 2012 to 2017, spending grew 81 which lowered the OPPS payment rates for some of percent for clinic visits and 84 percent for chemotherapy the chemotherapy techniques that are provided most administration. Most recently, from 2016 to 2017, volume frequently. of clinic visits grew 3.2 percent in HOPDs and Medicare spending rose by 6.0 percent. Volume of chemotherapy The shift of clinic visits and chemotherapy administration administration grew 5.6 percent, and Medicare spending from physician offices to HOPDs is important because rose 3.0 percent. In contrast, volume of office visits and it increases Medicare program spending and beneficiary chemotherapy administration provided in freestanding cost-sharing liability. Medicare payment rates for the same offices dropped 1.4 percent and 5.2 percent, respectively. or similar services are generally higher in HOPDs than in Hospital inpatient and outpatient services: Assessing payment adequacy and updating payments 76

107 Payments for off-campus outpatient departments (cont.) in excepted facilities. Therefore, the lower Medicare Medicare revenue in off-campus PBDs were different payment rates for services provided in nonexcepted from the 10 APCs that had the highest Medicare off-campus PBDs did not have much effect on reducing revenue in on-campus HOPDs (Table 3-3). Medicare spending. However, CMS has decided to Notably, the vast majority of services provided in off- expand the extent to which it pays for services provided campus PBDs in 2017 were in those with “excepted” in off-campus PBDs at the reduced rates currently paid status and thus paid at full OPPS payment rates. in nonexcepted PBDs: CMS will pay all clinic visits About 94 percent of the overall Medicare volume provided in excepted off-campus PBDs at the reduced and Medicare revenue in off-campus PBDs occurred rates starting in 2020. ■ TABLE Services with the highest OPPS revenue in off-campus PBDs and on-campus HOPDs 3–3 Off-campus PBDs On-campus HOPDs Share of Share of OPPS revenue APC APC OPPS revenue Clinic visits 18.0% Observation services 6.0% Level 4 drug administration 2.5 Clinic visits 4.4 Level 3 endovascular procedures 2.2 Level 4 imaging without contrast 3.6 Level 3 radiation therapy 2.2 Level 4 ED visits 3.1 2.1 2.9 Level 3 nuclear medicine Level 5 ED visits Level 2 ICD procedures 2.0 2.5 Level 2 imaging without contrast 1.6 Level 3 drug administration 2.1 Level 3 imaging without contrast Level 1 intraocular procedures 2.0 1.3 Level 4 musculoskeletal procedures 2.0 Level 4 nuclear medicine 1.2 Level 1 endovascular procedures Level 2 skin procedures 1.9 Level 3 electrophysiologic procedures 1.2 Note: OPPS (outpatient prospective payment system), PBD (provider-based department), HOPD (hospital outpatient department), APC (ambulatory payment classification), ED (emergency department), ICD (implantable cardioverter defibrillator). MedPAC analysis of hospital outpatient standard analytic claims files from 2017. Source: CMS has implemented lower OPPS payment rates for freestanding offices. For example, we estimate that the services provided in some hospitals’ off-campus PBDs. Medicare program spent $1.9 billion more in 2017 than CMS intends for the lower OPPS rates to equal rates it would have if payment rates for clinic visits in HOPDs paid in physician offices under the Medicare physician were the same as physician office rates. As a corollary, fee schedule, on average. For 2017 and 2018, the effects beneficiaries’ cost sharing was $480 million more in of this policy were limited and had a small effect on 2017 than it would have been under physician office rates spending under the OPPS (see text box). However, CMS because of the higher rates paid in HOPDs. However, decided to expand this policy substantially for 2019, and Section 603 of the Bipartisan Budget Act (BBA) of the likely effect will be a substantial reduction in OPPS 2015 has begun to have a small effect on the differences spending for clinic visits. in payments between HOPDs and physician offices for clinic visits. Under provisions in the BBA of 2015, March 2019 | Report to the Congress: Medicare Payment Policy 77

108 TABLE Trends in unadjusted and risk-adjusted rates of readmissions across all conditions 3–4 Type of readmission 2012 2013 2014 2015 2016 2017 15.8% Unadjusted unplanned readmissions 16.4% 15.9% 15.6% 15.5% 15.8% 15.0 15.2 15.3 15.7 16.3 Risk-adjusted unplanned readmissions 15.0 Note: The readmissions for 2017 reflect admissions during the first 11 months of fiscal year 2017 and readmissions after those admissions during the full 12 months of fiscal year 2017. Source: MedPAC analysis of Medicare claims files for Medicare fee-for-service beneficiaries ages 65 or older. Shift of some services from the inpatient to the outpatient incentive program (HVIP) that will be simpler and will setting has increased OPPS spending Growth in relatively produce more equitable results compared with existing complex services—such as spinal surgeries; endovascular quality payment programs. procedures; and removal, replacement, or insertion of In 2019, hospitals’ performance on quality metrics has defibrillator systems or pulse generators—suggests that the potential to increase a hospital’s base IPPS payment some of the growth in OPPS spending is from services rates by as much as 3 percent and lower payments by as migrating from the inpatient to the outpatient setting. For much as about 5.5 percent. Three payment adjustments example, from 2012 to 2017, spending on the services are responsible for these potential changes: the Hospital in APC 5464 (level 4 neurostimulator and related Readmission Reduction Program (HRRP) (which can procedures) increased 131 percent and from 2016 to 2017, reduce payments up to 3.0 percent), the HVIP (between by 22.4 percent. about a 3.0 percent increase and a 1.5 percent reduction to payments), and the Hospital-Acquired Condition Hospitals with excess capacity have a financial Reduction Program (which can reduce payments 1.0 incentive to serve Medicare beneficiaries percent for 25 percent of hospitals). (These programs Another measure of access is whether providers have a do not apply to outpatient payments.) In 2018, almost a financial incentive to expand the number of Medicare quarter of hospitals will see a net increase in payments beneficiaries they serve. This measure examines whether (averaging about $98,000) and a little less than three- Medicare payments cover the variable cost of treating quarters will see a net decrease in payments (averaging an additional Medicare patient, meaning the costs that about $456,000) under the combined effect of these vary with volume. On average, the marginal profit across programs. On net, these three programs lower Medicare hospital service lines was approximately 8 percent in 6 payments by about $970 million, equivalent to about 0.8 2017. Because hospitals would be expected to generate percent of Medicare’s inpatient payments or 0.5 percent of about 8 percent profit on a marginal increase in Medicare Medicare’s total hospital payments. volume, hospitals with excess capacity have a financial incentive to serve more Medicare beneficiaries. Key measures of quality demonstrate improvement Quality of care improved To assess aggregate trends in quality of care across all The quality of hospital care improved in recent years, IPPS hospitals, we use mortality rates, readmission rates, and at least part of this improvement appears to be due to and patient experience measures. From 2012 to 2017, various financial incentives included in recent years in the mortality rates, readmission rates, and patient experience Medicare program. Although these incentive programs measures (e.g., communication with nurses and doctors, could be improved, the data suggest that even imperfect quietness of hospital environment) have improved. The incentives can lead to improved quality. In Chapter 15 of share of patients rating their overall hospital experience a 9 this report, we discuss a redesign of Medicare’s hospital or 10 on a 10-point scale has increased from 71 percent to quality payment programs into a single hospital value 73 percent. Hospital inpatient and outpatient services: Assessing payment adequacy and updating payments 78

109 TABLE Risk-adjusted 30-day postdischarge mortality rates have declined 3–5 Mortality rate 2012 2013 2014 2015 2016 2017 8.4% Unadjusted mortality 8.1% 8.4% 8.4% 8.6% 8.4% 9.9 8.5 Expected mortality 9.1 9.4 10.2 10.7 Risk-adjusted mortality 7.0 7.7 7.5 6.7 7.2 6.4 MedPAC analysis of 2012 through 2017 Medicare claims using 3M all-patient refined–diagnosis related group risk of mortality V32 grouper and beneficiary age Source: and gender to calculate risk-adjusted mortality rates (using 2010 through 2012 data to set expected rates). points, including a decline of 0.3 percentage points in Readmission rates improved The Congress enacted the 2017 (Table 3-5). Over the five-year period, raw mortality HRRP in 2010, and since that time, readmission rates have fallen. In our recent analysis of the HRRP, we found rates were relatively constant, but expected mortality that the program gave hospitals an incentive to reduce increased, which suggests that beneficiaries admitted in inappropriate readmissions (Medicare Payment Advisory recent years tended to have more comorbidities and thus a Commission 2018). Our updated analysis of readmission higher risk of mortality. Other studies have found similar improvements for condition-specific mortality (Hines rates across all conditions for beneficiaries over age 65 found that between 2012 and 2017, the raw unplanned 2015, Krumholz 2015). The combination of a decline readmission rate declined by 0.6 percentage point, from in readmissions and a decline in hospital mortality is 16.4 percent to 15.8 percent (Table 3-4). Once risk evidence of steadily improving quality. adjusted, these rates declined from 16.3 percent to 15.0 Hospitals’ access to capital remained strong percent. An in-depth discussion of changes in readmission in 2017 rates is available in our June 2018 report to the Congress. In 2017, hospitals’ access to capital remained strong. In 2013, the Commission proposed a budget-neutral Nonprofit hospitals issued $35 billion in bonds, roughly package of improvements to the HRRP. The first proposal equivalent to the $38 billion of bond offerings issued in was to set a fixed target for readmission rates so aggregate 2016 (Figure 3-4, p. 80) (Thomson Reuters 2018). Both penalties would drop when industry performance years reflect higher bond issuance levels than any year improved. Second, we discussed changing the penalty since 2009. The 2017 bond issuances consisted of $23 formula to make the penalty per excess readmission close billion in new financing and $12 billion in refinancing of to the cost of each excess readmission. Third, to create existing debt. Between November 2017 and November greater precision in measuring relative performance 2018, the average interest rate for double-A tax-exempt 30- and offset the cost of changing the penalty formula, we year nonprofit hospital bonds increased from 3.2 percent to 7 discussed expanding the policy to cover all conditions. 3.92 percent (Cain Brothers 2018). Hospital construction Fourth, we proposed evaluating hospitals’ readmission spending was $24 billion, about the same level as 2016 and rates against rates for peer hospitals with similar shares of roughly equivalent to the level of bond issuances for new low-income patients as a way to adjust penalties for the financing (Census Bureau 2018). Construction spending effects of socioeconomic status on hospitals’ readmission for hospitals in 2017 was the lowest in over a decade rates (Medicare Payment Advisory Commission 2013), because the industry is focused on building less expensive which the Congress adopted in the 21st Century Cures outpatient capacity rather than inpatient capacity (Conn Act (Public Law 114–255). Aspects of these proposals are 2017). Several financial ratings agencies consistently incorporated in the HVIP design. observed increases in hospitals’ capital expenditures from 2016 to 2018 but note the current focus of hospitals on Mortality rates improving From 2013 to 2017, risk- building outpatient capacity, such as outpatient surgical adjusted mortality rates declined by 1.1 percentage March 2019 | Report to the Congress: Medicare Payment Policy 79

110 FIGURE FIGURE More hospitals opened... 3-X Nonprofit hospital bond offerings for new financing 3–4 roughly equal to hospital construction spending in 2017 60 Construction spending Bond offerings: Refinancing Bond offerings: New financing 50 21 40 12 8 15 12 30 9 9 Dollars (in billions) 40 41 5 39 20 9 34 32 30 4 33 33 32 30 27 6 25 24 26 23 23 20 18 10 18 14 15 10 0 2009 2010 2011 2012 2013 2014 2016 2015 2017 2007 2008 Source: Nonprofit hospitals’ bond offering data from Thomson Reuters and hospital construction spending data from the U.S. Census Bureau. Note: Note and Source are in InDesign. consisting largely of rural hospitals, by Apollo Global facilities and other outpatient access points (Barclays 2018, Source: Moody’s Investors Service 2018). management for $5.6 billion (Reed 2018). This acquisition suggests that some rural hospitals remain an attractive Mergers and acquisitions investment, despite years of declining rural inpatient volume. Hospitals and hospital systems continued to expand through acquisition. In 2017, 216 individual hospitals were Notes about this graph: Hospital employment increased acquired in 78 transactions, a decline from 2015 and 2016, • Data is in the datasheet. Make updates in the datasheet. when 267 and 241 hospitals, respectively, were acquired Between October 2015 and October 2018, the number of • WATCH FOR GLITCHY RESETS WHEN YOU UPDATE DATA!!!! (Irving Levin Associates Inc. 2018). In 2017, hospital individuals employed by hospitals grew from 4.9 million acquisitions tended to be slightly larger hospitals than to 5.2 million, an increase of 5.6 percent, slower than in • The column totals were added manually. in previous years, and a larger share of the transactions the rest of the health care sector (6.8 percent), but faster • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. involved single facilities (71 percent) rather than systems than the rest of the economy (4.8 percent) (Bureau of • I can’t delete the legend, so I’ll just have to crop it out in InDesign. of hospitals. In addition, the 2017 acquisitions tended Labor Statistics 2018b). Over 10 years (2008 to 2018), to occur across regions rather than in the same market. hospital employment increased 12.0 percent while • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph These acquisitions have resulted in greater market power employment in the rest of the economy increased 8.7 default when you change the data. for hospitals, in both the individual market and regional percent. • Use paragraph styles (and object styles) to format. context, in negotiating contracts with insurers, physicians, Hospitals have hired individuals in certain high-skill and drug and device manufacturers. Not included in • Data was from: R:\Groups\MGA\data book 2007\data book 2007 chp1 occupational categories and reduced the number of staff in the information above is the more recent acquisition certain lower-skilled occupations. From 2015 to 2017, the of LifePoint Health Inc., a for-profit hospital system, Hospital inpatient and outpatient services: Assessing payment adequacy and updating payments 80

111 FIGURE Title here... FIGURE 1-XX Hospitals’ all-payer financial performance has remained stable since 2010 3–5 14 12 10.9 10.8 10.6 10.4 10.4 10.4 10.3 10.2 10 8 7.1 7.2 7.1 6.8 6.4 6.4 6.3 6.0 6 6.4 6.0 6.0 5.9 5.8 5.6 Margin (in percent) 5.1 5.2 4 EBITDA margin Total all-payer margin 2 Operating margin 0 2013 2017 2012 2011 2010 2016 2015 2014 Note: EBITDA (earnings before interest, taxes, depreciation, and amortization). A margin is calculated as revenues minus costs, divided by payments. Analysis excludes critical access hospitals. MedPAC analysis of Medicare hospital cost report data. Source: Note: Note and Source are in InDesign. number of physicians employed by hospitals increased 5.3 insured patients. In 2017, total margins (which include Source: percent but varied by type of physician (Bureau of Labor investment income) were 7.1 percent, near an all-time Statistics 2018a). Overall, the number of registered nurses high (Figure 3-5). Other measures of all-payer profitability employed by hospitals rose 6.2 percent during this period, are also strong. Cash flow—as measured by earnings Notes about this graph: increasing by roughly 100,000 individuals. Hospitals also before interest, taxes, depreciation, and amortization • Data is in the datasheet. Make updates in the datasheet. increased the number of physician assistants hired by (EBITDA)—has remained steady and strong for the past • I deleted the years from the x-axis and put in my own. nearly 20 percent and pharmacists by 9 percent. eight years, between 10 percent and 11 percent. Financial ratings agencies consistently reported in 2018 that for- • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. Total (all-payer) profitability remains strong profit and nonprofit financial balance sheets (which • The dashed line looked ok here, so I didn’t hand draw it. include measures such as EBITDA, days cash on hand, Hospitals’ access to capital for expansions and • I can’t delete the legend, so I’ll just have to crop it out in InDesign. and debt load) were at historically high levels for the acquisitions is largely dependent on their total (all-payer) industry (Barclays 2018, Fitch Ratings 2018, Moody’s profitability. All-payer margins remain strong because the • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph Investors Service 2018, S&P Global Ratings 2018). growth of private payer rates continues to rise faster than default when you change the data. costs (Health Care Cost Institute 2018). While Medicare In 2017, total margins varied across hospital types. For the • Use paragraph styles (and object styles) to format. represents about one-third of all-payer revenues and 44 10th year in a row, for-profit hospitals had a higher total percent of all admissions, commercially insured patients (all-payer) margin compared with nonprofit hospitals, represent more than 40 percent of patient revenues and totaling 10.8 percent, almost 5 percentage points higher generate almost all of the operating profits for a typical than in 2007. In addition, the frontier IPPS hospitals (those 8 hospital. Operating margins (which exclude investment in low population-density counties) had an average total income) peaked in 2015 at 6.4 percent after a growth in March 2019 | Report to the Congress: Medicare Payment Policy 81

112 margin of 10.1 percent, 3 percentage points higher than in Between 2016 and 2017, three key changes to inpatient other IPPS hospitals, which suggests that isolated hospitals payments occurred: can do well financially in frontier areas when they have • a 1.0 percent increase in base payment rates sufficient volumes of insured individuals. While overall (consisting of a 1.65 percent update, adjustments for profitability was relatively high, margins on Medicare documentation and coding, and other changes); patients remained negative. • a 0.6 percent increase in inpatient case mix; and Medicare payments and providers’ costs a $0.4 billion reduction in disproportionate share • In assessing payment adequacy, the Commission also (DSH) hospital and uncompensated care payments. considers the relationship between Medicare payments and the costs of providing care to Medicare patients. Medicare continues to see growth in the use of outpatient We assess the adequacy of Medicare payments for the services. Growth resulted from a combination of factors: hospital as a whole (across all Medicare services), thus a rise in the number of beneficiaries, a rise in outpatient measuring the relationship between payments and costs visits per beneficiary, and 19 percent growth in payments using an overall Medicare margin. This margin includes for separately payable Part B drugs administered in all Medicare payments and all Medicare-allowable costs hospitals’ outpatient departments. for the six largest hospital departments covered by the inpatient, outpatient, and post-acute PPS systems as well Growth in Part B drug spending improved hospital as uncompensated care payments and graduate medical profitability The 19 percent increase in Part B drug 9 education payments and costs. spending was a result of new drugs coming on the market, increases in volumes of Part B drugs used, a shift in We report the overall Medicare margin across service lines the site of administration toward hospitals or hospital- because no hospital service line is a purely independent owned practices, and increases in Part B drug prices. business. For example, we find that operating any in- Because hospitals and the Medicare program do not hospital post-acute care (PAC) provider improves the set pharmaceutical prices, manufacturer price increases profitability of acute inpatient care services because such for Part B drugs can also drive up Medicare program a provider allows a hospital to safely discharge patients payments. sooner from their acute care beds, thus reducing the cost of the inpatient stay. The overall Medicare margin also takes However, as the volume and price of Part B drugs into account revenues that are not included in the service- increased from 2016 to 2017, hospital profits on these line payments for inpatient and outpatient care. These drugs also increased. In 2017, Medicare paid hospitals revenues include Medicare payments for uncompensated 106 percent of pharmaceutical companies’ average sales 10 care beginning in fiscal year 2014. Excluding these prices for most Part B drugs. Over 50 percent of hospitals’ Medicare revenues would understate Medicare payments Part B drug administration takes place at hospitals under to hospitals. Another benefit of focusing on overall the 340B Drug Pricing Program, which mandates that Medicare margins is that we can avoid the challenges of pharmaceutical companies provide substantial discounts precisely allocating overhead and administrative costs to certain hospitals. These discounts resulted in 340B among the different service lines. hospitals often having drug acquisition costs that were 30 percent or more below the average sales price (and Medicare payment growth thus below the 2017 payments from the Medicare program) (Government Accountability Office 2015). This Changes in Medicare inpatient hospital payments per difference between the Medicare price paid for drugs and discharge under the IPPS depend primarily on three the hospitals’ acquisition cost of drugs allowed many factors: (1) annual updates to base payment rates; (2) hospitals to generate substantial profits on Part B drugs, changes in reported patient case mix (i.e., a measure of which contributed to hospitals’ profit margin on outpatient relative patient complexity); and (3) policy changes that services increasing between 2016 and 2017 from –15.3 are not implemented in a budget-neutral manner. In 2017, percent to –14.2 percent. The increasing profit on Part B the average Medicare inpatient payment per case increased drugs offset part of hospitals’ losses on other outpatient 2.0 percent and uncompensated care payments declined services. Starting in 2018, CMS reduced payments to because of an increase in the number of insured patients. 340B hospitals for many Part B drugs (other than new Hospital inpatient and outpatient services: Assessing payment adequacy and updating payments 82

113 TABLE Cost growth, case-mix change, and hospital input price inflation, 2013–2017 3–6 Annual cost growth Average annual cost growth 2014 Cost measure 2015 2016 2017 2013 2013–2017 2.6 Inpatient costs per discharge % 2.6 % 1.8 % 4.0 % 2.3 % 2.3 % 1.8 0.6 3.4 Inpatient case-mix-index change 2.0 2.0 0.8 2.4 1.8 1.6 1.9 1.9 Input price inflation 1.7 Cost-growth numbers are not adjusted for reported changes in case mix. Analysis excludes critical access hospitals and Maryland hospitals. “Input price inflation” Note: reflects a four-quarter moving and weighted average of changes in the hospital operating and capital market basket indexes calculated for the second quarter of each year. MedPAC analysis of Medicare cost reports, claims files, and hospital input price inflation estimates from CMS. Source: pass-through drugs) to 22.5 percent below the average rest of the labor market (4.8 percent) (Bureau of Labor sales price to more closely align Medicare payments with Statistics 2017). how much these hospitals pay to acquire drugs. At the From 2016 to 2017, the reported resource needs across same time, CMS enacted an offsetting increase in payment all inpatient cases (or case-mix index (CMI)) increased rates for other services. The net result is that while we 0.6 percent. This increase in overall CMI was the result expect to find that hospital profits on Part B drugs declined of increases in CMI for both medical cases and surgical from 2017 to 2018, profit margins on other services likely cases. However, medical cases, which have a lower increased, resulting in no material effect on hospitals’ average case mix than surgical cases, increased as a share overall outpatient margins. of all cases, and this increase moderated the case-mix growth. Rate of cost growth remains close to rate of input price inflation The modest 1.8 percent increase in costs per inpatient Hospitals’ per case cost increases were relatively low from discharge reflects a modest growth in routine costs (e.g., 2013 to 2015. Then in 2016, costs per discharge increased nursing labor) and ancillary services. Ancillary services by about 4.0 percent, in large part reflecting an unusual made up about half of inpatient cost growth, with the one-year shift in services toward inpatient surgeries, which largest share of growth from implantable devices, which have a higher case mix (Table 3-6). However, in 2017, reflects 10 percent of total hospital costs and grew by 5 the per case cost increased by 1.8 percent, lower than at percent from 2016 to 2017 (Table 3-7, p. 84). The higher any other point in the last two decades, reflecting low cost of implantable devices reflects, in part, the increase in underlying cost growth and more case mix changes. joint replacement surgeries. The lower underlying cost growth in 2017 is a result of In contrast to the 2014 to 2016 time frame, when drug several factors, including shorter lengths of stay and lower costs per discharge rose by an average of 6 percent per input price inflation for hospitals, reflecting low economy- year, drug costs per discharge did not materially increase wide inflation and slow wage growth. Hospitals benefited from 2016 to 2017. We did not include a separate estimate from this slow wage growth, with compensation costs of drug costs per discharge in Table 3-7 because such for hospital workers growing by less than 2 percent per estimates from year to year are imprecise due to two year from 2013 through 2015 (Bureau of Labor Statistics unique factors in pharmacy cost accounting. First, 340B 2016). From 2016 through 2017, compensation costs for discounts apply to outpatient drugs but not inpatient drugs, hospital workers grew 4.4 percent, slower than that of the which can result in biasing downward the cost of inpatient March 2019 | Report to the Congress: Medicare Payment Policy 83

114 TABLE Change in cost per inpatient discharge from 2016 to 2017 3–7 2016 and 2017 Share of total inpatient costs per discharge Medicare costs and change Percent change Cost category 2017 2016–2017 in cost (in dollars) $13,377 2016 inpatient cost per discharge Categories comprising growth in inpatient costs per discharge from 2016 to 2017 Routine (e.g., room, nursing) $87 2% 33% Special care (e.g., intensive care) 21 1 11 Ancillary 119 2 56 14 Operating room 1 8 Cardiac catheterization 7 1 4 11 6 Medical supplies 1 60 5 Implantable devices 10 Dialysis 6 6 1 Emergency 13 4 3 Observation 6 1 5 All other 4 0 27 2017 inpatient cost per discharge $13,605 Note: Analysis excludes critical access hospitals and Maryland hospitals. Data are based on a cohort of hospitals included in the margin analysis from 2015 through 2017. Components may not sum to totals due to rounding. Source: MedPAC analysis of Medicare cost reports. spending and cost sharing increased 18.2 percent. Growth drugs by reducing the cost-to-charge ratio for all drugs in in outpatient spending was for cancer drugs administered the hospitals’ cost centers for pharmacy. Second, markups on an outpatient basis. differ among drugs. Although the markup percentage is smaller on high-cost drugs, the expansion of new high-cost Trend in the overall Medicare margin Part B drugs could cause an increase in the cost-to-charge ratio for the pharmacy cost center and cause an upward We define Medicare margins as Medicare payments minus bias in cost estimates for inpatient drugs. It is not clear the allowable costs of treating Medicare patients divided the degree to which the two potential biases offset each by Medicare payments. In analyzing hospital margins, we other. Given these limitations, we also examined changes compute an overall (aggregate) margin with and without in raw charges per inpatient discharge. From 2016 to critical access hospitals (CAHs), which are 1,300 rural 2017, charges for inpatient drugs per discharge increased hospitals whose payments are based on their incurred by less than 2 percent. Coupled with the slight decline in costs. We also exclude hospitals in Maryland, which are hospitals’ pharmacy cost-to-charge ratio, pharmacy costs excluded from the IPPS and paid under a statewide all- per discharge may have risen less than 2 percent or even payer prospective payment system. From 2009 to 2014, declined. The lack of cost growth in the inpatient setting is the overall Medicare margin held relatively steady, varying 11 in stark contrast to the outpatient sector, where charges for from –4.9 to –5.7 percent (Figure 3-6). From 2014 to drugs increased over 20 percent and combined program 2016, the Medicare margin dropped from –5.6 percent Hospital inpatient and outpatient services: Assessing payment adequacy and updating payments 84

115 FIGURE Title here... FIGURE 1-XX Overall Medicare margin dropped slightly from 2016 to 2017 3–6 2 0 –2 –4 –4.9 –5.0 –5.3 –5.4 –5.6 –5.7 –6 –7.2 –7.6 –8 –9.7 –9.9 Overall Medicare margin (in percent) –10 –12 2009 2010 2011 2012 2013 2014 2015 2016 2017 2008 Note: A margin is calculated as payments minus costs, divided by payments; margins are based on Medicare-allowable costs. Analysis excludes critical access hospitals and Maryland hospitals. Medicare inpatient margins include services covered by the acute inpatient prospective payment systems. “Overall Medicare margin” covers acute inpatient, outpatient, hospital-based skilled nursing facility (including swing beds), hospital-based home health, and inpatient psychiatric and rehabilitation services, plus graduate medical education and electronic health record incentive payments and payments for uncompensated care. MedPAC analysis of Medicare cost reports from CMS. Source: Note: Note and Source are in InDesign. Source: to –9.7 percent. This decline was not unexpected given for urban hospitals (Table 3-8, p. 86). Major teaching several payment adjustments required by statute, including hospitals (i.e., hospitals with a high resident-to-bed ratio) Notes about this graph: reductions to the annual payment update, adjustments for had a Medicare margin of –9.0 percent. Major teaching • Data is in the datasheet. Make updates in the datasheet. documentation and coding improvement, decreases in hospitals had higher Medicare margins than the average incentive payments for the adoption of electronic health IPPS hospital in large part because of the extra payments • I deleted the years from the x-axis and put in my own. records, and decreases in uncompensated care payments they receive through the indirect medical education and • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. that correspond with increases in the insured population. DSH adjustments and uncompensated care payments. • The dashed line looked ok here, so I didn’t hand draw it. From 2016 to 2017, the overall Medicare margin again In 2017, for-profit hospitals had the highest Medicare dropped, albeit at a lower rate than in prior years, from • I can’t delete the legend, so I’ll just have to crop it out in InDesign. margins (–2.6 percent), well above the –11.0 percent –9.7 percent to –9.9 percent. The tempered reduction in • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph Medicare margin for nonprofit hospitals (Table 3-8, p. margin was primarily due to historically low cost growth 86). Much of this differential reflects lower outpatient default when you change the data. from 2016 to 2017, coupled with increased revenue from costs at for-profit hospitals. In 2017, hospitals that treated Part B drugs. • Use paragraph styles (and object styles) to format. the highest shares of low-income patients (high-DSH hospitals) had a –8.1 percent Medicare margin. In contrast, Medicare margins by hospital type, 2017 hospitals treating the lowest share of low-income patients In 2017, rural IPPS hospitals (excluding CAHs) had a (non-DSH hospitals) had the lowest Medicare margins –8.2 percent overall Medicare margin, which was 1.8 (–16.4 percent). The difference in margins was attributable percentage points higher than the −10.0 percent margin March 2019 | Report to the Congress: Medicare Payment Policy 85

116 TABLE Overall Medicare margins by hospital type 3–8 2015 2014 2013 2017 2011 2010 Hospital group 2012 2016 –9.7% All hospitals (excluding CAHs) –4.9% –5.7% –5.4% –5.0% –5.6% –9.9% –7.6% –10.0 –5.1 –6.0 –5.9 –5.9 –5.8 –7.9 –9.9 Urban Rural Excluding CAHs –2.6 –2.6 –1.0 2.7 –3.5 –4.9 –7.5 –8.2 –1.9 2.7 0.3 –1.4 –1.7 Including CAHs –5.4 –5.9 –3.2 –6.5 –11.0 –11.1 –9.1 –7.1 –7.0 –7.1 –6.2 Nonprofit 0.8 For profit –2.6 –2.1 –1.3 1.2 1.0 –0.3 –0.1 Major teaching –8.5 –6.3 –3.7 –9.0 –3.5 –2.8 –2.1 –0.8 –4.6 –5.4 –5.0 –4.8 –8.2 –5.0 –6.3 –8.6 Other teaching Nonteaching –12.2 –11.7 –9.9 –7.7 –6.5 –7.9 –8.6 –8.1 –1.5 0.2 High DSH –1.3 –1.5 –2.3 –4.6 –7.2 –8.1 –8.1 Moderate-to-low DSH –9.9 –6.6 –10.0 –6.4 –6.0 –6.7 –7.0 No DSH –15.7 –15.3 –13.3 –16.4 –12.5 –13.4 –13.5 –12.9 CAH (critical access hospital), DSH (disproportionate share). Data are for all hospitals covered by the Medicare acute inpatient prospective payment system in Note: 2017 and for CAHs where indicated. A margin is calculated as payments minus costs, divided by payments; margins are based on Medicare-allowable costs. “Overall Medicare margins” covers acute inpatient, outpatient, hospital-based skilled nursing facility (including swing beds), hospital-based home health, and inpatient psychiatric and rehabilitation services, plus uncompensated care, graduate medical education, and electronic health record incentive payments. The rural margins are shown with and without 1,300 CAHs, which are paid 101 percent of costs for inpatient and outpatient services. The margins without CAHs illustrate the profitability of rural inpatient prospective payment system hospitals; the rural margins with CAHs give a fuller picture of rural hospital profitability. “High DSH” incudes hospitals with the highest disproportionate share adjustments (top quartile). “Moderate-to-low DSH” includes hospitals with disproportionate share adjustments that exceed zero but are not included in the top quartile. Source: MedPAC analysis of Medicare cost reports, Medicare Provider Analysis and Review files, and impact files from CMS. in part to the DSH adjustments and uncompensated care from private payers: high, medium, and low, based on their payments received by hospitals. In addition, hospitals with median non-Medicare profit margins and other factors high shares of Medicare and Medicaid patients tend to from 2012 to 2016. For these years, the hospitals under have more pressure to control costs and therefore tend to high pressure historically had non-Medicare profit margins have lower costs per discharge. of less than 1 percent, while the low-pressure hospitals had non-Medicare profit margins of more than 5 percent. We Fiscal pressure constrains costs found that hospitals under high pressure during the five- year period ended up with lower standardized Medicare Hospitals under financial pressure tend to have lower costs per discharge in 2017 than hospitals under low levels costs. To illustrate this tendancy, we compare hospitals of financial pressure. For more details on our analytic under low and high financial pressure in the analysis methods, see our earlier analysis of payment adequacy below. In addition to financial pressure affecting the level (Medicare Payment Advisory Commission 2011). of costs, the literature shows that changes in Medicare rates can affect the rate of cost growth. Hospitals that The following are key findings from our analysis of receive larger increases in Medicare payment rates financial pressure on hospitals: tend to have larger increases in costs. To determine the association between financial pressure and costs, we • The 25 percent of High pressure equals low cost. grouped hospitals into three levels of financial pressure hospitals under the most financial pressure had median standardized Medicare costs per case that were 6 Hospital inpatient and outpatient services: Assessing payment adequacy and updating payments 86

117 percent lower than the national median for the 2,798 the 508 program “treated more patients, increased payroll, IPPS hospitals with available data. Because of their hired nurses, added new technology, raised CEO pay, and lower Medicare costs, hospitals under pressure had ultimately increased their spending by over $100 million only slight losses on Medicare (–2 percent margin). annually” (Cooper et al. 2017). The implication of these These hospitals tended to have slightly higher shares studies is that constraining Medicare prices should help of patients paying at government rates (50 percent constrain hospital costs. of inpatient days were attributed to Medicare and Relatively efficient hospitals Medicaid FFS patients). The Commission follows two principles when identifying • The 62 percent of Low pressure equals high cost. a set of efficient providers. First, the providers must hospitals under a low level of financial pressure had do relatively well on cost and quality metrics. Second, median standardized Medicare costs per case that the performance has to be consistent, meaning that the were 3 percent above the national median. Because of provider cannot have poor performance on any metric over higher costs, they generated a median Medicare profit the past three years. In the hospital sector, the variables we margin of –11 percent, about 2 percentage points use to identify relatively efficient hospitals are hospital- below the national median. These hospitals tended ® level mortality rates (3M risk-adjusted all-condition to have a slightly smaller share of patients paying at ® mortality), readmission rates (3M potentially preventable government rates (46 percent of inpatient days were readmissions), and standardized inpatient Medicare costs attributed to Medicare and Medicaid FFS patients). per case. Our assessment of efficiency is not in absolute terms, but rather, relative to other IPPS hospitals. In addition to cost differences at the hospital level, cost differences appear at the state level. The literature We assigned Categorizing hospitals as relatively efficient generally finds that a dominant insurer in a state can hospitals to the relatively efficient group or the control reduce the relative market power of hospitals and the group according to each hospital’s performance relative prices commercial insurers pay hospitals (Trish and to the national median on a set of risk-adjusted cost and Herring 2015). We find that lower commercial prices 12 quality metrics for the period 2014 to 2016. We then can result in lower costs. For example, in North Dakota examined the performance of the two hospital groups in and Alabama, where there is one dominant insurer and fiscal year 2017. relatively low commercial payment rates, hospital wage rates are relatively low. (By relatively low, we mean Hospitals were identified as relatively efficient if they met that the ratio of hospital wages to wages paid by other four criteria in each year from 2014 to 2016: employers for comparable employees is lower in Alabama Risk-adjusted mortality rates were among the best • and North Dakota than in the average state) (Medicare two-thirds of all hospitals. Payment Advisory Commission 2007). • Risk-adjusted readmission rates were among the best Another way to examine the relationship between financial two-thirds of all hospitals. pressure and costs is to see how changes in financial pressure affect changes in costs. For example, White and Standardized costs per discharge were among the best • Wu found that hospitals that received higher Medicare two-thirds of all hospitals. payment increases resulting from policy changes tended to have higher cost growth (White and Wu 2014). Contrary Risk-adjusted mortality or standardized costs per • to “cost-shift” theory, they also found that lower Medicare discharge were among the best one-third of all price growth did not cause hospitals to increase prices hospitals. negotiated with commercial insurers. Instead, they found The objective was to identify hospitals that consistently lower Medicare prices led to lower cost growth (White performed at an above-average level on at least one 2013). Similar findings have been reported by others measure (cost or quality) and that always performed (Clemens and Gottlieb 2017, Frakt 2015). A recent study reasonably well on all measures. The rationale for this examined how hospitals responded when they received a methodology and the details of computing the various large increase in their wage index through Section 508 of measures are discussed in our March 2011 report the Medicare Modernization Act. The study found that the (Medicare Payment Advisory Commission 2011). As a hospitals that received higher Medicare payments through March 2019 | Report to the Congress: Medicare Payment Policy 87

118 TABLE 3–9 Performance of relatively efficient hospitals Type of hospital Other Relatively efficient, 2014–2016 Relative performance measure hospitals Number of hospitals 291 1,860 86% Share of hospitals 14% Historical performance, 2014–2016 (percent of national median) Risk-adjusted: TM 89% ) 101% Composite 30-day mortality (3M Readmission rates (3M) 93 102 Standardized Medicare costs per discharge 89 101 Performance metrics, 2017 (percent of national median) Risk-adjusted: 93% 102% Composite 30-day mortality (3M) 103 Composite 30-day readmission (3M) 95 101 Standardized Medicare costs per discharge 91 Median: –2% –9% Overall Medicare margin, 2017 11 9 Non-Medicare margin, 2017 8 Total (all-payer) margin, 2017 5 Note: Relative measures are the median for the group as a share of the median of all hospitals. Per case costs are standardized for area wage rates, case-mix severity, prevalence of outlier and transfer cases, interest expense, low-income shares, and teaching intensity. Composite mortality was computed using the 3M methodology to compute risk-adjusted mortality for all conditions. We removed hospitals with low Medicaid patient loads (the bottom 10 percent of hospitals) and hospitals in markets with high service use (top 10 percent of hospitals) because of concerns that socioeconomic conditions and aggressive treatment patterns can influence unit costs and risk-adjusted quality metrics. MedPAC analysis of 2014 to 2017 Medicare cost report and claims-based quality data. Source: secondary check on hospital quality, we also require that at for the efficient group was 7 percent below the national least 60 percent of the hospital’s patients rated the hospital median. The standardized Medicare cost per discharge for 13 a 9 or 10 on a 10-point scale. the efficient group was 11 percent lower than the national median. These relatively efficient hospitals were spread Examining performance of relatively efficient and other across the country and had a diverse set of characteristics, hospitals from 2014 to 2016 Of the 2,151 hospitals that but they were more likely to be larger nonprofit hospitals met our screening criteria during the 2014 to 2016 period, because those hospitals tend to have better performance 14 291 (14 percent) were found to be relatively efficient. on the quality metrics we analyzed. For a more complete We examined the performance of relatively efficient description of the methodology and other characteristics hospitals on three measures by reporting the group’s of relatively efficient providers, see online Appendix 3-B median performance divided by the median for the set of from our 2016 report to the Congress, available at http:// hospitals in our analysis (Table 3-9). The median efficient www.medpac.gov. hospital’s relative risk-adjusted 30-day mortality rate for the 3-year assessment period was 89 percent of the Historically strong performers had lower mortality and national median, meaning that the 30-day mortality rate costs in 2017 Lower costs allowed the relatively efficient for the efficient group was 11 percent below (that is, better hospitals to generate better Medicare margins. In 2017, than) the national median. The median readmission rate the median hospital in the efficient group had a Medicare Hospital inpatient and outpatient services: Assessing payment adequacy and updating payments 88

119 margin of –2 percent while the median hospital in the (HVIP) that aligns with the Commission’s principles for comparison group had a Medicare margin of −9 percent quality measurement and would replace existing quality (Table 3-9). The relatively efficient group also continued incentive programs. The following recommendation would increase hospital payments by increasing the base payment to perform better on quality metrics, with risk-adjusted rate and by increasing the average rewards hospitals mortality equal to 93 percent of the national median and risk-adjusted readmissions equal to 95 percent of the receive under the proposed Medicare HVIP. national median (Table 3-9). RECOMMENDATION 3 How would current-law changes for 2018, The Congress should: 2019, and 2020 affect hospitals’ Medicare payments and beneficiaries’ access? Replace Medicare’s current hospital quality programs • with a new hospital value incentive program (HVIP) We project Medicare margins for 2019 based on margins that: in 2017 and policy changes that take place in 2018 and includes a small set of population-based outcome, • 2019. The 2018 update for inpatient and outpatient patient experience, and value measures; payments was 1.35 percent. In 2019, the update is also 1.35 percent for both inpatient and outpatient services. • scores all hospitals based on the same absolute Other changes in payment policy largely offset each other. and prospectively set performance targets; For example, in 2018, CMS reduced Medicare payments accounts for differences in patients’ social risk • for separately payable Part B drugs at 340B hospitals, but factors by distributing payment adjustments CMS offset those decreases by increasing payments for through peer grouping, and other outpatient services. Some other regulatory changes • For 2020, update the 2019 Medicare base payment increased payments (e.g., higher uncompensated care rates for acute care hospitals by 2 percent. The payments in 2018 and 2019 due to expected increases in difference between the update recommendation and uninsured patients), but others decreased payments (e.g., the amount specified in current law should be used to reducing evaluation and management payment rates in increase payments in a new HVIP. 2019). The net result is that, from 2017 to 2019, payment rates increased by about 5 percent over two years after RATIONALE 3 accounting for case-mix change. We expect cost growth In examining our payment adequacy indicators, we found per discharge of about 3 percent per year in 2018 and that, in 2017, beneficiaries had good access to care, 2019, slightly faster than the past several years due to hospitals maintained strong access to capital markets, tighter labor markets. Given that costs are expected to and hospital quality improved, despite negative Medicare increase about 1 percent faster than payments, we expect margins for most providers. Looking forward, we expect overall Medicare margins to decline from –9.9 percent beneficiaries’ access to care to remain adequate given in 2017 to about –11 percent in 2019. We also expect the hospitals’ modest occupancy rates and good access to efficient provider margins to remain negative. The change capital. However, the aggregate Medicare profit margin in Medicare margins for 2020 will depend on whether cost is expected to decline to approximately –11 percent growth exceeds hospitals’ payment rate growth on a case- in 2019. Given these payment adequacy indicators, an mix-adjusted basis. update of 2 percent coupled with enhanced payments for hospitals with strong performance under the proposed HVIP would be high enough to maintain beneficiaries’ access to care and move payment rates closer toward the How should Medicare payment rates cost of efficiently delivering high-quality care. However, change in 2020? the 2 percent update would still be below the projected rate of input price inflation to maintain some pressure The Commission’s update recommendation for 2020 is on hospitals to constrain costs while improving quality. based on indicators of beneficiaries’ access to hospital The 2 percent update (rather than current law) would care, hospitals’ access to capital, hospital quality, and the also limit the growth in the differential between rates relationship between Medicare payments and hospital paid for physician office visits on a hospital campus and costs. As we discuss in Chapter 15, the Commission is also rates paid to freestanding physician offices. We expect recommending a new hospital value incentive program March 2019 | Report to the Congress: Medicare Payment Policy 89

120 the combination of a 2 percent update and the replacing payments that are equal to current law. In addition, eliminating the current readmissions penalty program of current quality incentives (which currently reduce and hospital acquired condition penalty would remove hospitals’ Medicare payments in aggregate) with the new HVIP (which would increase Medicare payments these penalties from hospital payment rates and thus in aggregate) would cause hospital Medicare margins to increase spending by between $750 million and $2 billion in 2020 and by $5 billion to $10 billion over improve from 2019 to 2020 given expected levels of cost five years. On net, hospital payment rates would be growth. We discuss the rationale for and implications of implementing a new HVIP in Chapter 15. expected to increase by an average of 3.3 percent. Beneficiary and provider IMPLICATIONS 3 We do not expect the recommendation to materially • Spending affect beneficiaries’ access to care or providers’ willingness to treat Medicare beneficiaries relative to • Current law is expected to increase payment rates by current law. Beneficiaries may benefit from hospitals’ 2.8 percent (a 3.3 percent market basket less a 0.5 enhanced incentives to improve the quality of care percent productivity adjustment). The recommended they provide and work with providers outside of the update of 2.0 percent with an increase in quality hospital to lower cost and improve outcomes. ■ incentive payments would result in total hospital Hospital inpatient and outpatient services: Assessing payment adequacy and updating payments 90

121 Endnotes because we do not consider any potential labor costs that Across all inpatient discharges, a handful of Medicare 1 are fixed. Using a cost-accounting approach, we find that severity–diagnosis related groups accounted for over half of approximately 20 percent of hospital costs are fixed, resulting the spending growth between 2016 and 2017. Specifically, in a marginal profit of about 8 percent. In our March 2015 heart failure and shock cases rose, increasing costs by $1.1 report to the Congress, we also took an econometric approach billion; chronic obstructive pulmonary disease cases, $600 to estimating hospitals’ marginal costs and found that fixed million; and septicemia cases, $600 million. costs were about 20 percent of overall costs for medium and Payments include roughly $7 billion of inpatient and 2 large hospitals. Small hospitals tend to have a lower share of outpatient payments to critical access hospitals (CAHs), costs that are variable and thus have higher marginal profits. which are paid 1 percent over their costs of inpatient, The finding that about 20 percent of costs are fixed at large outpatient, and post-acute care services in swing beds. CAHs hospitals also matches the 20 percent figure used in the do not receive disproportionate share hospital payments or Medicare outlier policy. For a discussion of our econometric uncompensated care payments. results and the literature on hospital marginal costs, see the online appendix to our 2015 report, available at http://www. In 2015 and 2016, CMS implemented several RAC program 3 medpac.gov (Medicare Payment Advisory Commission policies aimed at improving the accuracy of RACs’ auditing 2015b). of inpatient hospital claims. A list of these policies can be viewed at the following website: https://www.cms. Recent analysis performed by the Office of the Assistant 7 gov/Research-Statistics-Data-and-Systems/Monitoring- Secretary for Planning and Evaluation found that moving Programs/Medicare-FFS-Compliance-Programs/Recovery- making without to an all-condition hospital readmission Audit-Program/Downloads/Recovery-Audit-Program- any of the other changes suggested in our March 2013 Improvements-November-24-2017.pdf. Preceding these CMS package of recommendations would result in higher annual policy changes, the Commission recommended in its June penalties (Zuckerman et al. 2017). It is important to note 2015 report that the Secretary make several improvements to that any increase in penalties resulting from expanding to the RAC program (Medicare Payment Advisory Commission all conditions would be fully offset by the other changes we 2015a). discussed. Seven cancer drugs account for most of the increase in 4 Between 2010 and 2015, the Medicare share of hospital 8 OPPS spending on Part B drugs between 2016 and 2017: admissions rose from 42 percent to 44 percent. However, pembrolizumab, daratumumab, nivolumab, atezolizumab, during that period, because Medicare prices rose more slowly denosumab, rituximab, and trastuzumab. In aggregate, than commercial prices and due to additional revenue from payments to hospitals under the OPPS for these drugs the newly insured, Medicare’s share of all hospital revenues increased by approximately $1 billion from 2016 to 2017. remained at 33 percent. 5 Data concerning hospital outpatient observation care reflect The six largest departments in order of Medicare patient 9 services that are separately paid for under the Medicare revenues are inpatient acute care (60 percent), outpatient care OPPS system and not included in other APCs. While we (30 percent), inpatient rehabilitation (2.1 percent), inpatient report a decline from 2016 to 2017 in separately payable psychiatric (1.3 percent), home health care (0.7 percent), and outpatient observation visits, the volume of all outpatient skilled nursing services (0.4 percent). observation visits (separately paid or packaged with other 10 From fiscal year 2011 through 2017, we also considered outpatient services) increased 3 percent from 2016 to 2017, Medicare payments for health information technology; and 32 percent from 2012 to 2017. These figures indicate however, these payments ended for most IPPS hospitals as of that placement of patients in outpatient observation status is fiscal year 2016. increasingly common for beneficiaries. 11 The services included in the overall Medicare margin are 6 If we approximate marginal cost as total Medicare costs minus fixed building and equipment costs, then marginal Medicare’s acute inpatient, outpatient, graduate medical education, skilled nursing facility (including swing beds), profit can be calculated as follows: Marginal profit = hospital-based home health care, and inpatient psychiatric (payments for Medicare services – (total Medicare costs – fixed building and equipment costs)) / Medicare payments. and inpatient rehabilitation services. Also included in the overall margin are special payments for health information This comparison is a lower bound on the marginal profit technology, temporary extra payments to hospitals located in March 2019 | Report to the Congress: Medicare Payment Policy 91

122 coding. It is possible that overly aggressive coding by some low-spending counties, and uncompensated care payments (as providers could artificially lower their risk-adjusted cost and of fiscal year 2015). risk-adjusted mortality metrics. We use medians rather than means to limit the influence of 12 14 The 2,151 hospitals that met our screening criteria had levels outliers on our set of efficient providers. of profitability similar to the overall population of hospitals. ® (Hospital–Consumer Assessment of While HCAHPS 13 However, these hospitals tended to be larger than the average ® Healthcare Providers and Systems )—and similar patient hospital for two reasons. First, we screened out hospitals with satisfaction surveys—have the limitation of being subjective, fewer than 500 discharges due to instability in their costs we add it as another way to screen out low-value providers and quality indicators. Second, we excluded critical access because it has the advantage of not being dependent on hospitals due to their different cost accounting rules. Hospital inpatient and outpatient services: Assessing payment adequacy and updating payments 92

123 References Medicare Part B drugs: Government Accountability Office. 2015. Appleby, J. 2018. ‘Holy cow’ moment changes how Montana’s Action needed to reduce financial incentives to prescribe 340B state health plan does business. Kaiser Health News , June 20. . Washington, DC: GAO. drugs at participating hospitals Barclays. 2018. Health care services: Initiating coverage of VA health care: Government Accountability Office. 2013. hospital sector. August 14. Management and oversight of fee basis care need improvement . Berenson, R. A., J. H. Sunshine, D. Helms, et al. 2015. Why GAO–13–441. Washington, DC: GAO. Medicare Advantage plans pay hospitals traditional Medicare 2016 health care cost and Health Care Cost Institute. 2018. prices. Health Affairs 34, no. 8 (August): 1289–1295. Washington, DC: HCCI. utilization report. Bureau of Labor Statistics, Department of Labor. 2018a. May Hines, A. 2015. Trends in observed adult inpatient mortality for 2017 national occupational employment and wage estimates high-volume conditions, 2002–2012. Washington, DC: Agency United States. http://www.bls.gov/oes/. for Healthcare Research and Quality. Bureau of Labor Statistics. 2018b. National current employment The health care services Irving Levin Associates Inc. 2018. statistics, October 2018. https://www.bls.gov/ces/data.htm. acquisition report: 24th edition. Norwalk, CT: Irving Levin Bureau of Labor Statistics, Department of Labor. 2017. Associates Inc. Employment cost index – September 2017. Table 4. News Japsen, B. 2018. Tenet Healthcare signals outpatient acquisitions release. October 31. https://www.bls.gov/news.release/archives/ , February ahead amid value-based care push. Modern Healthcare eci_10312017.pdf. 27. Bureau of Labor Statistics, Department of Labor. 2016. May 2015 Krumholz, H. 2015. Mortality, hospitalizations, and expenditures national occupational employment and wage estimates United for the Medicare population aged 65 years or older 1999–2013. States. http://www.bls.gov/oes/. 314 no. 4 (July 28): 355–365. JAMA Industry Insights . November 20. https:// Cain Brothers. 2018. An analysis of private-sector Maeda, J., and L. Nelson. 2017. www.cainbrothers.com/research-type/newsletters/. Congressional Budget Office prices for hospital admissions. Census Bureau. 2018. Value of construction put in place survey working paper 2017–02. Washington, DC: CBO. at a glance. https://www.census.gov/construction/c30/c30index. Medicaid and CHIP Payment and Access Commission. 2016. html. Report to Congress on Medicaid disproportionate share hospital Clemens, J., and J. Gottlieb. 2017. In the shadow of a giant: Washington, DC: MACPAC. payments. Journal of Medicare’s influence on private physician payments. Medicare Payment Advisory Commission. 2018. Report to Political Economy 125, no. 1 (February): 1–39. the Congress: Medicare and the health care delivery system. Conn, J. 2017. Consumers fueling outpatient construction. Washington, DC: MedPAC. Modern Healthcare , March. Report to Medicare Payment Advisory Commission. 2015a. Politics, hospital Cooper, Z., A. Kowalski, E. Powell, et al. 2017. the Congress: Medicare and the health care delivery system. . NBER working paper no. behavior, and health care spending Washington, DC: MedPAC. 23748. Cambridge, MA: National Bureau of Economic Research. Medicare Payment Advisory Commission. 2015b. Report to the Fitch Ratings. 2018. 2018 median ratios for nonprofit hospitals Washington, DC: MedPAC. Congress: Medicare payment policy. and healthcare systems. September 20. Medicare Payment Advisory Commission. 2013. Report to Frakt, A. 2015. Hospitals are wrong about shifting costs to private the Congress: Medicare and the health care delivery system. , March 23. New York Times insurers. Washington, DC: MedPAC. Goldberg, S. 2018. As it scopes out its post-merger identity, Report to the Medicare Payment Advisory Commission. 2011. Modern Healthcare Amita looks beyond hospitals. , October 29. Congress: Medicare payment policy. Washington, DC: MedPAC. Report to the Congress: Medicare Payment Policy | March 2019 93

124 Medicare Payment Advisory Commission. 2007. Report to the Tosczak, M. 2018. No more negotiations: State health plan wants Congress: Promoting greater efficiency in Medicare. Washington, clearer process, lower prices. , North Carolina Health News DC: MedPAC. November 1. Moody’s Investors Service. 2018. Non-profit and public health Trish, E. E., and B. J. Herring. 2015. How do health insurer care: Medians–Operating pressures persist as growth in expenses market concentration and bargaining power with hospitals affect exceeds revenue. August 28. health insurance premiums? Journal of Health Economics 42 (July): 104–114. Reed, T. 2018. LifePoint Health leaves Nasdaq following completed merger with RCCH Healthcare Partners. Fierce White, C. 2013. Contrary to cost-shift theory, lower Medicare , November 16. http://www.fiercepharma.com/pharma/ Healthcare hospital payment rates for inpatient care lead to lower private big-three-pbms-cut-loose-more-specialty-pharmacies-as-scrutiny- payment rates. Health Affairs 32, no. 5 (May): 935–943. grows. White, C., and V. Y. Wu. 2014. How do hospitals cope with S&P Global Ratings. 2018. US not-for-profit acute health care Health Services sustained slow growth in Medicare prices? ratios: Sector is buffeted by disruption, yet 2017 median trends 49, no. 1 (February): 11–31. Research remain unchanged from last year. July 17. Zuckerman, R. B., K. E. Joynt Maddox, S. H. Sheingold, et al. Thomson Reuters. 2018. Hospital municipal bond issuances: As 2017. Effect of a hospital-wide measure on the Readmissions of October 29, 2018. 377, no. New England Journal of Medicine Reduction Program. 16 (October 19): 1551–1558. Hospital inpatient and outpatient services: Assessing payment adequacy and updating payments 94

125 CHAPTER 4 Physician and other health professional services

126 RECOMMENDATION 4 For calendar year 2020, the Congress should increase the calendar year 2019 Medicare payment rates for physician and other health professional services by the amount specified in current law. COMMISSIONER VOTES: YES 17 • NO 0 • NOT VOTING 0 • ABSENT 0

127 CHAPTER 4 Physician and other health professional services In this chapter Chapter summary Physicians and other health professionals deliver a wide range of services— • Are Medicare fee schedule including office visits, surgical procedures, and diagnostic and therapeutic payments adequate in 2019? services—in a variety of settings. In 2017, Medicare paid $69.1 billion for How should Medicare • physician and other health professional services, accounting for 14 percent of payments change in 2020? fee-for-service (FFS) Medicare benefit spending. About 985,000 clinicians billed Medicare: roughly 596,000 physicians and 389,000 nurse practitioners, physician assistants, therapists, chiropractors, and other practitioners. Medicare pays for the services of physicians and other health professionals using a fee schedule. Under current law, there is no update to Medicare’s conversion factor for the fee schedule on January 1, 2020. Assessment of payment adequacy We use the following factors to assess payment adequacy for physicians and other health professionals: beneficiaries’ access to care, the supply of providers, volume growth, quality, and Medicare payments and providers’ costs. Beneficiaries’ access to care— Overall, beneficiary access to physician and other health professional services is comparable with prior years. Most beneficiaries continue to report that they are able to find a new doctor without a problem. A small number of beneficiaries report more difficulty, with a March 2019 | Report to the Congress: Medicare Payment Policy 97

128 higher share reporting problems obtaining a new primary care doctor than problems obtaining a new specialist. • Supply of providers— The number of physicians per beneficiary declined slightly, the number of advanced practice registered nurses and physician assistants per beneficiary rose, and the share of providers enrolled in Medicare’s participating provider program remains high. • Volume of services— In 2017, across all services, volume per beneficiary grew by 1.6 percent. Among broad service categories, growth rates were 1.2 percent for evaluation and management services, 1.3 percent for imaging services, 2.1 percent for major procedures, 2.1 percent for other procedures, and 2.4 percent for tests. CMS assesses the quality of Medicare-billing physicians and Quality of care— other health professionals based on clinician-reported individual quality measures. We report three population-based measures: patient experience measures, avoidable hospitalizations for ambulatory care–sensitive conditions, and rates of low-value care in Medicare. Patient experience scores in Medicare FFS remain high, and rates of avoidable hospitalizations for ambulatory care–sensitive conditions continue to decline modestly from prior years, but there is substantial use of low-value care. CMS currently projects that the increase Medicare payments and providers’ costs— in 2020 in the Medicare Economic Index (which measures input prices) will be 2.4 percent. In 2017, Medicare FFS payment rates for physician and other health professional services averaged 75 percent of commercial rates paid by preferred provider organizations, unchanged from 2016. Median compensation in 2017 was much lower for primary care physicians than for physicians in certain specialties, such as radiology and nonsurgical, procedural specialties, continuing to raise concerns about fee schedule mispricing and its impact on primary care. The evidence suggests that Medicare payments for physicians and other health professionals are adequate. Therefore, the Commission recommends that the 2020 payment rate for physician and other health professional services be updated by the amount specified in current law. ■ Physician and other health professional services: Assessing payment adequacy and updating payments 98

129 are adjusted for variation in the input prices in different Background markets, and the sum is multiplied by the fee schedule’s conversion factor (average payment amount) to produce 1 Physicians and other health professionals billing a total payment amount. The conversion factor will be under Medicare’s fee schedule deliver a wide range $36.04 in 2019, up from $36.00 in 2018. of services—office visits, surgical procedures, and The Medicare Access and CHIP Reauthorization Act diagnostic and therapeutic services—in a variety of of 2015 (MACRA) established a set of updates for settings. clinicians billing under the Medicare fee schedule and The Medicare program paid $69.1 billion for physician repealed the prior framework that set the conversion and other health professional services in 2017, or 14 factor—the sustainable growth rate (SGR) formula. percent of benefit spending in Medicare’s traditional The SGR was established to limit total fee schedule fee-for-service (FFS) program. In 2017, about 985,000 spending by restraining annual updates when spending health professionals billed Medicare through the fee exceeded certain parameters. MACRA established two schedule—roughly 596,000 physicians and 389,000 paths for clinicians: one payment path for clinicians nurse practitioners, physician assistants, therapists, who participate in advanced alternative payment models chiropractors, and other practitioners. (A–APMs) and, for other clinicians, the Merit-based Incentive Payment System (MIPS) (Table 4-1). In 2020, Medicare uses a fee schedule to pay for physician and there is no statutory update for clinicians. Clinicians other health professional services based on a list of over qualifying for the A–APM incentive payment will receive 7,000 services and their payment rates. In determining an incentive payment of 5 percent of their professional payment rates for each service, CMS considers services payments in a lump sum. Clinicians remaining in the amount of clinician work required to provide a MIPS can receive payment adjustments of –5 percent to service, expenses related to maintaining a practice, and +5 percent (or higher) in 2020, based on performance. professional liability insurance costs. These three factors TABLE Statutory payment updates and incentive payments 4–1 for physicians and other health professionals 2026 2021 2022 and later 2024 2025 2019 2020 2023 A–APM clinicians Update 0.25% 0.75% 0% 0% 0% 0% 0% 0% 5% 5% 5% 5% 5% N/A N/A 5% APM bonus Other clinicians Update 0.25% 0% 0% 0% 0% 0% 0% 0.25% (–5% (–9% (–9% (–9% (–9% (–7% (–9% Potential MIPS (–4% to to to to to to to to adjustments +4%) +9%) +9%) +5%) +7%) +9%) +9%) +9%) A–APM (advanced alternative payment model), N/A (not applicable), MIPS (Merit-based Incentive Payment System). Clinicians who are subject to MIPS can receive Note: upward or downward adjustments of up to 4 percent in 2019, 5 percent in 2020, 7 percent in 2021, and 9 percent in 2022 and later. The MIPS maximum upward adjustment may exceed these limits or be less than these amounts because of scaling factors and an additional increase for exceptional performance. The basic MIPS adjustments are budget neutral, and there is an additional $500 million per year from 2019 to 2024 for exceptional performance under MIPS. The 5 percent incentive payment for A–APM participation expires after 2024. Source: Medicare Access and CHIP Reauthorization Act of 2015 and Bipartisan Budget Act of 2018, www.congress.gov. March 2019 | Report to the Congress: Medicare Payment Policy 99

130 the broader health care delivery system. This year’s survey TABLE Satisfaction with the overall was fielded in the summer and fall of 2018. 4–2 quality of health care received in all settings in the past 12 months, 2018 The Commission also conducts focus groups in markets around the country to provide a qualitative description of Private Medicare beneficiary and provider experiences with the Medicare insurance (ages 65 (ages 50–64) and older) program. This year, we conducted nine focus groups of Medicare beneficiaries in three markets, and we conducted 55% 68% Very satisfied a primary care physician focus group in each location. In 20 Somewhat satisfied 25 these markets, we also conducted site visits and interviews Somewhat dissatisfied 3 5 with various providers. Very dissatisfied 2 1 Overall, findings from our survey and focus groups are Note: Table excludes the following responses: “Did not receive health care 2 consistent with one another and similar to prior years. in past 12 months,” “Don’t know,” and “Refused.” It does not include Medicare beneficiaries under the age of 65. Medicare beneficiaries generally have adequate access to clinician services, and their reported access is largely Source: MedPAC-sponsored telephone survey conducted in 2018. comparable with (or in some cases, better than) access for privately insured individuals. Medicare beneficiaries’ overall satisfaction with care is similar to satisfaction among privately insured patients In our telephone survey, a slightly higher share of Are Medicare fee schedule payments Medicare beneficiaries reported that they were very or adequate in 2019? somewhat satisfied with their care (88 percent) compared with those who have private insurance (80 percent) (Table We assess payment adequacy by reviewing beneficiaries’ 4-2). access to care provided by physicians and other health professionals, the supply of physicians and other health Most beneficiaries report that they are able to see a doctor when they need to professionals, volume growth, quality of care, Medicare’s payment rates relative to commercial rates paid by Indicators from our 2018 telephone survey of access are preferred provider organizations, physician compensation largely comparable with prior years’ surveys. In particular, across specialties, and the change in input prices for in 2018, 70 percent of Medicare beneficiaries reported that physician and other health professional services. Overall, they never had to wait longer than they wanted for routine most indicators show no significant change from prior care, and 79 percent reported the same for illness or injury years. care (Table 4-3). Medicare beneficiaries were less likely to report trouble obtaining either type of care when needed Beneficiaries’ access to care than privately insured individuals (the rates for privately We use a number of measures to assess beneficiary access insured individuals were 64 percent for routine care and 74 to timely, appropriate care, including direct reporting from percent for illness or injury care). beneficiaries (through, for example, our own beneficiary Rates of access to timely regular or routine care for telephone survey); focus groups with beneficiaries; and both Medicare and privately insured individuals were health facility site visits conducted yearly. slightly worse in 2018 than in 2017, but Medicare access Each year, the Commission sponsors a telephone survey continued to be slightly better than access for privately of 4,000 Medicare beneficiaries ages 65 and over and insured individuals (Figure 4-1, p. 102). 4,000 privately insured individuals ages 50 to 64. The goal Medicare beneficiaries were also less likely than privately in surveying these two populations is to assess whether insured individuals to report that they waited longer than access concerns reported by Medicare beneficiaries are they wanted for care for illness or injury (Figure 4-2, p. unique to the Medicare population or are part of trends in 102). Physician and other health professional services: Assessing payment adequacy and updating payments 100

131 TABLE 4–3 Most aged Medicare beneficiaries and older privately insured individuals had good access to physician care, 2014–2018 Private insurance Medicare (ages 50–64) (ages 65 and older) 2017 2017 2014 2014 2016 2018 2018 2016 Survey question 2015 2015 Among those who needed an appointment in the past 12 months, “How often did you Unwanted delay in getting an appointment: have to wait longer than you wanted to get a doctor’s appointment?” For routine care a ab ab ab ab b ab a ab 68% 69% 73% 67% 72% 69% 70% 64% Never 69% 72% b a b a ab b ab a a 20 20 23 22 23 19 22 20 26 Sometimes 23 b b 3 4 3 4 5 Usually 4 5 4 5 4 b a a 3 3 4 3 4 3 3 Always 3 3 3 b b b b a a 2 2 1 1 2 2 1 1 1 2 Don’t know/Refused For illness or injury a a ab a a ab a ab a ab 79 79 75 80 76 79 74 82 77 Never 83 ab ab a a ab a a a 13 16 12 17 16 19 Sometimes 18 19 15 15 ab b a b Usually 2 2 2 2 3 3 3 2 2 3 a a ab ab b Always 1 2 2 1 2 3 2 2 2 2 b b b 2 1 1 1 Don’t know/Refused 2 1 1 1 1 2 Not accessing a doctor for medical problems: “During the past 12 months, did you have any health problem or condition about which you think you should have seen a doctor or other medical person, but did not?” a a b a a b Share answering “Yes” 11 11 11 12 12 14 12 10 11 11 Looking for a new doctor: “In the past 12 months, have you tried to get a new...?” (Share answering “Yes”) a ab ab b a a 8 Primary care doctor 7 10 10 11ª 10 8 9 9 8 a b a b b b b a a Specialist 17 19 18 18 20 17 21 16 18 17 Among those who tried to get an appointment with a new primary care physician or a specialist in the past 12 Getting a new physician: months, “How much of a problem was it finding a primary care doctor/specialist who would treat you? Was it...” Primary care physician a ab 63 63 63 59 71 67 64 69 No problem 67 67 b b b b Share of total insurance group 5.5 6.2 7.1 4.9 5.7 5.1 6.1 6.5 6.7 4.7 16 18 16 18 16 13 Small problem 16 18 15 13 a a Share of total insurance group 1.2 1.2 1.2 1.3 1.3 1.7 1.5 2.0 1.6 1.3 a a 19 20 22 17 16 15 14 20 14 Big problem 14 a a 1.4 1.5 1.5 1.9 2.4 1.7 Share of total insurance group 1.3 1.2 1.6 1.0 Specialist a b a No problem 87 85 85 79 82 82 83 81 80 84 b b b b 14.4 Share of total insurance group 14.2 14.7 14.8 14.1 16.1 14.5 16.2 17.1 14.4 b Small problem 7 7 10 11 7 9 8 9 11 9 1.4 Share of total insurance group 1.8 1.9 1.4 1.1 1.5 1.6 2.2 2.0 1.2 a a a a 11 8 8 6 9 6 7 8 Big problem 10 5 a a b a ab 1.2 1.0 Share of total insurance group 1.0 0.9 1.7 1.4 2.0 1.6 1.5 2.0 Note: Numbers may not sum to 100 percent because of rounding. Sample sizes for each group (Medicare and privately insured) are 4,000. Sample sizes for individual questions varied. “Aged” beneficiaries are those ages 65 or older. a Statistically significant difference between the Medicare and privately insured groups in the given year (at a 95 percent confidence level). b Statistically significant difference from 2018 within the same insurance category (at a 95 percent confidence level). Source: MedPAC-sponsored telephone surveys conducted from 2014 to 2018. March 2019 | Report to the Congress: Medicare Payment Policy 101

132 FIGURE FIGURE Title here... Among patients seeking care, share who ever waited longer than X-X 4–1 wanted for regular or routine care, Medicare and private insurance Medicare Private insurance 40 40 Always Always Usually Usually Sometimes Sometimes 30 30 20 20 Percent Percent 10 10 0 0 2010 2008 2006 2016 2018 2014 2012 2006 2008 2010 2016 2014 2012 2018 Source: MedPAC-sponsored telephone surveys, 2006–2018. FIGURE FIGURE Title here... Among patients seeking care, share who ever waited longer than 4–2 X-X wanted for illness or injury care, Medicare and private insurance Note: Note and Source are in InDesign. Medicare Private insurance Source: 30 30 Always Always Usually Usually Notes about this graph: Sometimes Sometimes • Data is in the datasheet. Make updates in the datasheet. 20 20 • I deleted the years from the x-axis and put in my own. • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. • The dashed line looked ok here, so I didn’t hand draw it. Percent Percent • I can’t delete the legend, so I’ll just have to crop it out in InDesign. 10 10 • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph default when you change the data. • Use paragraph styles (and object styles) to format. 0 0 2016 2014 2012 2010 2008 2006 2008 2010 2012 2014 2016 2018 2018 2006 Source: MedPAC-sponsored telephone surveys, 2006–2018. Physician and other health professional services: Assessing payment adequacy and updating payments 102 Note: Note and Source are in InDesign. Source: Notes about this graph: • Data is in the datasheet. Make updates in the datasheet. • I deleted the years from the x-axis and put in my own. • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. • The dashed line looked ok here, so I didn’t hand draw it. • I can’t delete the legend, so I’ll just have to crop it out in InDesign. • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph default when you change the data. • Use paragraph styles (and object styles) to format.

133 FIGURE FIGURE Among those looking, share of respondents who indicated trouble 4–3 Title here... finding a new primary care doctor, Medicare and private insurance X-X Medicare Private insurance 50 50 A small problem A small problem A big problem A big problem 40 40 30 30 Percent Percent 20 20 10 10 0 0 2006 2008 2012 2012 2014 2016 2018 2014 2016 2018 2010 2008 2006 2010 Note: The share of respondents looking for a new doctor each year is about 10 percent for primary care. Therefore, the share of Medicare respondents facing a problem (small or big) in obtaining a new primary care doctor was 2.7 percent in 2018, and the share of private insurance respondents facing a problem (small or big) was 4.3 percent in 2018. Source: MedPAC-sponsored telephone surveys, 2006–2018. Note: Note and Source are in InDesign. Source: big problem, meaning that, on net, 1.5 percent of the total Beneficiaries report more difficulty accessing primary care than specialty care Medicare population reported a big problem. Notes about this graph: We also ask respondents whether, when they are looking This pattern of greater difficulty for Medicare for a new doctor, they are able to find one without • Data is in the datasheet. Make updates in the datasheet. beneficiaries (among those looking) in finding a new difficulty. Most beneficiaries reported that they were able primary care doctor relative to finding a specialist is • I deleted the years from the x-axis and put in my own. to find a new doctor without a problem. consistent with prior years, other surveys, and our • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. beneficiary focus groups. Consistent with prior years, beneficiaries looking for a • The dashed line looked ok here, so I didn’t hand draw it. new doctor generally reported more problems finding one However, overall, Medicare beneficiaries continue to be • I can’t delete the legend, so I’ll just have to crop it out in InDesign. when seeking a new primary care doctor than seeking slightly less likely than individuals with private insurance a new specialist (Table 4-3, p. 101). For primary care, • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph to report problems obtaining primary and specialty care 10 percent were looking for a new doctor, and of those default when you change the data. (Figure 4-3, this page, and Figure 4-4, p. 104). looking, 14 percent reported a big problem, meaning that, • Use paragraph styles (and object styles) to format. on net, 1.4 percent of the Medicare population reported a Beneficiaries in the Commission’s telephone survey big problem. For specialty care, 19 percent were looking reported difficulty with certain specialty referrals, namely for a new doctor; of those looking, 8 percent reported a dermatologists (likely due to specialization in cosmetic dermatology vs. medical dermatology), psychiatrists, and neurologists. March 2019 | Report to the Congress: Medicare Payment Policy 103

134 FIGURE FIGURE Among those looking, share of respondents indicating 4–4 Title here... trouble finding a new specialist, Medicare and private insurance X-X Medicare Private insurance 25 25 A small problem A small problem A big problem A big problem 20 20 15 15 Percent Percent 10 10 5 5 0 0 2014 2016 2018 2014 2016 2018 2008 2012 2006 2006 2008 2010 2012 2010 Note: The share of respondents looking for a new doctor each year is about 20 percent for specialty care. Therefore, the share of Medicare respondents facing a problem (small or big) in obtaining a new specialist was 2.8 percent in 2018, and the share of private insurance respondents facing a small or big problem was 4.0 percent in 2018. Source: MedPAC-sponsored telephone surveys, 2006–2018. Note: Note and Source are in InDesign. Source: Specifically, minority Medicare beneficiaries were more Some groups of beneficiaries report more difficulty obtaining care likely than non-Hispanic White Medicare beneficiaries to report that they always had to wait longer than they In our telephone survey, minority beneficiaries were more Notes about this graph: wanted for a routine doctor’s appointment (5 percent vs. likely than (non-Hispanic) White beneficiaries to report • Data is in the datasheet. Make updates in the datasheet. 2 percent, respectively). Similar to prior years’ findings, that they could not obtain care as quickly as they wanted. minority Medicare beneficiaries were also more likely • I deleted the years from the x-axis and put in my own. As in prior years, differences in reported access between than non-Hispanic White beneficiaries to say that they did • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. urban and rural beneficiaries were minimal. not receive care when they thought they should have (15 • The dashed line looked ok here, so I didn’t hand draw it. percent for minority beneficiaries vs. 10 percent for non- Minority beneficiaries reported more difficulty receiving Hispanic White beneficiaries). • I can’t delete the legend, so I’ll just have to crop it out in InDesign. care as soon as they wanted and higher rates of forgoing care We continue to find through the Commission’s • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph Minority Medicare beneficiaries also reported higher rates telephone survey that Medicare beneficiaries who belong of problems finding a specialist, and a similar pattern default when you change the data. to racial or ethnic minority groups are more likely to report exists for privately insured minority individuals. Although • Use paragraph styles (and object styles) to format. waiting longer than they want for regular or routine care the small sample sizes of the Commission’s survey than non-Hispanic White beneficiaries, consistent with generally do not permit us to detect significant differences general trends in poorer access to health care among racial in reported access among Black (or African American) 3 and ethnic minority groups (Table 4-4). and Hispanic (or Latinx) beneficiaries separately, Physician and other health professional services: Assessing payment adequacy and updating payments 104

135 TABLE Medicare beneficiaries had similar access to physicians compared with privately insured 4–4 individuals, but minorities in both groups reported problems more frequently, 2018 Medicare Private insurance (ages 65 and older) (ages 50–64) White Minority All White Minority Survey question All Among those who needed an appointment in the past 12 months, “How often did you Unwanted delay in getting an appointment: have to wait longer than you wanted to get a doctor’s appointment?” For routine care a ab b b ab a 70% Never 65% 61% 65% 71% 64% a a a a a a Sometimes 26 20 25 20 29 21 Usually 5 5 5 4 5 5 b a ab ab b a Always 3 5 4 2 6 4 a a a ab a ab 3 2 2 1 Don’t know/Refused 2 * For illness or injury a ab b a ab b Never 79 75 75 71 74 80 a a a a a 15 15 19ª 19 22 15 Sometimes Usually 2 3 3 3 4 2 b b Always 2 2 2 3 2 3 ab ab a a Don’t know/Refused 1 2 2 1 2 3 Not accessing a doctor for medical problems: “During the past 12 months, did you have any health problem or condition about which you think you should have seen a doctor or other medical person, but did not?” b a a a ab 15 14 10 13 16 Share answering “Yes” 11 “In the past 12 months, have you tried to get a new...?” (Share answering “Yes”) Looking for a new doctor: Primary care physician 10 10 10 9 11 9 b a b b a b Specialist 19 23 21 19 15 20 Getting a new physician: Among those who tried to get an appointment with a new primary care physician or a specialist in the past 12 months, “How much of a problem was it finding a primary care doctor/specialist who would treat you? Was it...” Primary care physician b b 71 71 69 67 72 No problem 59 Share of total insurance group, by race 7.1 7.2 6.2 6.7 6.7 6.7 14 Small problem 13 17 14 16 15 Share of total insurance group, by race 1.3 1.4 1.3 1.6 1.4 1.9 b b Big problem 14 15 14 16 14 23 b b 1.3 1.7 1.3 Share of total insurance group, by race 2.5 1.5 1.4 Specialist b b b b 84 86 No problem 82 80 74 77 b b b b 14.0 11.9 Share of total insurance group, by race 17.1 18.5 16.1 17.3 Small problem 10 9 9 11 7 7 1.4 1.4 1.6 2.0 2.1 2.1 Share of total insurance group, by race b b b b 8 7 Big problem 8 13 13 10 Share of total insurance group, by race 2.5 1.4 2.0 2.0 1.8 1.5 Respondents who did not report race or ethnicity were not included in “White” or “Minority” results but were included in “All” results. “White” in the table refers to non- Note: Numbers may not sum to 100 percent because of rounding. Sample sizes for each group (Medicare and privately insured) were 4,000 Hispanic White respondents. in 2018. Sample sizes for individual questions varied. a Statistically significant difference between the Medicare and privately insured populations in the given year (at a 95 percent confidence level). b Statistically significant difference by race within the same insurance category in the given year (at a 95 percent confidence level). Source: MedPAC-sponsored telephone surveys conducted in 2018. March 2019 | Report to the Congress: Medicare Payment Policy 105

136 TABLE 4–5 Access to physician care for Medicare beneficiaries was similar to or slightly better than access for privately insured individuals in urban and rural areas, 2018 Private insurance Medicare (ages 50–64) (ages 65 and older) Rural All Urban Survey question All Urban Rural Among those who needed an appointment in the past 12 months, “How often did you Unwanted delay in getting an appointment: have to wait longer than you wanted to get a doctor’s appointment?” For routine care a b a ab a 64% 68% 68% 70% 63% Never 70% a a a a a a 26 19 20 24 20 Sometimes 26 ab b a 4 5 6 Usually 5 5 4 a a a a 4 3 4 3 3 Always 4 a a 2 2 2 2 1 2 Don’t know/Refused For illness or injury a a a a a a Never 74 78 73 79 74 79 a a a a a a Sometimes 15 15 19 21 19 15 a a Usually 2 2 3 3 3 3 Always 2 2 2 2 2 2 Don’t know/Refused 1 1 1 2 2 2 Not accessing a doctor for medical problems: “During the past 12 months, did you have any health problem or condition about which you think you should have seen a doctor or other medical person, but did not?” a a a a (Share answering “Yes”) 11 11 13 14 11 14 Looking for a new primary care physician: “In the past 12 months, have you tried to get a new...?” ( Share answering “Yes”) 10 10 10 Primary care physician 10 9 9 ab a a a b Specialist 21 22 19 19 18 17 Getting a new physician: Among those who tried to get an appointment with a new primary care physician or a specialist in the past 12 months, “How much of a problem was it finding a primary care doctor/specialist who would treat you? Was it...” Primary care physician 68 67 No problem 72 68 64 71 7.1 Share of total insurance group, by area 6.0 6.7 6.8 5.9 7.1 15 16 12 13 Small problem 15 13 Share of total insurance group, by area 1.3 1.2 1.2 1.6 1.6 1.4 Big problem 14 13 18 16 15 21 1.4 Share of total insurance group, by area 1.6 1.4 1.7 1.5 1.9 Specialist 86 84 No problem 84 80 81 82 b b 15.7 17.6 16.1 16.1 Share of total insurance group, by area 17.1 14.0 7 5 Small problem 9 9 8 8 Share of total insurance group, by area 1.4 1.6 0.9 2.0 2.1 1.4 Big problem 7 9 10 9 10 8 a a 2.0 2.1 1.7 Share of total insurance group, by area 1.5 1.3 1.6 Note: Numbers may not sum to 100 percent because of rounding. Sample sizes for each group (Medicare and privately insured) were 4,000 in 2018. Sample sizes for individual questions varied. The Commission uses the Census Bureau definitions of “urban” and “rural.” The Census Bureau classifies as urban all territory, population, and housing units located within an urbanized area (UA) or an urban cluster (UC). It delineates UA and UC boundaries to encompass densely settled territory, which consists of core census block groups or blocks that have a population density of at least 1,000 people per square mile and surrounding census blocks that have an overall density of at least 500 people per square mile. In addition, under certain conditions, less densely settled territory may be part of each UA or UC. The Census Bureau’s classification of rural consists of all territory, population, and housing units located outside of UAs and UCs. a Statistically significant difference between the Medicare and privately insured populations in a given year (at a 95 percent confidence level). b (at a 95 percent confidence level). Statistically significant difference by area type within the same insurance category in a given year Source: MedPAC-sponsored telephone survey conducted in 2018. Physician and other health professional services: Assessing payment adequacy and updating payments 106

137 as payment in full); and the share of claims that are paid Hispanic Medicare beneficiaries were more likely to on assignment. report waiting longer than wanted for regular or routine care than non-Hispanic White beneficiaries. In addition, Supply of physicians and other health Black beneficiaries are significantly more likely than non- professionals billing Medicare has kept pace with Hispanic White beneficiaries to report that they should enrollment growth have seen a doctor but did not (15 percent vs. 10 percent Our analysis of Medicare FFS claims data for 2015 to for non-Hispanic White beneficiaries) (data not shown). 2017 shows that the number of physicians and other health Few reported differences in access between urban professionals furnishing services to Medicare beneficiaries and rural beneficiaries Similar to prior years, the is growing and has generally kept pace with enrollment Commission’s telephone survey showed no major growth in Medicare (Table 4-6, p. 108). Between 2016 and differences in access between urban and rural beneficiaries 2017, the number of primary care physicians increased 4 (Table 4-5). There was no significant difference between by 1 percent, from almost 185,000 to just over 186,000. the share of urban and rural beneficiaries experiencing Because the number of beneficiaries grew by almost 3 an unwanted delay in getting an appointment. Urban percent between 2016 and 2017 (data not shown), the ratio beneficiaries reported more timely access to routine of primary care physicians to the number of beneficiaries care than urban individuals with private insurance. dropped slightly, from 3.6 per 1,000 beneficiaries to 3.5 5 However, differences between rural beneficiaries and rural per 1,000. Similarly, the number of physicians in other individuals with private insurance were minimal and not specialties grew by 1 percent between 2016 and 2017, statistically significant in most cases. from nearly 406,000 to almost 410,000, but the number per 1,000 beneficiaries declined slightly from 7.8 to Nearly all beneficiaries have a regular source of 7.7. Meanwhile, during the same period, the number care, with more use of nurse practitioners and of advanced practice registered nurses and PAs billing physician assistants in rural areas Medicare grew by 10 percent, and the number per 1,000 Nearly all beneficiaries in the Commission’s survey beneficiaries rose from 3.9 to 4.2. reported that they had a regular source of primary care— 94 percent in 2018 (data not shown). Beneficiaries in focus Most physicians and other health professionals groups generally responded that they could access their are part of Medicare’s participating provider provider the same day or within a few days. program, and nearly all claims are paid on assignment In the Commission’s telephone survey, 16 percent of In 2018, 96 percent of physicians and other health beneficiaries responded that they saw a nurse practitioner professionals billing Medicare signed an agreement (NP) or physician assistant (PA) for all or most of their with Medicare to be part of the participating provider primary care, and 29 percent said that they saw an NP or program. Participating providers agree to take assignment PA for some of their primary care (these numbers have for all claims, which means they accept the fee schedule continued to rise gradually over time). Similar to prior amount as payment in full (almost all claims are paid on years, rural beneficiaries were more likely than urban 6 assignment). Providers who do not elect to participate beneficiaries to report seeing NPs and PAs for all or most receive a 5 percent lower payment amount and can choose of their primary care (21 percent for rural beneficiaries vs. whether to take assignment for their claims on a claim-by- 14 percent for urban beneficiaries) (data not shown). claim basis. If they do not assign a claim, providers may “balance bill” up to 109.25 percent of the fee schedule Supply of physicians and other health amount, with the beneficiary paying the difference professionals billing Medicare has kept pace between 95 percent of the fee schedule amount and the with enrollment growth, and most clinicians amount billed. Clinicians can also opt out of the Medicare are in a participating provider arrangement with Medicare program and treat patients entirely outside of the Medicare benefit. Opt-out clinicians are concentrated in the provider Other indicators of access include the supply of clinicians specialties of dentistry and behavioral health (including billing Medicare; the share of physicians and other health psychiatry). The number of clinicians who opted out of professionals who are participating providers (which Medicare in 2018 is largely consistent with prior years means that they are required to accept Medicare’s payment (data not shown). March 2019 | Report to the Congress: Medicare Payment Policy 107

138 TABLE Number of physicians and other health professionals billing Medicare, 2015–2017 4–6 Physicians Advanced practice registered nurses and Other specialties Primary care specialties Other practitioners physician assistants Number Number Number Number per 1,000 per 1,000 per 1,000 per 1,000 Year beneficiaries beneficiaries Number Number beneficiaries Number beneficiaries Number 182,949 3.1 155,310 3.6 7.9 400,303 3.6 182,767 2015 184,905 3.6 405,780 7.8 202,874 3.9 160,661 3.1 2016 2017 186,193 409,995 7.7 223,567 4.2 165,486 3.1 3.5 Note: Specialty is self-reported by physicians and other health professionals when they enroll in the Medicare program. “Primary care specialties” are specialties that were eligible for the Primary Care Incentive Payment program: family medicine, internal medicine, pediatric medicine, and geriatric medicine. In 2017, CMS introduced a new physician specialty code for hospitalists. Most of the physicians who billed Medicare as hospitalists in 2017 billed as a primary care specialty in 2016. To maintain consistency across years, we assigned physicians who billed as hospitalists in 2017 to the “primary care specialties” group. “Other practitioners” includes physical and occupational therapists, chiropractors, optometrists, psychologists, social workers, and podiatrists. The number billing Medicare includes those with a caseload of more than 15 different beneficiaries during the year. Beneficiary counts used to calculate numbers per 1,000 include those in fee-for-service Medicare and Medicare Advantage on the assumption that physicians and other health professionals are furnishing services to beneficiaries in both programs. Figures for 2015 and 2016 may vary from figures that appeared in prior Commission reports due to minor technical changes. Figures exclude nonperson providers such as suppliers or clinical laboratories. MedPAC analysis of Medicare claims data for 100 percent of beneficiaries and the 2018 annual report of the Boards of Trustees of the Medicare trust funds. Source: change in service use: units of service per beneficiary and Higher growth in the volume of clinician volume of services per beneficiary. Volume is measured as services units of service multiplied by each service’s relative value We analyze annual changes in use of services provided units (RVUs) from the fee schedule. Our volume growth by physicians and other health professionals as another measure thus accounts for changes in both the number of indicator of payment adequacy. However, we recommend services and the complexity, or intensity, of those services. caution in interpreting such data because factors unrelated For example, growth in the volume of imaging services to Medicare’s payment rates can influence service volume. would account not just for any change in the number of Evidence indicates that volume decreases could be related such services but also for any change in intensity (e.g., to the movement of services from freestanding offices if providers substitute computed tomography (CT) scans to hospitals and to general practice pattern changes. For for less complex X-rays). We used RVUs for 2017 to example, the number of echocardiograms per beneficiary put service volume for all years on a common scale. administered in freestanding offices declined in 2017 by We grouped individual service codes into broad service 0.3 percent, while the number administered in hospital categories that are clinically meaningful (e.g., evaluation outpatient departments (HOPDs) rose by 2.0 percent. and management (E&M)). Each broad service category Increases in volume can signal overpricing if practitioners contains multiple subcategories of similar services (e.g., favor certain services because they are relatively E&M includes office/outpatient services, hospital inpatient profitable, but other factors—including changes in the services, and other subcategories). population, disease prevalence, Medicare benefits, site of care, technology, and beneficiaries’ preferences—can also Between 2016 and 2017, across all services, volume explain volume increases. per beneficiary grew by 1.6 percent (Table 4-7). Among broad service categories, growth rates were 1.2 percent We used claims data from 2012, 2016, and 2017 to analyze for E&M, 1.3 percent for imaging services, 2.1 percent volume changes. We identified the services furnished for major procedures, 2.1 percent for other procedures, by physicians and other professionals billing under and 2.4 percent for tests. The 2017 growth rates for all Medicare’s fee schedule and calculated two measures of Physician and other health professional services: Assessing payment adequacy and updating payments 108

139 TABLE Use of clinician services per FFS beneficiary, 2012–2017 4–7 Change in volume Change in units of service per beneficiary per beneficiary Share of 2017 allowed Average annual Average annual Type of service charges 2012–2016 2016–2017 2012–2016 2016–2017 1.3% All services 1.6% 100.0% 0.7% 1.0% 0.2 1.0 1.2 52.8 Evaluation and management 0.6 1.0 0.5 1.6 1.0 26.9 Office/outpatient services –1.8 Hospital inpatient services –0.5 11.5 –1.2 –1.5 0.4 1.2 0.0 3.3 Emergency department services –0.8 1.9 1.0 2.6 2.1 3.0 Nursing facility services –0.2 0.1 0.2 2.8 –0.1 Ophthalmological services N/A 3.0 N/A 2.9 1.9 Behavioral health services Critical care services 0.9 3.9 0.9 3.8 1.5 Care management/coordination 20.1 30.3 40.7 0.8 31.7 8.2 7.9 1.1 0.7 Observation care services 0.7 –0.6 –3.6 –0.5 –3.4 0.3 Home services Imaging 0.2 –0.2 1.3 11.4 1.4 Standard X-ray 0.8 –0.3 –0.1 3.0 –0.8 Ultrasound –0.1 –0.1 –1.3 –0.1 3.0 CT 3.0 2.7 4.9 2.1 4.8 MRI 2.4 1.3 2.3 1.3 1.9 Nuclear –3.8 –1.4 –3.8 1.0 1.3 Major procedures 0.5 0.6 2.2 2.1 7.8 2.9 Musculoskeletal 2.4 1.3 3.2 2.2 Vascular –1.1 1.3 9.5 8.0 0.0 Cardiovascular 1.5 0.5 1.0 2.0 1.6 –1.1 –0.6 0.4 1.0 Other organ systems 0.3 –2.7 –1.5 –1.9 –1.4 0.8 Digestive/gastrointestinal 0.4 0.1 –0.9 0.5 –0.4 Skin –0.5 –1.8 –0.4 –1.7 0.2 Eye Other procedures 1.8 2.4 1.4 2.1 23.1 Skin 1.6 1.7 1.3 4.6 0.8 Physical, occupational, and speech therapy 5.8 5.5 6.2 3.9 5.0 Musculoskeletal 0.9 0.2 1.3 2.6 2.6 Eye 1.5 2.1 1.0 1.5 2.4 2.0 Radiation oncology –1.8 3.5 –1.4 2.3 Other organ systems 1.7 2.5 0.7 1.5 2.4 Digestive/gastrointestinal –0.5 0.1 0.0 1.3 –0.3 Dialysis –1.7 0.2 –0.3 1.2 –1.4 Vascular –0.7 –0.5 3.2 2.7 1.1 Chiropractic –1.9 –2.6 –2.0 –2.7 0.8 –0.7 Injections and infusions: non-oncologic –2.5 0.5 –2.5 –0.8 Chemotherapy administration –3.3 –3.2 –5.7 0.5 –4.9 Tests 1.2 –0.2 2.4 4.6 0.3 Anatomic pathology –0.2 1.0 –0.4 1.7 1.3 Cardiography –0.8 1.3 –1.6 4.2 1.3 0.9 1.8 Neurologic 1.7 0.3 0.7 FFS (fee-for-service), CT (computed tomography), MRI (magnetic resonance imaging), N/A (not available). Volume is measured as units of service multiplied by Note: each service’s relative value units (RVUs) from Medicare’s fee schedule for physicians and other health professionals. To put service use in each year on a common scale, we used the RVUs for 2017. For billing codes not used in 2017, we imputed RVUs based on the average change in RVUs for each type of service. Use of behavioral health services is not reported for 2012 to 2016 because of a change in billing codes implemented in 2013. Some low-volume categories are not shown but are included in the summary calculations. Totals may not sum due to rounding. Source: MedPAC analysis of claims data for 100 percent of Medicare beneficiaries. March 2019 | Report to the Congress: Medicare Payment Policy 109

140 other procedures to treat peripheral artery disease (PAD) in FIGURE FIGURE Volume growth has raised... 4-X the lower extremities. These procedures have higher RVUs Growth in the volume of 4–5 clinician services per FFS than other procedures in the same subcategory. beneficiary, 2000–2017 Among the service subcategories, care management/ 100 coordination had the highest rate of volume growth: 30.3 Major procedures percent per year from 2012 to 2016 and 40.7 percent in 2017. However, this subcategory had a very low level 80 E&M of volume in 2012 (data not shown). CMS created new billing codes for transitional care management (TCM) Other procedures 60 in 2013 and chronic care management (CCM) in 2015. In 2016, CMS established a billing code for monthly Tests enhanced oncology services for the Oncology Care Model 40 (OCM). The OCM, CCM, and TCM services account for Imaging most of the growth in care management/coordination. In Tests Imaging Cumulative percent change 20 2017, the volume of OCM services increased by 147.8 Other procedures percent, CCM increased by 59.9 percent, and TCM E&M services increased by 19.4 percent (data not shown). At the same Major procedures 0 time, the volume of the other services in this subcategory 2000 2004 2008 2010 2012 2006 2014 2016 2002 (physician certification and recertification of home health care, home health care supervision, and hospice care Note: FFS (fee-for-service). E&M (evaluation and management). Volume growth for supervision) decreased by 2.8 percent (data not shown). E&M from 2009 to 2010 is not directly observable because of a change Although care management/coordination experienced high in payment policy for consultations. To compute cumulative volume growth for E&M through 2017, we used a growth rate for 2009 to 2010 of 1.85 volume growth, it accounts for less than 1 percent of total percent, which is the average of the 2008 to 2009 growth rate of 1.7 fee schedule spending. percent and the 2010 to 2011 growth rate of 2.0 percent. The type-of- service categories were restructured starting with the growth rates for 2016. While volume growth for imaging in 2017 was slightly MedPAC analysis of claims data for 100 percent of Medicare Source: lower than the average increase for all services and follows beneficiaries. Note and Source in InDesign. Note: a slight decrease from 2012 to 2016, use of imaging services remains much higher than it was in 2000 (Figure 4-5). Cumulative growth in the volume of imaging per beneficiary from 2000 to 2017 totaled 75 percent, which was much higher than cumulative growth during the services and for broad service categories were higher than Notes about this graph: same period for major procedures and E&M services (47 the average annual growth rates from 2012 to 2016, except • Data is in the datasheet. Make updates in the datasheet. percent and 45 percent, respectively). In addition, volume for major procedures (2.1 percent increase in 2017 vs. 2.2 increases in 2017 were higher for certain types of imaging • I had to force return the items on the x-axis. They will reflow if I update the data. percent average annual growth from 2012 to 2016). than others. For example, in 2017, the volume of CT grew • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. 4.9 percent (Table 4-7, p. 109). By contrast, from 2012 Subcategories of a broad service category sometimes to 2016, average annual volume growth of CT was 2.7 • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph experienced more rapid volume growth in 2017 than the percent. Similarly, in 2017, MRI volume increased 2.3 default when you change the data. broad service category. For example, volume growth in percent, compared with an average annual increase from the other procedures category was 2.1 percent, but volume • Use paragraph styles (and object styles) to format. 2012 to 2016 of 1.3 percent (Table 4-7, p. 109). growth in the subcategory of physical, occupational, and speech therapy was 6.2 percent (physical therapy In response to concerns about overuse of imaging, tests, accounted for most of this growth). and procedures, the American Board of Internal Medicine (ABIM) Foundation developed the “Choosing Wisely” Some service subcategories exhibited large increases campaign. As part of this ongoing effort, more than 80 in intensity. For example, within major procedures, the specialty societies have identified over 550 tests and vascular procedures subcategory had no change in units procedures that are often overused (ABIM Foundation of service in 2017 but a 9.5 percent increase in volume. 2016). The goal of Choosing Wisely is to promote The difference was due to rapid growth of angioplasty and Physician and other health professional services: Assessing payment adequacy and updating payments 110

141 a freestanding office. For example, in 2018, the total and inform conversations between clinicians and their payment for the most common E&M office/outpatient patients about appropriate tests and treatments. As part of visit for an established patient when provided in an HOPD Choosing Wisely, the Society for Vascular Medicine has (other than certain off-campus HOPDs) was $166 ($52 cautioned that patients with PAD usually do not need to for the fee schedule payment to the clinician plus $114 have a procedure (ABIM Foundation 2017). Nevertheless, for the facility payment to the HOPD) compared with $74 the number of procedures to treat PAD in the lower (the nonfacility fee schedule payment) for this visit when extremities grew rapidly in 2017. provided in a freestanding office. In addition, CMS is developing the Appropriate Use In recent years, there has been a trend toward billing for Criteria (AUC) Program that will require clinicians some services in hospitals instead of freestanding offices. to use clinical decision support (CDS) software when From 2013 to 2017, for example, the number of outpatient they order advanced diagnostic imaging services for hospital–based E&M visits per beneficiary grew by 19.4 beneficiaries. Under this program, clinicians who order percent, compared with a 3.5 percent decline in physician these services will need to consult with CDS software office–based E&M visits. During the same period, the and obtain feedback on whether the services adhere to number of chemotherapy administration services per AUC developed by medical societies or other provider-led beneficiary delivered in HOPDs grew 28.7 percent, while entities. CMS is in the process of developing this program, the number provided in physician offices declined 13.1 which is scheduled to begin on January 1, 2020. percent. This change in the billed setting increases overall Volume changes reflect shift in billing from Medicare program spending and beneficiary cost sharing freestanding offices to hospitals because Medicare generally pays more for the same or similar services in HOPDs than in freestanding offices Measuring volume growth as units of service multiplied (Medicare Payment Advisory Commission 2014, Medicare by each service’s RVUs has two advantages. First, volume Payment Advisory Commission 2013, Medicare Payment growth accounts for changes not just in the number Advisory Commission 2012). For example, we estimate of services but also in the intensity of services (e.g., that the Medicare program spent $1.9 billion more in substitution of CT scans for X-rays). Second, volume 2017 than it would have if payment rates for E&M office/ growth is important because it has a significant impact on outpatient visits in HOPDs were the same as freestanding spending growth, along with changes in payment rates. office rates. In addition, beneficiaries’ cost sharing for Volume growth, however, is sensitive to shifts in the site of E&M office/outpatient visits in HOPDs was $480 million care. The RVUs used to calculate volume include practice higher in 2017 than it would have been had payment rates expenses, which are often lower for services provided in a been the same in both settings. facility setting, such as an HOPD, compared with services To address the increased spending that results when in a nonfacility setting, such as a freestanding office. services shift from freestanding offices to HOPDs, the In 2018, for example, the most common type of E&M Commission recommended adjusting payment rates in office/outpatient visit for an established patient (Current the outpatient prospective payment system (OPPS) so Procedural Terminology code 99213) had an average 7 that Medicare pays the same amount for E&M office/ nonfacility fee schedule payment of $74. By contrast, the outpatient visits in freestanding physician offices and average fee schedule payment for this visit when provided HOPDs (Medicare Payment Advisory Commission 2012). in a facility setting was $52 because the practice expense As of 2019, Medicare pays a comparable amount for RVUs are lower. Thus, the shift of E&M office/outpatient E&M office/outpatient visits in freestanding physician visits from freestanding offices to HOPDs reduces volume offices and off-campus HOPDs; however, Medicare growth because the RVUs are lower for these services continues to pay a higher amount for these visits when when they are delivered in HOPDs. 8 provided in on-campus HOPDs. The Commission Medicare makes both a fee schedule payment and a also recommended adjusting OPPS rates for services in facility payment when a service is provided in an HOPD ambulatory payment classification (APC) groups that (the facility payment accounts for the cost of the service meet certain criteria so that payment rates are equal or in an HOPD). However, the program makes only a more closely aligned between HOPDs and freestanding 9 fee schedule payment when a service is furnished in offices (Medicare Payment Advisory Commission 2014). March 2019 | Report to the Congress: Medicare Payment Policy 111

142 Volume growth, which accounts for most of the difference FIGURE between the payment updates and spending growth, is FIGURE Growth in the volume of clinician 4–6 Volume growth has raised... 4-X influenced, among other things, by changes in clinical services caused fee schedule spending to increase faster than input practice, such as the diffusion of new technologies. It may prices and payment updates, 2000–2017 also be related to an increase in the treated prevalence (the share of the population receiving treatment) of many 80 Updates chronic conditions. For example, rapid growth in the Spending per beneficiary 70 proportion of beneficiaries treated for five or more chronic MEI MEI Payment updates conditions between 1987 and 2002 fueled an increase in 60 Medicare spending during this period (Thorpe and Howard Spending per FFS beneciary 2006). Reasons for growth in the treated prevalence 50 of chronic conditions included higher rates of obesity, 40 advances in technology for diagnosing and treating conditions, and changes in the definition of some diseases 30 (Medicare Payment Advisory Commission 2007). Volume growth could also reflect changes in the demographic 20 Cumulative percent change status of beneficiaries, although the effect of changes in 10 age and sex on Medicare spending for physician and other health professional services has generally been small in the 0 recent past, and spending on physician services varies less 2014 2012 2010 2008 2006 2016 2004 2002 2000 by age than spending for other services, such as inpatient hospital and post-acute care. MEI (Medicare Economic Index). The MEI measures the change in Note: clinician input prices. Spending per beneficiary includes only services paid under the fee schedule for physicians and other health professionals In 2017, per beneficiary spending for fee schedule services and excludes services paid under the clinical laboratory fee schedule. 10 increased slightly, by 0.8 percent. Several factors 2018 annual report of the Boards of Trustees of the Medicare trust funds; Source: influenced this increase: the small increase in volume Centers for Medicare & Medicaid Services 2017; Clemens 2014. (1.6 percent), the small increase in the fee schedule conversion factor (0.5 percent), a larger penalty for Note and Source in InDesign. Note: clinicians who did not meet the electronic health record (EHR) meaningful use requirement, and smaller incentive payments for clinicians who met the EHR meaningful use 11 APCs that meet these criteria are those that are unlikely requirement. to have costs associated with operating an emergency department; do not have extra costs associated with higher Quality of care Notes about this graph: patient complexity in HOPDs; and include services that For the past decade, CMS has assessed the quality of • Data is in the datasheet. Make updates in the datasheet. are frequently performed in physicians’ offices (which Medicare-billing physicians and other health professionals indicates that these services are likely safe and appropriate • I had to force return the items on the x-axis. They will reflow if I update the data. based largely on clinician-reported individual quality to provide in a physician’s office). measures and clinician attestation of participation in • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. certain activities. In 2019, CMS is implementing the Volume growth has contributed to an increase in • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph Merit-based Incentive Payment System (MIPS), which spending, 2000 to 2017 default when you change the data. entails clinician-level payment adjustments based on these The growth in service volume has contributed significantly • Use paragraph styles (and object styles) to format. clinician-reported and -attested quality measures and to an increase in spending for fee schedule services participation activities (see text box for first year results, (Figure 4-6). From 2000 to 2017, payment updates for pp. 114–115). these services did not keep pace with growth in input prices. Payment updates increased cumulatively by 11 The Commission has established a set of principles percent—less than the 33 percent cumulative increase in for quality measurement in Medicare; we believe that the Medicare Economic Index (MEI), which measures the MIPS measures are neither effective in assessing changes in input prices. However, spending per beneficiary true clinician quality nor appropriate for Medicare’s for these services grew at a cumulative rate of 67 percent. value-based purchasing programs. Specifically, quality Physician and other health professional services: Assessing payment adequacy and updating payments 112

143 TABLE ® performance rates, 2013–2017 Medicare FFS CAHPS 4–8 CAHPS composite measure 2014 2015 2016 2017 2013 % 87 % 86 % 85 % 84 % 84 Getting needed care and seeing specialists 77 75 76 75 77 Getting appointments and care quickly Care coordination (e.g., personal doctor always or usually discusses medication, has relevant medical records, helps with managing care) 86 86 85 86 86 85 Rating of health plan (FFS Medicare) 84 82 84 83 Rating of health care quality 86 86 86 85 85 ® ® ). Questions in rows 1 to 3 have responses of “Never,” “Sometimes,” Note: FFS (fee-for-service), CAHPS (Consumer Assessment of Healthcare Providers and Systems “Usually,” and “Always.” CMS converts these to a linear mean score on a 0 to 100 scale. Questions in rows 4 and 5 have responses of 1 to 10 (which CMS converts to a linear mean score on a 0 to 100 scale). “Plan” in the fourth row refers to the Medicare FFS program. FFS CAHPS mean scores provided by CMS. Source: measurement should be patient oriented, encourage Avoidable hospitalizations coordination across providers and time, and promote To assess rates of avoidable hospitalizations for change in the delivery system. Medicare quality programs ambulatory care–sensitive conditions, we use the should include population-based measures such as Prevention Quality Indicators (PQIs), a set of population- outcomes, patient experience, and value. Along these based measures of potentially avoidable hospital lines, this chapter reports three measures assessing admissions developed by the Agency for Healthcare the ambulatory care environment for Medicare FFS Research and Quality. The PQIs, which are based on beneficiaries: patient experience (measured using the national data, can help gauge the quality of a community’s ® Consumer Assessment of Health Providers and Systems ambulatory care environment. Lower rates can be one ® (CAHPS )), population-based measures assessing indication of higher quality. However, this measure is also avoidable hospitalizations for ambulatory care–sensitive sensitive to secular trends over time in the site of care. conditions, and rates of low-value care in Medicare. Figure 4-8 (p. 116) presents results for three common Patient experience measures conditions among the Medicare population—diabetes, congestive heart failure, and bacterial pneumonia. The CAHPS surveys are a suite of surveys that assess Consistent with prior years, the rates show general patient experience and reported access. CAHPS results declines across all three conditions and the age categories, for Medicare Advantage plans are used in the Part C and likely due to continuing declines in inpatient admissions. Part D star ratings that are intended to measure quality in The modest increase for heart failure may be the result of the Medicare Advantage program, and a CAHPS survey CMS dramatically reducing its frequency of challenges to module is issued to a sample of beneficiaries in the FFS the medical necessity of short-stay cases. Medicare population. The Commission plans to continue to refine a set of Overall, how Medicare FFS beneficiaries rated their health population-based outcome measures that Medicare can care quality and reported their ability to get care quickly calculate using claims data. was generally stable between 2013 and 2017, although there was a slight decline in reporting that they could get Low-value care needed care and see specialists (Table 4-8). There was Low-value care is the provision of a service that has little also a slight decline in beneficiaries’ ratings of their health or no clinical benefit or care in which the risk of harm plans and health care quality. March 2019 | Report to the Congress: Medicare Payment Policy 113

144 The Merit-based Incentive Payment System year 1 results • Permit clinicians to meet the 3-point MIPS s of 2019, the Merit-based Incentive Payment performance threshold by reporting minimal System (MIPS) adjusts payments in fee-for- information on one quality measure (or attesting to service (FFS) Medicare at the individual A one performance activity). clinician level based on performance in four areas: quality; resource use; clinical practice improvement Weight the cost component at 0 points. • activities; and advancing care information (formerly meaningful use of electronic health records) (Centers Because the basic MIPS payment adjustments must for Medicare & Medicaid Services 2016). be weighted to be budget neutral, the decision to set the performance threshold very low means that the The payment changes that take place in 2019 are based payment increases will be very small (because there on clinician performance in 2017. On November 8, will be many clinicians meeting or exceeding the 2018, CMS released the initial summary of MIPS thresholds). The exceptional performance bonus is not performance data that will underlie the payment budget neutral and will linearly increase for clinicians adjustments in 2019. at a certain threshold above the MIPS threshold. For the first year of the program, CMS made a number In the first year of the program, just over 1 million of discrete policy decisions to reflect a phased approach MIPS-eligible clinicians (including physicians, nurse to implementation, which CMS refers to as “Pick Your practitioners, physician assistants, group practices, Pace.” Specifically, CMS used its regulatory authority to: and certain other nonphysician practitioners) reported MIPS information (Table 4-9). Most clinicians— • Set the MIPS performance threshold at 3 points over 700,000—reported sufficient information with (out of 100). Clinicians with a score above 3 are to sufficiently high scores to receive both a positive MIPS receive a neutral or positive payment adjustment, adjustment and qualify for the MIPS exceptional and clinicians with a score of 3 or below are to performance bonus. receive a negative payment adjustment. Figure 4-7 illustrates how the MIPS incentive • Set the MIPS exceptional performance bonus payment works for the first year. In concept, the threshold at 70 points (out of 100). TABLE MIPS performance information for eligible clinicians, 2017 4–9 Number of clinicians Payment adjustment –4 % 51,500 Did not report Reported Minimum required 20,100 0% Sufficient data to gain a positive update 221,400 Between 0% and 0.22% Sufficient data to gain a positive update and exceptional performance bonus 714,500 Between 0.28% and 1.88% Note: MIPS (Merit-based Incentive Payment System). This table includes all clinicians who reported MIPS information, even those who may qualify as “low volume” for MIPS purposes or are excluded from Table 4-6 (p. 108). CMS. http://qpp.cms.gov. Source: (continued next page) Physician and other health professional services: Assessing payment adequacy and updating payments 114

145 (cont.) The Merit-based Incentive Payment System year 1 results adjustment and the 70-point threshold for the negative payment adjustment applied to clinicians exceptional performance bonus. below the MIPS performance threshold must fund the bonuses applied to clinicians above the MIPS CMS is moving toward meeting an eventual statutory performance threshold. Under these circumstances, deadline in 2022 to set the MIPS performance score performance bonuses this year were predictably small: at the mean or median of clinician performance, The maximum MIPS bonus was 0.22 percent. When which will compress the range of positive payment the exceptional performance bonus was added, the adjustments such that small changes in MIPS maximum total bonus was 1.88 percent. performance scores will result in large swings in payment adjustments. In other words, because most Despite the low performance threshold, because clinicians have sufficiently high scores in the first year clinicians could choose which measures to report, of the program (with 71 percent qualifying for both most clinicians had very high performance scores a positive payment adjustment and the exceptional overall in the first year of the program. Specifically, performance bonus), the mean or median MIPS the mean performance score was 74 points, and performance scores will be very high. the median performance score was 89 points, well ■ in excess of the 3-point threshold for a positive FIGURE Freestanding Medicare margins... 4-X FIGURE The Merit-based Incentive Payment System adjustments, 2017 4–7 3 2 24% of clinicians qualified for a positive MIPS adjustment (but not an exceptional performance bonus) 1 0 71% of clinicians qualified for –1 a positive MIPS adjustment plus exceptional performance bonus –2 Adjustment (in percent) –3 5% of clinicians did not report any performance information and –4 so received the maximum penalty of –4% –5 3 6 0 75 9 12 15 18 99 96 93 90 87 84 81 78 21 72 69 66 63 60 57 54 51 48 45 42 39 36 33 30 27 24 MIPS performance score (0 to 100) Note: MIPS (Merit-based Incentive Payment System). Source: MedPAC analysis based on data from CMS. Note: Note and Source are in InDesign. Source: | March 2019 Report to the Congress: Medicare Payment Policy 115 Notes about this graph: • Data is in the datasheet. Make updates in the datasheet. • WATCH FOR GLITCHY RESETS WHEN YOU UPDATE DATA!!!! • The column totals were added manually. • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. • I can’t delete the legend, so I’ll just have to crop it out in InDesign. • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph default when you change the data. • Use paragraph styles (and object styles) to format. • Data was from: R:\Groups\MGA\data book 2007\data book 2007 chp1

146 FIGURE FIGURE Cumulative change... Trends in selected PQIs for inpatient admissions of FFS beneficiaries 4–8 X-X for ambulatory care–sensitive conditions, 2007–2016 3000 2007 2015 2016 2014 2013 2012 2011 2010 2009 2008 2500 2000 1500 1000 per 100,000 beneficiaries 500 Inpatient admissions of FFS beneficiaries 0 Under Ages Ages Under Ages Ages Ages Under Ages 75 and older 65 to 74 65 to 74 age 65 75 and older age 65 75 and older 65 to 74 age 65 PQI 3: PQI 11: PQI 8: Bacterial pneumonia Diabetes long-term complications Congestive heart failure PQI (Prevention Quality Indicator), FFS (fee-for-service). Figures represent the number of hospital admissions for the identified condition for Medicare beneficiaries in Note: each age range per 100,000 beneficiaries. Only FFS beneficiaries with both Part A and Part B are included. Beneficiaries who died during the year are included. Source: CMS data on geographic variation. Figures calculated by CMS from the Chronic Conditions Data Warehouse of 100 percent of claims. from the service outweighs its potential benefit (Chan percent and 37 percent of beneficiaries received at least et al. 2013, Kale et al. 2013). In addition to increasing one low-value service, and annual Medicare spending for health care spending, low-value care has the potential these services ranged from $2.4 billion to $6.5 billion. The to harm patients by exposing them to the risks of injury spending estimates are conservative because they do not from inappropriate tests or procedures and may lead to reflect the downstream cost of low-value services (e.g., a cascade of additional services that contain risks but follow-up tests and procedures). For more information on provide little or no benefit (Keyhani et al. 2013, Korenstein this analysis, see the Commission’s June 2018 report to et al. 2012). Because the current MIPS measure set has the Congress (Medicare Payment Advisory Commission few measures assessing low-value care and few clinicians 2018). report these measures, the Commission previously used a Medicare payments and providers’ costs set of 31 claims-based measures to assess low-value care in Medicare in 2014. Our analysis demonstrated that low- Because physicians and other health professionals do not value care was a significant issue in Medicare that year: report their costs to the Medicare program, we use other We found between 34 and 72 instances of low-value care measures to assess the adequacy of Medicare payments per 100 beneficiaries, depending on whether we used a relative to clinicians’ costs. The first measure is how Note: Note and Source are in InDesign. narrow or broad version of each measure. Between 23 Medicare’s payments compare with the commercial rates Source: Physician and other health professional services: Assessing payment adequacy and updating payments 116

147 paid by preferred provider organizations (PPOs). The By contrast, the average commercial price received by the second measure compares physician compensation across smallest independent practices for an E&M visit was about specialties and evaluates whether Medicare’s fee schedule equal to Medicare’s rate. These findings indicate that the contributes to an income disparity between primary care ratio of Medicare rates to commercial rates for physician clinicians and other specialties. The third measure assesses services varies by practice size within the same market the change in input prices for physician and other health because larger practices can obtain higher prices from professional services—the MEI. commercial payers than smaller practices. In addition to varying within markets, evidence suggests that commercial Ratio of Medicare payments to commercial PPO prices for physician services vary widely across markets. payments A study by the Congressional Budget Office found that the average ratio of commercial prices to Medicare prices In 2017, Medicare’s payment rates for physician and for 20 common services was at least 70 percent higher other health professional services (including cost sharing) in the most costly market than in the least costly market were 75 percent of commercial rates paid by PPOs, (Congressional Budget Office 2018). unchanged from 2016. The ratio has declined from 81 percent in 2010. The ratio in 2017 varied by type of Compensation is much higher for certain service. For example, Medicare rates were 80 percent of specialties than for primary care commercial rates for E&M office visits for established patients but 59 percent of commercial rates for coronary The Commission remains concerned that ambulatory artery bypass graft surgery. This analysis uses data on paid E&M visits, which make up a large share of the claims for PPO members of a large national insurer that services provided by primary care clinicians and certain covers a wide geographic area across the United States. other specialties (e.g., psychiatry, endocrinology, and The payments reflect the insurer’s allowed amount with rheumatology), are underpriced in the fee schedule relative allowed cost sharing. The data exclude any remaining to other services, such as procedures (Medicare Payment 12 balance billing and payments made outside of the claims Advisory Commission 2018). This factor contributes to process, such as bonuses or risk-sharing payments. an income disparity between primary care physicians and certain specialists. The ratio of Medicare rates to commercial rates has declined in recent years as commercial rates have risen For an analysis of the compensation received from all while Medicare rates have remained relatively stable. The payers by physicians—the largest subset of practitioners— growth of commercial prices could be a consequence of the Commission contracted with the Urban Institute, greater consolidation of physician practices, which gives working in collaboration with SullivanCotter. The physicians greater leverage to negotiate higher prices with contractor calculated median compensation based on commercial plans. In recent years, an increasing number 2017 data from SullivanCotter’s Physician Compensation of physicians have joined larger groups, hospitals, and and Productivity Survey. Median compensation across health systems. For example, between 2009 and 2014, all specialties was $300,000 in 2017. Compensation the share of physicians working in practices with more was much higher for some specialties than others. The than 50 physicians grew from 16 percent to 22 percent specialty groups with the highest median compensation (Medicare Payment Advisory Commission 2017). Recent were radiology ($460,000); the nonsurgical, procedural studies show that commercial prices for physician services group ($426,000); and surgical specialties ($420,000) 13 are higher in markets with larger physician practices and (Figure 4-9, p. 118). Median compensation for radiology in markets with greater physician–hospital consolidation was 90 percent higher than median compensation for (Baker et al. 2014, Clemens and Gottlieb 2017, Neprash primary care ($242,000), and median compensation et al. 2015). Our own research found that independent for nonsurgical, procedural specialties was 76 percent practices with larger market shares and hospital-owned higher than that of primary care. Psychiatry—which is practices received higher commercial prices for E&M in the nonsurgical, nonprocedural group—had median visits than other practices in their market (Medicare compensation of $241,000, slightly lower than primary 14 Payment Advisory Commission 2017). For example, care physicians (data not shown). Previous Commission independent practices with a large market share of E&M work using data from the Medical Group Management visits received an average commercial price for an E&M Association (MGMA) showed that such disparities also visit that was 41 percent higher than the Medicare rate. existed when compensation was observed on an hourly March 2019 | Report to the Congress: Medicare Payment Policy 117

148 FIGURE Freestanding Medicare margins... FIGURE 4-X Disparities in physician compensation are widest when primary care physicians 4–9 are compared with radiologists, nonsurgical proceduralists, and surgeons, 2017 500 460 426 420 400 300 283 300 242 200 (in thousands of dollars) Median annual compensation 100 0 Primary care Radiology All Nonsurgical, Surgical Nonsurgical, procedural nonprocedural Figure includes all physicians who reported their annual compensation in the survey (76,336). Note: Urban Institute 2018. Source: Note: Note and Source are in InDesign. Source: basis, thus accounting for variations in hours worked number of work RVUs per physician generated by primary 15 per week. care (4,833) (Table 4-10). Median compensation for From 2013 to 2017, median compensation surgical specialties was 76 percent higher than median for primary care physicians and surgeons increased at a compensation for primary care, and their median number cumulative rate of 15.4 percent, slower than nonsurgical, Notes about this graph: of work RVUs was 46 percent higher than primary care. procedural specialties (17.9 percent) and nonsurgical, • Data is in the datasheet. Make updates in the datasheet. Because primary care physicians are more likely to focus nonprocedural specialties (16.2 percent) but faster than 16 • WATCH FOR GLITCHY RESETS WHEN YOU UPDATE DATA!!!! on ambulatory E&M services than the other specialty radiology (9.6 percent) (data not shown). Across all groups and because these services tend to have lower work specialty groups, median compensation grew 15.9 percent • The column totals were added manually. RVUs than other services, the fee schedule’s RVUs for during this period. • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. ambulatory E&M services may be an important source of Three of the four specialty groups with higher annual • I can’t delete the legend, so I’ll just have to crop it out in InDesign. the disparities in compensation between primary care and compensation than primary care also generated more other specialty groups. • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph 17 work RVUs per year. For example, in 2017, median default when you change the data. The fee schedule’s work RVUs, which account for the compensation for radiology was nearly double the median amount of work required to provide a service, are based compensation for primary care, and radiology had the • Use paragraph styles (and object styles) to format. on an assessment of how much time and intensity (e.g., highest median number of cumulative work RVUs per • Data was from: R:\Groups\MGA\data book 2007\data book 2007 chp1 mental effort and technical skill) services require relative physician (8,862)—83 percent higher than the median Physician and other health professional services: Assessing payment adequacy and updating payments 118

149 TABLE Most specialty groups with higher median annual compensation than primary care 4–10 generate a higher median number of work RVUs than primary care, 2017 Ratio of median annual compensation Median number of annual work RVUs for specialty group to median Specialty group compensation for primary care per physician Radiology 1.99 8,862 7,070 Surgical 1.76 Nonsurgical, procedural 1.80 6,395 Primary care 1.00 4,833 Nonsurgical, nonprocedural 4,554 1.19 Note: RVU (relative value unit). The table includes only physicians who reported both their annual compensation and their annual number of work RVUs in the survey (44,605). Source: Urban Institute 2018. to one another. If estimates of time and intensity are not Input costs for physicians and other health professionals are projected to increase from 2019 kept up to date, especially for services that experience to 2020 efficiency improvements, the work RVUs become inaccurate. Because of advances in technology, technique, The MEI measures the change in the market basket of and clinical practice, efficiency improves more easily for input prices for physician and other health professional 18 procedures, imaging, and tests than for ambulatory E&M services and is adjusted for economy-wide productivity. services, which are composed largely of activities that As of the third quarter of 2018, CMS’s forecast is that the require the clinician’s time and so do not lend themselves MEI will increase by 2.4 percent in 2020. This projection to efficiency gains. When efficiency gains reduce the is subject to change. amount of work needed for a service, the work RVUs for the affected services should decline accordingly. Under the budget-neutral fee schedule, a reduction in the RVUs How should Medicare payments change of these services would raise the RVUs for all other in 2020? services, such as ambulatory E&M services. Because of problems with the process of reviewing overpriced The Commission’s deliberations on payment adequacy services and the data used to set prices, such as the lack for physician and other health professional services of current and objective data on clinician work time and are informed by beneficiary access to services, volume practice expenses, this two-step sequence tends not to growth, quality, and input prices for clinician services. We occur (Medicare Payment Advisory Commission 2018). find that, on the basis of these indicators, payments appear Therefore, ambulatory E&M services become passively adequate. devalued over time. On measures of access to the services of physicians and The Commission is concerned that this mispricing could other health professionals, the Commission continues to lead to problems with beneficiary access to E&M services find that beneficiaries’ access to care appears generally and, over the longer term, could even influence the pipeline stable. Overall, Medicare beneficiaries generally have of physicians in specialties that tend to provide a large comparable or slightly better access to clinician services share of these services. The Commission has made previous than privately insured individuals ages 50 to 64. A slight recommendations to improve the accuracy of the data decline in the number of physicians per beneficiary was used to set RVUs and to rebalance the fee schedule toward offset by an increase in the number of advanced practice primary care by establishing a per beneficiary payment for primary care practitioners (see text box, pp. 120–121). March 2019 | Report to the Congress: Medicare Payment Policy 119

150 Previous Commission recommendations to improve the accuracy of data for setting relative value units and establish a per beneficiary payment for primary care clinicians should be used to calculate the amount of time that he Commission has a long-standing concern that a clinician worked over the course of a week or ambulatory evaluation and management (E&M) month and compare it with the time estimates in the services, which make up a large share of the T fee schedule for all of the services that the clinician services provided by primary care clinicians and certain billed over the same period. If the fee schedule’s time other specialties (e.g., psychiatry, endocrinology, and estimates exceed the actual time worked, this finding rheumatology), are underpriced in the fee schedule could indicate that the time estimates—and, hence, for physician and other health professional services the work RVUs—are too high. CMS could use this compared with other services, such as procedures approach to identify groups of services that are likely (Medicare Payment Advisory Commission 2018). overpriced, carefully review those services, and adjust Validation of the relative value units (RVUs) in the the work RVUs accordingly. fee schedule could help correct the fee schedule’s inaccuracies and ensure that ambulatory E&M visits— Practice expense RVUs—which account for the cost of such as office visits, hospital outpatient department operating a practice—are based on data from a survey visits, nursing facility visits, and home visits—are not of total practice costs incurred by nearly all specialty underpriced. Addressing this mispricing could also groups. Because this survey was conducted in 2007 reduce disparities in compensation among specialties. and 2008, practice expense RVUs are not likely to reflect current practice costs. CMS has not developed In 2011, the Commission recommended that CMS use a strategy for updating practice cost data. However, a streamlined method to regularly collect data from CMS could regularly collect data on total practice a cohort of efficient clinician practices—including costs along with data on service volume and work time service volume and work time—to establish more from a cohort of efficient practices, as the Commission accurate work and practice expense RVUs (Medicare recommended in 2011 (Medicare Payment Advisory Payment Advisory Commission 2011a, Medicare Commission 2011a). Payment Advisory Commission 2011b). These data (continued next page) registered nurses and physician assistants per beneficiary, Update recommendation and the share of providers enrolled in Medicare’s In recommending an update for physicians and other participating provider program remains high. health professionals, the Commission balanced the following objectives: In 2017, across all services, volume per beneficiary grew 1.6 percent. Among broad service categories, growth • maintaining beneficiary access to physician and other rates were 1.2 percent for E&M, 1.3 percent for imaging health professional services; services, 2.1 percent for major procedures, 2.1 percent for other procedures, and 2.4 percent for tests. • minimizing the burden on taxpayers and beneficiaries, who finance the Medicare program; and As of the third quarter of 2018, input prices for physicians and other health professionals were projected to increase • ensuring adequate payments for the efficient provision by 2.4 percent in 2020. In 2017, compensation was much of services. lower for primary care physicians than for physicians In balancing these objectives with the overall findings that in certain specialties, which raises concerns about fee payments appear adequate, the Commission recommends schedule mispricing and its impact on primary care. an update for 2020 consistent with current law. Physician and other health professional services: Assessing payment adequacy and updating payments 120

151 Previous Commission recommendations to improve the accuracy of data for setting relative value units and establish a per beneficiary payment for primary care clinicians (cont.) This method of funding would be budget neutral and In addition to concern about the mispricing of would help rebalance the fee schedule toward primary ambulatory E&M services, the Commission is also care clinicians. concerned that the fee schedule—with its orientation toward discrete services that have a definite beginning In the Commission’s June 2018 report, we described and end—is not well designed to support primary care, another budget-neutral approach to rebalance the which requires ongoing care coordination for a panel fee schedule that would increase payment rates for of patients. Consequently, in 2015, the Commission ambulatory E&M services while reducing payment recommended that the Congress establish a per rates for other services (e.g., procedures, imaging, and beneficiary payment for primary care clinicians to tests) (Medicare Payment Advisory Commission 2018). replace the expired Primary Care Incentive Payment Under this approach, the increased payment rates would (PCIP) program, which provided a 10 percent bonus apply to ambulatory E&M services provided by all payment on fee schedule payments for certain E&M clinicians, regardless of specialty, and would not change visits provided by primary care clinicians (Medicare the current fee-for-service system. This change would Payment Advisory Commission 2015). A monthly be a one-time price adjustment to the fee schedule per beneficiary payment based on the total amount to address several years of passive devaluation of of PCIP payments in 2015 would initially amount to ambulatory E&M services. We modeled the impact of a about $2.35. The Commission recommended that the 10 percent payment rate increase for ambulatory E&M additional payments to primary care clinicians be in the services, although a higher or lower increase could be form of a per beneficiary payment to move away from considered. A 10 percent increase would raise annual the approach of paying separately for each discrete spending for ambulatory E&M services by $2.4 billion. service. The payment would provide funds to support To maintain budget neutrality, payment rates for all other the investment in infrastructure and staff that facilitate fee schedule services would be reduced by 3.8 percent. care management and care coordination. Funding for These payment changes could be implemented in one the per beneficiary payment would come from reducing year or phased in gradually over multiple years. ■ payment rates for all services in the fee schedule other than ambulatory E&M visits provided by any clinician. RECOMMENDATION 4 payment adequacy are stable and consistent with prior years. Therefore, the Commission does not see a reason to For calendar year 2020, the Congress should increase the diverge from the current-law policy of no update for 2020. calendar year 2019 Medicare payment rates for physician and other health professional services by the amount IMPLICATIONS 4 specified in current law. Spending RATIONALE 4 • No change as compared with current law. The Medicare Access and CHIP Reauthorization Act of Beneficiary and provider 2015 established a set of statutory updates for clinicians, • The Commission’s recommendation of the current-law including no statutory update for calendar year 2020. update should not affect beneficiaries’ access to care or Overall, access to clinician services for Medicare providers’ willingness and ability to furnish care. ■ beneficiaries appears stable and comparable with that for privately insured individuals. Other measures of March 2019 | Report to the Congress: Medicare Payment Policy 121

152 Endnotes 9 For the OPPS, CMS classifies services into APC groups on 1 For further information, see the Commission’s Payment the basis of clinical and cost similarity; all services within an Basics: Physician and Other Health Professionals Payment APC group have the same payment rate. System at http://medpac.gov/docs/default-source/payment- basics/medpac_payment_basics_18_physician_final_sec. 10 This figure is based on incurred spending, rather than cash pdf?sfvrsn=0. spending, for fee schedule services. Cash spending for fee schedule services declined slightly between 2016 and 2017 This year’s survey results continue to be largely consistent 2 because of a lag between incurred and cash spending. with other surveys of Medicare beneficiaries and privately insured individuals. 11 Between 2016 and 2017, the penalty for clinicians who did not meet the EHR meaningful use requirement grew from 2 refers to those not of 3 White In this section, the category percent of payments to 3 percent of payments, and the total Hispanic origin. See the U.S. Census Bureau’s “Explanation amount of incentive payments for clinicians who met the EHR of Race and Hispanic Origin Categories” at https://www. meaningful use requirement dropped from $932 million to census.gov/population/estimates/rho.txt. $437 million. The penalties and incentive payments under the 4 Primary care physicians include specialties that were eligible EHR program are mandated by statute. for the Primary Care Incentive Payment Program: family Ambulatory E&M services include visits in offices, hospital 12 medicine, internal medicine, pediatric medicine, and geriatric outpatient departments, certain other settings such as nursing medicine. In 2017, CMS introduced a new physician specialty facilities, and patients’ homes. code for hospitalists. Most of the physicians who billed Medicare as hospitalists in 2017 billed as a primary care The nonsurgical, procedural specialties in the analysis are 13 specialty in 2016. Therefore, to maintain consistency across cardiology, dermatology, gastroenterology, pulmonary years, we assigned physicians who billed as hospitalists in medicine, and hematology/oncology. 2017 to the primary care physicians group for this analysis. In addition to psychiatry, the nonsurgical, nonprocedural 14 5 The number of beneficiaries used to calculate the ratio group includes emergency medicine, endocrinology, hospital of physicians and other health professionals per 1,000 medicine, nephrology, neurology, physical medicine, beneficiaries includes those in FFS Medicare and Medicare rheumatology, and other internal medicine/pediatrics. The Advantage because we assume that clinicians are furnishing primary care specialties in the analysis are family medicine, services to beneficiaries covered under either program. internal medicine, and general pediatrics. Services that are less likely to be assigned include osteopath 6 15 To account for differences among specialties in hours worked services and chiropractor services (although the assignment per week, an earlier analysis based on MGMA data from rates are still about 90 percent for both service types). 2007 included comparisons of hourly compensation. Hourly compensation for nonsurgical, procedural specialties and 7 When this type of visit is provided in an HOPD, it is billed as radiology was more than double the hourly compensation rate Healthcare Common Procedure Coding System code G0463. for primary care. 8 Section 603 of the Bipartisan Budget Act of 2015 prohibits To control for annual changes in survey respondents, the 16 HOPDs that began billing under the OPPS on or after percentage changes are based on a cohort analysis in which November 2, 2015, and are located off a hospital campus the sample was restricted to physicians who were present in from billing under the OPPS after January 1, 2017. In 2018, both the 2013 and 2017 data. the facility payment rate for services provided at these off- campus HOPDs was equal to 40 percent of the rate under the 17 The exception was nonsurgical, nonprocedural specialties, OPPS. On-campus HOPDs; off-campus HOPDs that began which had median annual compensation that was 19 percent billing before November 2, 2015; and dedicated emergency higher than primary care but generated 6 percent fewer work departments are permitted to continue billing under the OPPS. RVUs per year than primary care. However, as of 2019, Medicare pays all off-campus HOPDs (regardless of when they began billing under the OPPS) 18 The MEI measures the weighted average annual price an amount equal to 40 percent of the OPPS rate for office/ change for various inputs used by physicians and other health outpatient E&M visits. professionals to furnish services. Physician and other health professional services: Assessing payment adequacy and updating payments 122

153 References Korenstein, D., R. Falk, E. A. Howell, et al. 2012. Overuse ABIM Foundation. 2017. Choosing Wisely: Treating blocked of health care services in the United States: An understudied leg arteries. http://www.choosingwisely.org/wp-content/ 172, no. 2 (January 23): Archives of Internal Medicine problem. uploads/2018/02/Treating-Blocked-Leg-Arteries-SVM.pdf. 171–178. ABIM Foundation. 2016. Choosing Wisely: Facts and figures. Medicare Payment Advisory Commission. 2018. Report to http://www.choosingwisely.org/wp-content/uploads/2016/07/ the Congress: Medicare and the health care delivery system. Choosing-Wisely-Facts-and-Figures.pdf. Washington, DC: MedPAC. Baker, L. C., M. K. Bundorf, A. B. Royalty, et al. 2014. Physician Report to Medicare Payment Advisory Commission. 2017. practice competition and prices paid by private insurers for office the Congress: Medicare and the health care delivery system. 312, no. 16 visits. Journal of the American Medical Association Washington, DC: MedPAC. (October 22–29): 1653–1662. Medicare Payment Advisory Commission. 2015. Report to the Centers for Medicare & Medicaid Services, Department of Health Congress: Medicare payment policy. Washington, DC: MedPAC. and Human Services. 2017. Market basket data. https://www.cms. gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and- Report to the Medicare Payment Advisory Commission. 2014. Reports/MedicareProgramRatesStats/MarketBasketData.html. Congress: Medicare payment policy. Washington, DC: MedPAC. Centers for Medicare & Medicaid Services, Department of Health Medicare Payment Advisory Commission. 2013. Report to and Human Services. 2016. Medicare program; Merit-based . the Congress: Medicare and the health care delivery system Incentive Payment System (MIPS) and Alternative Payment Washington, DC: MedPAC. Model (APM) Incentive under the physician fee schedule, and criteria for physician-focused payment models. Final rule. Federal Report to the Medicare Payment Advisory Commission. 2012. 81, no. 214 (November 4): 77008–77831. Register Washington, DC: MedPAC. Congress: Medicare payment policy. Chan, K. S., E. Chang, N. Nassery, et al. 2013. The state of Medicare Payment Advisory Commission. 2011a. Moving overuse measurement: A critical review. Medical Care Research forward from the sustainable growth rate (SGR) system. Letter to 70, no. 5 (October): 473–496. and Review the Congress. October 14. Clemens, J., and J. Gottlieb. 2017. In the shadow of a giant: Report to Medicare Payment Advisory Commission. 2011b. Journal of Medicare’s influence on private physician payments. the Congress: Medicare and the health care delivery system. 125, no. 1 (February): 1–39. Political Economy Washington, DC: MedPAC. Clemens, K., Centers for Medicare & Medicaid Services, Report to the Medicare Payment Advisory Commission. 2007. Department of Health and Human Services. 2014. Estimated Washington, Congress: Promoting greater efficiency in Medicare. sustainable growth rate and conversion factor, for Medicare DC: MedPAC. payments to physicians in 2015. Fact sheet. https://www. cms.gov/medicare/medicare-fee-for-service-payment/ Neprash, H. T., M. E. Chernew, A. L. Hicks, et al. 2015. sustainablegratesconfact/downloads/sgr2015p.pdf. Association of financial integration between physicians and hospitals with commercial health care prices. JAMA Internal Congressional Budget Office. 2018. An analysis of private- 175, no. 12 (December): 1932–1939. Medicine . Working paper 2018–01. sector prices for physicians’ services Washington, DC: CBO. Thorpe, K. E., and D. H. Howard. 2006. The rise in spending among Medicare beneficiaries: The role of chronic disease Kale, M. S., T. F. Bishop, A. D. Federman, et al. 2013. Trends 25, Health Affairs prevalence and changes in treatment intensity. in the overuse of ambulatory health care services in the United no. 5 (September–October): w378–388. 173, no. 2 (January 28): 142– JAMA Internal Medicine States. 148. Urban Institute. 2018. Analysis of disparities in physician compensation. Report prepared for the Medicare Payment Keyhani, S., R. Falk, E. A. Howell, et al. 2013. Overuse and Advisory Commission. Washington, DC: MedPAC. Medical Care 51, no. 6 systems of care: A systematic review. (June): 503–508. Report to the Congress: Medicare Payment Policy | March 2019 123

154

155 CHAPTER 5 Ambulatory surgical center services

156 RECOMMENDATIONS The Congress should eliminate the calendar year 2020 update to the Medicare conversion 5-1 factor for ambulatory surgical centers. COMMISSIONER VOTES: YES 16 • NO 0 • NOT VOTING 0 • ABSENT 1 ... 5-2 The Secretary should require ambulatory surgical centers to report cost data. COMMISSIONER VOTES: YES 16 • NO 0 • NOT VOTING 0 • ABSENT 1

157 CHAPTER 5 Ambulatory surgical center services Chapter summary In this chapter Ambulatory surgical centers (ASCs) provide outpatient procedures to patients • Are Medicare payments who do not require an overnight stay after the procedure. In 2017, 3.4 million adequate in 2019? fee-for-service (FFS) Medicare beneficiaries were treated in the 5,603 ASCs How should Medicare • certified to provide services to Medicare beneficiaries. Medicare program and payments change in 2020? beneficiary spending on ASC services was about $4.6 billion. Assessment of payment adequacy Our results indicate that beneficiaries’ access to ASC services is adequate. Most of the available indicators of payment adequacy for ASC services, discussed below, are positive. Our analysis of facility supply and volume Beneficiaries’ access to care— of services indicates that beneficiaries’ access to ASC services has generally been adequate. From 2012 to 2016, the number of • Capacity and supply of providers— ASCs increased by an average annual rate of 1.0 percent. In 2017, the number of ASCs increased 2.4 percent. Most new ASCs in 2017 (about 94 percent) were for-profit facilities. • From 2012 through 2016, the volume of services Volume of services— per FFS beneficiary increased by an average annual rate of 1.2 percent. In 2017, volume increased by 1.7 percent. March 2019 | Report to the Congress: Medicare Payment Policy 127

158 Quality of care— The first four years of ASC-reported quality data show improvement in performance, but the measures used within the ASC Quality Reporting (ASCQR) Program will change substantially in the next few years. Among the 11 quality measures for which data were available through 2016, performance among the ASCs that reported data improved for most measures. CMS will be making several changes to the ASCQR Program for 2019 and beyond. While the Commission concurs with CMS’s decision to eliminate process measures and measures of limited utility, we remain concerned about the delayed use of Consumer ® Assessment of Healthcare Providers and Systems measures and the lack of claims- based outcome measures that apply to all ASCs. For example, CMS could add measures targeting the frequency of ASC patients receiving subsequent hospital care or rates of surgical site infection. Because the number of ASCs has continued to Providers’ access to capital— increase and hospital systems and others have significantly incorporated ASCs into their business strategies, access to capital appears to be adequate. From 2012 to 2016, Medicare payments Medicare payments and providers’ costs— for ASC services per FFS beneficiary increased by an average annual rate of 3.5 percent. By contrast, in 2017, payments for ASC services increased by 7.7 percent. ASCs do not submit data on the cost of services they provide to Medicare beneficiaries. Therefore, we cannot calculate a Medicare margin as we do for other provider types to help assess payment adequacy. On the basis of these indicators, the Commission concludes that ASCs can continue to provide Medicare beneficiaries with access to ASC services with no update to the payment rates for 2020. In addition, the Commission continues to recommend that the Secretary of Health and Human Services collect cost data from ASCs without further delay. ■ Ambulatory surgical center services: Assessing payment adequacy and updating payments 128

159 indicates a procedure’s resource intensity relative to Background other procedures, is based on its relative weight under the OPPS. Although the ASC payment system is linked An ambulatory surgical center (ASC) is a distinct entity to the OPPS, payment rates for all services covered that primarily provides outpatient surgical procedures under both systems are lower in ASCs for two reasons. to patients who do not require an overnight stay after First, relative weights are lower under the ASC system the procedure. In addition to ASCs, hospital outpatient compared with the OPPS relative weights because CMS departments (HOPDs) and, in some cases, physicians’ makes proportional adjustments to the relative weights of offices perform outpatient surgical procedures. the OPPS to maintain budget neutrality in the ASC system. In 2019, this adjustment results in ASC relative weights Since 1982, Medicare has covered and paid for surgical that are 12.0 percent lower than the relative weights in the procedures provided in ASCs. Medicare covers surgical OPPS. Second, for most procedures covered under the procedures represented in about 3,500 Healthcare ASC system, the payment rate is the product of its relative Common Procedure Coding System (HCPCS) codes weight and an ASC conversion factor, set at $46.53 for under the ASC payment system. However, ASC volume 2019, which is lower than the OPPS conversion factor set for services covered under Medicare is concentrated in a at $79.49 for 2019. relatively small number of HCPCS codes. For example, in 2017, 28 HCPCS codes accounted for 75 percent of the The ASC conversion factor is lower than the OPPS ASC volume for surgical services provided to Medicare conversion factor because it started at a lower level in beneficiaries. For procedures performed in an ASC, 2008 and has been updated since then at a lower rate than Medicare makes two payments: one to the facility through the OPPS conversion factor. CMS set the initial ASC the ASC payment system and the other to the physician conversion factor in 2008 such that total payments to for his or her professional services through the payment ASCs under the revised payment system would equal what system for physicians and other health professionals, also they would have been under the pre-2008 ASC payment known as the physician fee schedule (PFS). According to system. In addition, from 2010 through 2018, CMS surveys, most ASCs have partial or complete physician updated the ASC conversion factor based on the consumer ownership (Ambulatory Surgery Center Association price index for all urban consumers (CPI–U), while it 2017, Medical Group Management Association 2009). used the hospital market basket (MB) index to update the Physicians who perform surgeries in ASCs they own OPPS conversion factor. The CPI–U has generally been receive a share of the ASC’s facility payment in addition lower than the hospital MB index. Therefore, the ASC to payment for their professional services. To receive conversion factor has been updated by smaller percentages payments from Medicare, ASCs must meet Medicare’s than the OPPS conversion factor. conditions of coverage, which specify standards for In a change of regulatory policy, CMS has decided to administration of anesthesia, quality evaluation, operating update the ASC conversion factor using the hospital MB and recovery rooms, medical staff, nursing services, and index from 2019 through 2023. Under this change, in other aspects of care. 2019 the update to the ASC conversion factor is higher Medicare pays ASCs for a bundle of facility services and than the update to the OPPS conversion factor because items—such as nursing, recovery care, anesthetics, and the update to the ASC conversion factor is the hospital supplies—through a system that is linked primarily to the MB index minus a multifactor productivity adjustment, outpatient prospective payment system (OPPS), which while the OPPS conversion factor is the hospital MB Medicare uses to set payment rates for most services index minus a multifactor productivity adjustment minus provided in HOPDs. The ASC payment system is also a statutory adjustment of 0.75 percentage points from the partly linked to the PFS. A more detailed description of Patient Protection and Affordable Care Act of 2010. From the ASC payment system can be found online at http:// 2020 through 2023, both the ASC and OPPS conversion medpac.gov/docs/default-source/payment-basics/medpac_ factors will be the hospital MB index minus a multifactor payment_basics_18_asc_final_sec.pdf?sfvrsn=0. productivity adjustment. For most covered procedures, payment rates in the ASC We are concerned that neither the CPI–U nor the hospital payment system are the product of a relative weight and MB index reflects ASCs’ cost structure (see text box, a conversion factor. The ASC relative weight, which March 2019 | Report to the Congress: Medicare Payment Policy 129

160 p. 145). The Commission has recommended that CMS OPPS but not for the ASC system. C–APCs combine all collect cost data from ASCs to identify a price index that hospital outpatient services reported on a claim that are covered under Medicare Part B into a single payment, with would be an appropriate proxy for ASC costs (Medicare a few exceptions. CMS has not implemented C–APCs Payment Advisory Commission 2010). However, the ASC industry has opposed the collection of cost data for in the ASC system because the system of processing ASC claims does not allow for the type of packaging of this purpose (Ambulatory Surgery Center Association ancillary items necessary to create C–APCs. Therefore, 2012), and CMS does not yet collect these data. In 2018, the payment bundles for services in the C–APCs under CMS requested comments on whether the Secretary the OPPS have greater packaging of ancillary items should collect cost data from ASCs to use in determining than the same services under the ASC payment system. ASC payment rates. Representatives of individual ASCs Consequently, a disconnect exists between OPPS provided comments that generally opposed a policy payment rates and ASC payment rates for the services that would require ASCs to submit formal cost reports, but were willing to complete surveys on the condition that are in C–APCs under the OPPS. The magnitude of that they would not be administratively burdensome this disconnect has grown over time because CMS has (Centers for Medicare & Medicaid Services 2017). The substantially expanded the number of C–APCs. Currently, about 72 percent of HCPCS codes for surgical procedures Commission asserts, however, that all other institutional that are covered under the ASC payment system are providers submit at least abbreviated versions of cost in C–APCs under the OPPS. The Commission supports reports to CMS, including small entities such as hospices the use of C–APCs in the OPPS and encourages CMS to and home health agencies. Moreover, the ASCs in Pennsylvania are able to submit revenue and cost data each implement them in the ASC payment system because the year to the Pennsylvania Health Care Cost Containment greater packaging of ancillary items that occurs with C– Council, so it is clear that submission of cost data is APCs gives providers an incentive to furnish care more feasible for ASCs. Indeed, submitting revenue and cost efficiently. data does not appear to adversely affect ASC participation: Although we do not have recent ASC cost data that In Pennsylvania, there were seven more ASCs in 2017 would allow us to quantify cost differences between than in 2016. settings, evidence suggests that ASCs are a lower cost CMS uses a different method from the one described setting than HOPDs. The Government Accountability above to determine payment rates for procedures that are Office (GAO) compared ASC cost data from 2004 with predominantly performed in physicians’ offices and were HOPD costs and found that costs were, on average, lower first covered under the ASC payment system in 2008 in ASCs than in HOPDs (Government Accountability 2 or later. Payment for these “office-based” procedures is Office 2006). In addition, studies that used data from the the lesser of the amount derived from the standard ASC National Survey of Ambulatory Surgery found that the method or the practice expense portion of the PFS rate average time for ambulatory surgical visits for Medicare that applies when the service is provided in a physician’s patients was 25 percent to 39 percent lower in ASCs office (the nonfacility practice expense, which covers the than in HOPDs, which likely contributes to lower costs 3 equipment, supplies, nonphysician staff, and overhead in ASCs (Hair et al. 2012, Munnich and Parente 2014). 1 costs of a service). An additional study using data from a facility that has The physicians who provide these both an ASC and a hospital found that surgeries took services receive a separate payment under the PFS. CMS 17 percent less time in the ASC (Trentman et al. 2010). set this limit on the rate for office-based procedures to Trentman and colleagues and Munnich and Parente prevent migration of these services from physicians’ estimated less time savings in ASCs than did Hair and offices to ASCs for financial reasons. colleagues, likely because Trentman and colleagues The ASC payment system generally parallels the OPPS in and Munnich and Parente accounted for differences in terms of which ancillary services are paid separately and health status between patients treated in ASCs and those which are packaged into the payment of the associated treated in HOPDs, while Hair and colleagues did not. surgical procedure. In 2015, however, the connection Beneficiaries who are sicker may require more time between the ASC payment system and the OPPS was to treat. We have found that, on average, beneficiaries weakened when CMS implemented comprehensive receiving surgical services in HOPDs are not as healthy ambulatory payment classifications (C–APCs) for the Ambulatory surgical center services: Assessing payment adequacy and updating payments 130

161 TABLE Number of ASCs grew, 2012–2017 5–1 Average annual percent change Type of ASC 2012 2016 2017 2012–2016 2016–2017 2.4% Total 5,216 5,474 5,603 1.0% N/A 189 New 176 159 N/A 60 N/A Closed or merged 114 90 N/A ASC (ambulatory surgical center), N/A (not applicable). The average annual percentage change data for the “new” and “closed or merged” categories are shown Note: as “N/A” because they are outside the purpose of this table, which is to show the growth in the total number of ASCs. Source: MedPAC analysis of Provider of Services file from CMS, 2018. beneficiaries has been fairly stable. Access to ASCs as beneficiaries receiving those services in ASCs, as may be beneficial to patients and physicians compared indicated by risk scores from the CMS hierarchical with HOPDs, the provider type most similar to ASCs. condition categories risk adjustment model. For patients, ASCs can offer more convenient locations, shorter waiting times, and easier scheduling relative to HOPDs. ASCs offer physicians more control over their Are Medicare payments adequate in work environment and specialized staff. In addition, 2019? beneficiaries’ cost sharing is lower in ASCs than in HOPDs. However, these same qualities could lead to To address whether payments for the current year (2019) overuse of surgical procedures. are adequate to cover the costs of efficient providers and how much payments should change in the coming Capacity and supply of providers: Number of ASCs year (2020), we examine several measures of payment is increasing adequacy. We evaluate beneficiaries’ access to care by From 2016 to 2017, the number of ASCs increased 2.4 examining the supply of ASC facilities and changes percent to 5,603 ASCs (Table 5-1). This annual growth over time in the volume of services provided, providers’ rate was faster than the period from 2012 to 2016, when access to capital, and changes in ASC revenue from the the number of ASCs increased 1.0 percent per year. In Medicare program. However, our assessment of quality 2017, the number of new ASCs increased by 189, while of care (another measure of payment adequacy) is limited 60 ASCs closed or merged with other facilities. The and does not fully represent quality in ASCs. Most of our number of ASCs that closed or merged has declined each available indicators of payment adequacy are positive. year from 2012 to 2017, and the number of new ASCs has outnumbered closed ASCs. In addition, through the Beneficiaries’ access to care: Supply of ASCs first three-quarters of 2018, a reported 106 new ASCs and volume of services indicate adequate have opened in several states (Dyrda 2018a, Dyrda access 2018b). Beneficiaries have adequate access to care in ASCs, Several factors may explain the relatively slower growth although some groups—such as beneficiaries dually of ASCs between 2012 and 2016 and faster growth eligible for Medicare and Medicaid, African Americans, from 2016 to 2017. From 2012 to 2016, to expand their and beneficiaries under age 65—are less likely than outpatient surgery capacity, many hospitals acquired the average beneficiary to receive care in ASCs than in and integrated ASCs into the hospital or developed HOPDs. The number of ASC facilities has increased, new surgery centers that were part of the hospital. and the volume of services provided to Medicare March 2019 | Report to the Congress: Medicare Payment Policy 131

162 However, in response to provisions in the Bipartisan TABLE Budget Act of 2015 (Section 603), CMS has aligned Most ASCs are for profit and urban 5–2 payment rates for facilities established as off-campus ASCs that were: PBDs after November 2, 2015, with PFS payment rates, which are typically lower than ASC rates. Therefore, there New in Open in Open in is little incentive for a hospital system to acquire an ASC Type of ASC 2017 2017 2012 and convert it to an off-campus PBD. Instead, it is more 93.8% For profit 93.6% 92.6% financially beneficial to maintain the facility as an ASC. 5.8 3.5 3.8 Nonprofit 2.7 2.7 Government 1.6 The number of operating rooms (ORs) in ASCs is also growing. In 2017, there were nearly 17,000 ORs in ASCs, Urban 92.9 94.2 92.5 or an average of 3.0 per facility. From 2012 to 2016, the 7.4 Rural 7.1 5.8 total number of ASC ORs increased 0.8 percent per year, a slightly slower rate than the growth in the number of ASCs Note: ASC (ambulatory surgical center). Some totals do not sum to 100 percent over the same period (1.0 percent per year). However, because of rounding. from 2016 to 2017, the number of ORs in ASCs increased MedPAC analysis of Provider of Services file from CMS, 2018. Source: by about 1.6 percent, also a slightly slower rate than the growth rate in the number of ASCs from 2016 to 2017, which suggests the size of ASCs has declined since 2012. For example, ASCs that entered the market in 2017 had an average of 2.7 ORs, while those operating in 2012 had an average of 3.1 ORs. This approach may have limited the market for new freestanding ASCs (Jacobson 2014, Kochman 2014, Consistent with previous years, most ASCs in 2017 were Levingston 2014, Moody 2014, Sowa 2014). During for profit (93.8 percent) and located in urban areas (92.9 this time, hospitals’ decisions to increase their outpatient percent) (Table 5-2). However, ASCs that were new in surgery capacity may have been influenced by the 2017 were slightly more likely to be nonprofit and urban higher rates Medicare pays for ambulatory surgical (including urban and suburban areas) compared with services provided in HOPDs relative to ASCs (in 2019, existing ASCs. Beneficiaries who do not live near an ASC Medicare’s rates are 94 percent higher in HOPDs than in can obtain ambulatory surgical services in HOPDs and, 4 ASCs). In addition, during this period, physicians were in some cases, physicians’ offices. Beneficiaries who live increasingly choosing to be employed by hospitals rather in rural areas can travel to urban areas to receive care in than work in an independent practice (American Medical ASCs. In addition, most ASCs are freestanding, located off Association 2017, Berenson et al. 2012, Mathews 2012, a hospital campus (99 percent) (data not shown). Medicare Payment Advisory Commission 2013a, Merritt Hawkins 2014, Physicians Advocacy Institute 2018). Geographic distribution of ASCs is uneven In general, these physicians are more likely to provide In addition to being much more common in urban areas ambulatory procedures in the hospitals that employ them than rural areas, the concentration of ASCs varies widely than in freestanding ASCs. However, from 2016 to 2017 among states. In 2017, Maryland had the most ASCs and beyond, hospital systems such as Tenet Healthcare per Medicare beneficiary (40 ASCs per 100,000 Part B Corporation and HCA Healthcare Inc. have invested more beneficiaries), followed by Georgia, Alaska, and Wyoming substantially in outpatient surgical capacity and ASCs. (approximately 20 ASCs per 100,000 beneficiaries) Some believe this new strategy is intended to respond to (Figure 5-1, p. 133). Kentucky, the District of Columbia, the trend toward value-based care and the associated desire Alabama, West Virginia, and Vermont had the fewest to conduct surgeries in lower cost settings such as ASCs ASCs per beneficiary (fewer than 4 ASCs per 100,000 (Barclays 2018, Japsen 2018, Moody’s Investors Service beneficiaries). Availability in Vermont was especially low, 2018). Last, hospital systems that acquire ASCs have the with less than 1 ASC per 100,000 beneficiaries, and only 1 option of maintaining the facility as an ASC or converting 5 ASC in the entire state. it to an off-campus provider-based department (PBD) of a hospital (most likely an outpatient surgery department). Even though beneficiaries can largely receive the same services in HOPDs if an ASC is not located near them, the Ambulatory surgical center services: Assessing payment adequacy and updating payments 132

163 FIGURE Freestanding Medicare margins... FIGURE X-X Number of ASCs per beneficiary varies widely across states 5–1 45 40 35 30 25 20 15 10 Number of ASCs per 100,000 beneficiaries 5 0 MD NJ NE AZ NV MS DE UT IN LA TX ND AK PA OH MT MN SC IL MI RI VT NM NY KY AL KS HI WV MO SD CT WA WI OR OK WY NC TN ME CO MA AR GA IA FL VA CA DC ID NH State Note: ASC (ambulatory surgical center). Source: MedPAC analysis of CMS Provider of Services file for 2018 and Medicare denominator file for 2017. Note: Note and Source are in InDesign. Source: beneficiaries—defined as those living in an MSA. In 2017, small number of ASCs in some states and rural areas may 7.2 percent of rural beneficiaries received care in an ASC raise concerns about beneficiaries’ access to ambulatory versus 10.4 percent of urban beneficiaries. surgical services in the context of site-neutral payments Notes about this graph: between ASCs and HOPDs. In its June 2013 report, the • Data is in the datasheet. Make updates in the datasheet. Specialization of ASCs largely unchanged, some Commission identified surgical services that are viable for growth in pain management site-neutral payments between the ASC payment system • WATCH FOR GLITCHY RESETS WHEN YOU UPDATE DATA!!!! and the OPPS (Medicare Payment Advisory Commission The majority of ASCs that billed Medicare in 2017 • The column totals were added manually. 2013a). The impact of site-neutral payments between specialized in a single clinical area, with gastroenterology • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. ASCs and HOPDs would be to lower payment for some and ophthalmology being the most common, and ASCs services in HOPDs. Hospitals could respond by reducing specializing in pain management services are growing • I can’t delete the legend, so I’ll just have to crop it out in InDesign. the extent to which they provide these services. In areas as a share of ASCs. Overall, in 2017, 61 percent of • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph that have low ASC concentration, site-neutral payments ASCs were single-specialty facilities and 40 percent default when you change the data. could make it more difficult for beneficiaries to access were multispecialty facilities, providing services in 6 ambulatory surgical services. • Use paragraph styles (and object styles) to format. more than one clinical specialty (Table 5-3, p. 134). In 2017, the most common single-specialty ASCs focused • Data was from: R:\Groups\MGA\data book 2007\data book 2007 chp1 We found that rural beneficiaries—defined as those who on gastroenterology (21 percent) and ophthalmology live outside metropolitan statistical areas (MSAs)—are (21 percent). The most common multispecialty ASCs less likely to receive care in an ASC than are urban March 2019 | Report to the Congress: Medicare Payment Policy 133

164 TABLE Specialization of ASCs billing Medicare in 2015 and 2017 5–3 2015 2017 Number of Share of Number of Share of all ASCs Type of ASC ASCs all ASCs ASCs 61% 61% 2,890 Single specialty 2,878 1,019 1,027 22 21 Gastroenterology Ophthalmology 1,020 22 1,022 21 Pain management 355 8 368 8 191 Dermatology 4 179 4 125 3 3 124 Urology Podiatry 95 2 88 2 29 Orthopedics/musculoskeletal 23 0 1 24 1 Respiratory 0 16 0 9 0 11 OB/GYN 0 10 Cardiology 18 0 Neurology 5 0 6 0 1 0 3 Other 0 Multispecialty 1,802 38 1,878 40 More than 2 specialties 1,421 30 1,415 30 5 6 Pain management and either neurology or orthopedics 221 288 4 Gastroenterology and ophthalmology 160 3 175 4,768 100 100 Total 4,680 Note: ASC (ambulatory surgical center), OB/GYN (obstetrics and gynecology). A “single-specialty ASC” is defined as one with more than 67 percent of its Medicare claims in one clinical specialty. A “multispecialty ASC” is defined as one with more than 67 percent of its Medicare claims in more than one clinical specialty. ASCs included in this analysis are limited to those in the 50 states and the District of Columbia with a paid Medicare claim in 2017. Columns containing the share of all ASCs may not sum to 100 percent due to rounding. MedPAC analysis of Medicare carrier file claims, 2017. Source: for this trend to continue as momentum grows for that focused on two specialties in 2017 were those knee and hip arthroplasty (knee and hip replacement) specializing in pain management and either neurology or to be done in ambulatory settings. orthopedic services (6 percent of all ASCs). From 2015 to 2017, ASCs specializing in pain management services ASCs can offer patients greater convenience than • grew most rapidly. Across both single-specialty and HOPDs, such as the ability to schedule surgery more multispecialty ASCs in 2017, there were roughly 100 more quickly. pain management ASCs than in 2015. • For most procedures covered under the ASC payment Continued growth in the number of ASCs suggests that system, beneficiaries’ coinsurance is lower in ASCs Medicare’s payment rates have been adequate. Other 7 than in HOPDs. factors also have likely influenced the long-term growth in the number of ASCs: Physicians have greater autonomy in ASCs than in • HOPDs, which enables them to design customized • Changes in clinical practice and health care surgical environments and hire specialized staff. technology have expanded the provision of surgical procedures in ambulatory settings. There is potential Ambulatory surgical center services: Assessing payment adequacy and updating payments 134

165 TABLE Volume of ASC services per FFS beneficiary increased in 2017 5–4 Average annual change 2012 2016 2017 2012–2016 2016–2017 1.4% Volume of services (in millions) 6.0 6.4 6.5 1.8% 1.7 1.2 193.3 190.1 181.2 Volume per 1,000 FFS beneficiaries ASC (ambulatory surgical center), FFS (fee-for-service). The volume of services for 2012 and 2016 has been modified to reflect the volume of services covered Note: under the ASC payment system in 2017 that was provided in those years. MedPAC analysis of physician/supplier standard analytic files from CMS, 2012–2017. Source: • Physicians who invest in ASCs and perform surgeries through 2016 and increased by 1.7 percent in 2017 (Table on their patients in those ASCs can increase their 5-4). On average, the number of services per beneficiary revenue by receiving a share of ASC facility receiving care in ASCs increased at an average annual rate payments. The federal anti-self-referral law (also of 0.8 percent from 2012 through 2016 and 1.0 percent in known as the Stark Law) does not apply to ASC 2017 (data not shown). services. Services that have historically contributed the most to Because physicians are able to perform more • overall ASC volume continued to be a large share of the procedures in ASCs than in HOPDs in the same total in 2017. For example, the HCPCS code for cataract amount of time, they can earn more revenue from removal with intraocular lens insertion (HCPCS 66984) professional fees. had the highest volume in both 2012 and 2017, accounting for 18.9 percent of the total in 2012 and 18.8 percent in Increased interest across the health care industry in • 2017. Moreover, 19 of the 20 most frequently provided the concept of value-based care and the provision of HCPCS codes in 2012 were among the 20 most frequently care in lower cost settings has increased the strategic provided in 2017 (Table 5-5, p. 137). These services made investment interest of hospital systems, insurers, and up about 71 percent of ASC Medicare volume in 2012 and private equity firms in ASCs (Barclays 2018, Japsen 70 percent in 2017. 2018). A potential concern about the services most frequently Number of beneficiaries treated and volume of provided in ASCs is the extent to which they are services per beneficiary increased from 2016 to unnecessary or low value, such as spinal injections and 2017 other pain management services (Pinto et al. 2012). The number of fee-for-service (FFS) beneficiaries treated We have found that the volume of pain management in ASCs and the volume of ASC surgical services per FFS services grew robustly from 2012 to 2017. Table 5-5 beneficiary increased from 2016 to 2017. Because ASC shows that from 2012 to 2017, injections of foramen services are covered under Part B, we limited our analysis epidural into either the lumbar or sacral area, injecting to FFS beneficiaries who have Part B coverage. The the paravertebral facet joint in the lumbar or sacral area, number of FFS beneficiaries who received ASC services injecting an anesthetic into the sacroiliac joint, and grew by an average of 1.0 percent per year from 2012 destruction of nerves in the lumbar or sacral facet joint through 2016 and increased by 0.4 percent in 2017 (data all grew strongly. Moreover, the volume of insertion or not shown). The volume of services per FFS beneficiary replacement of spinal neurostimulators increased sharply increased by an average of 1.2 percent per year from 2012 from about 2,000 in 2012 to 9,500 in 2017 (data not shown). March 2019 | Report to the Congress: Medicare Payment Policy 135

166 ASCs. This issue is pertinent to the ASC sector because Volume of outpatient surgical procedures increased by a higher percentage in ASCs than in among even the most frequently provided services in HOPDs in 2017 ASCs, a substantial volume is provided in HOPDs. For example, 434,000 Medicare-covered cataract surgeries For the first time in several years, surgical volume in 2017 with intraocular lens insertion were performed in HOPDs increased at a faster rate in ASCs than in HOPDs. From in 2017, which was 26 percent of the total volume for this 2012 through 2016, average annual growth in volume service. per FFS beneficiary of surgical services covered by the ASC payment system was 1.2 percent in ASCs compared Concern remains, however, about services provided in with 2.4 percent in HOPDs. In 2017, volume per FFS ASCs rather than HOPDs because most ASCs have beneficiary increased by 1.7 percent in ASCs and by 0.7 some degree of physician ownership. Studies offer some percent in HOPDs. evidence that physicians who have an ownership stake in an ASC perform a higher volume of certain procedures The higher growth in ASCs in 2017 relative to HOPDs is than physicians who do not (Hollingsworth et al. 2010, a reversal of what occurred in previous years when growth Mitchell 2010, Strope et al. 2009). Other studies suggest in HOPDs was higher than in ASCs. This change is likely that the presence of an ASC in a market is associated a reflection of the same factors that contributed to the with a higher volume of outpatient surgical procedures faster growth in the number of ASCs in 2017, discussed (Hollenbeck et al. 2014, Hollingsworth et al. 2011, Koenig earlier. That is, the higher volume growth in ASCs in and Gu 2013). Although none of these studies assessed the 2017 was a response to the trend toward value-based care appropriateness of the additional procedures, they suggest and the associated desire to conduct surgeries in lower that the presence of ASCs might increase overall surgical cost settings, such as ASCs (Barclays 2018, Japsen 2018, volume. Moody’s Investors Service 2018). Also, beginning in 2017, when a hospital system acquires an ASC, provisions in Another setting that has a substantial overlap of services Section 603 of the Bipartisan Budget Act of 2015 have with ASCs is physician offices. In general, Medicare made it more financially advantageous to maintain the payment rates are higher in ASCs than in physician offices facility as an ASC rather than convert it to an off-campus for the same procedure. Services that are frequently PBD of a hospital. provided in both ASCs and physician offices include cystoscopy, pain management, and, to a lesser extent, Maintaining or expanding access to ASCs can be cataract procedures. Cystoscopy is performed much more beneficial for patients and Medicare frequently in offices than in ASCs, pain management is Maintaining beneficiaries’ access to ASCs has some about equally common in these two settings, and cataract benefits because services provided in this setting are procedures are done more frequently in ASCs than in less costly to Medicare and beneficiaries than services physician offices. 8 delivered in HOPDs. Medicare payment rates for surgical services performed in HOPDs are almost twice as high Quality of care: ASC-reported quality data as in ASCs. For example, the payment rate in 2019 for demonstrate modest improvement cataract surgery with intraocular lens insertion (the service ASC-reported quality data demonstrated modest most frequently provided in ASCs) is $1,917 in HOPDs improvement in recent years. CMS established the ASC compared with $977 in ASCs. The lower payment rate in Quality Reporting (ASCQR) Program in 2012 (Centers ASCs for this service has been financially beneficial to for Medicare & Medicaid Services 2011). Under this Medicare and beneficiaries. Other recent studies similarly system, ASCs that do not successfully submit data that find that ASCs are less costly than HOPDs in the Medicare measure quality have their payment update for that year and non-Medicare context and that the recent price growth reduced by 2 percentage points. Actual performance on at ASCs has been slower than price growth at HOPDs these quality measures does not affect an ASC’s payments; (Carey 2015, Robinson et al. 2015). ASCs are required only to submit the data to receive a full update. The Commission has recommended a value- Medicare program spending and overall beneficiary cost based purchasing program for ASCs that would reward sharing could be reduced if more surgical services were high-performing providers and penalize low-performing provided in ASCs than HOPDs or if HOPD payment providers (see text box, p. 140). rates are reduced to the level that Medicare sets for Ambulatory surgical center services: Assessing payment adequacy and updating payments 136

167 TABLE The 20 most frequently provided ASC services 5–5 in 2017 were similar to those provided in 2012 2017 2012 Percent Percent Rank Surgical service of volume of volume Rank Cataract surgery w/ IOL insert, 1 stage 1 18.8% 1 18.9% Upper GI endoscopy, biopsy 2 8.0 2 8.9 6.9 6.5 3 3 Colonoscopy and biopsy Lesion removal colonoscopy (snare technique) 5.0 4 5.9 4 Inject foramen epidural: lumbar, sacral 4.3 5 4.7 5 4.3 After cataract laser surgery 6 4.3 6 8 2.9 7 3.7 Injection spine: lumbar, sacral (caudal) Diagnostic colonoscopy 10 1.9 8 3.3 9 2.6 Inject paravertebral: lumbar, sacral 3.3 7 9 2.0 10 2.1 Colorectal screen, high-risk individual Colorectal screen, not high-risk individual 1.8 11 1.8 11 Cataract surgery, complex 1.5 13 1.4 12 18 1.3 13 0.9 Upper GI endoscopy, diagnosis Revision of upper eyelid 1.1 14 0.9 19 23 1.0 Lesion removal colonoscopy (hot biopsy forceps) 0.6 15 Inject spine, cervical or thoracic 1.0 16 1.0 16 Injection procedure for sacroiliac joint, anesthetic 1.0 17 1.4 14 18 17 Cystoscopy 1.0 1.0 20 Upper GI endoscopy, insertion of guide wire 0.9 19 0.8 1.6 20 12 Destroy lumbar/sacral facet joint 0.8 Total 71.1 70.0 Note: ASC (ambulatory surgical center), IOL (intraocular lens), GI (gastrointestinal). Components do not sum to totals due to rounding. MedPAC analysis of physician/supplier standard analytic files from CMS, 2012 and 2017. Source: (ASC–1 through ASC–4) and the delay of measures of The quality measures for which ASCs submit data 9 patient experience. continue to evolve. Over the past year, changes made to For 2022, CMS will implement two the ASCQR Program are the result of CMS’s Meaningful new claims-based measures of beneficiaries’ visits to Measures Initiative. In the last two years, CMS made a hospital subsequent to an ASC orthopedic or urology several revisions to the initial ASCQR measure set, which procedure, respectively (ASC–17 and ASC–18). resulted in CMS measuring ASC quality based on eight Results from reported ASC quality data measures (plus one voluntary measure) for 2019 and four measures (plus one voluntary measure) for 2021 Data reported by ASCs for four years (2013 to 2016) (Table 5-6, p. 138). In recent years, CMS has chosen to suggest improvement in ASC quality of care. Among the discontinue or delay several measures that were considered 11 quality measures for which data were available in 2016, “topped out” (meaning full or nearly full compliance performance improved for most measures. For the four with these measures has been reached), demonstrated adverse event measures (ASC–1, ASC–2, ASC–3, and less utility, or were not ready for use, including the ASC–4), the data show consistently low levels of these discontinuation of the current adverse event measures events in each of the four years and gradual improvement March 2019 | Report to the Congress: Medicare Payment Policy 137

168 TABLE Quality measures used in the ASC Quality Reporting Program 5–6 Required in: 2021 Description of quality measure 2019 a ASC–1: Patient burn Ye s No a Ye s ASC–2: Patient fall No a Wrong site, wrong side, wrong patient, wrong procedure, wrong implant ASC–3: No Ye s a Hospital transfer/admission Ye s No ASC–4: b No Prophylactic intravenous antibiotic timing ASC–5: No b ASC–6: No No Safe-surgery checklist use b ASC–7: ASC facility volume data on selected ASC surgical procedures No No c Influenza vaccination coverage among health care personnel ASC–8: No Ye s ASC–9: Endoscopy/polyp surveillance: Appropriate follow-up interval for normal colonoscopy in Ye s Ye s average-risk patients ASC–10: Endoscopy/polyp surveillance: Colonoscopy interval for patients with a history of d Ye s No adenomatous polyps—avoid inappropriate use Voluntary Cataracts: Improvement in patient’s visual function within 90 days following cataract surgery Voluntary ASC–11: Ye s Ye s ASC–12: Facility seven-day risk standardized hospital visit rate after outpatient colonoscopy ASC–13: Normothermia outcome: Percentage of patients under anesthesia who are normothermic within e 15 minutes of arrival in the post-anesthesia care unit No Ye s ASC–14: Unplanned anterior vitrectomy: Percentage of cataract surgery patients who have an e No Ye s unplanned removal of the vitreous ASC–15: Five patient experience measures from the Consumer Assessment of Healthcare Providers and ® survey measures: Systems ASC–15a: About facilities and staff ASC–15b: Communication about procedure ASC–15c: Preparation for discharge and recovery ASC–15d: Overall rating of facility f No No ASC–15e: Recommendation of facility f 16: Toxic anterior segment syndrome (TASS) – No No ASC g No – No 17: Hospital visits after orthopedic ASC procedures ASC g No No – 18: Hospital visits after urology ASC procedures ASC Note: ASC (ambulatory surgical center). a Retained in the ASC Quality Reporting (ASCQR) Program, but data collection is suspended by CMS starting in 2019. As a result, the measure will not be used for payment year 2021. b Discontinued by CMS from the ASCQR Program beginning in 2018. c Discontinued by CMS from the ASCQR Program beginning in 2020. d Discontinued by CMS from the ASCQR Program beginning in 2021. e CMS will activate this measure in 2020. f CMS has delayed the implementation of this ASCQR measure indefinitely. g CMS will activate this measure in 2022. Final rule for outpatient prospective payment system and ambulatory surgical center payment system, 2019. Source: (Table 5-7). Specifically, the share of ASCs reporting zero increased from 88 percent to 92 percent, and the share of adverse events increased over time. For example, from ASCs without any patient falls increased from 91 percent 2013 to 2016, the share of ASCs without any patient burns to 94 percent (data not shown). Ambulatory surgical center services: Assessing payment adequacy and updating payments 138

169 TABLE ASC quality measure levels, 2013–2016 5–7 Estimated Mean percent among ASCs number of events ASC quality measure 2014 2013 2015 in 2016* 2016 0.43% 0.49% 0.24% 11,500 0.36% ASC–1: Share of patients suffering burns ASC–2: Share of patients suffering falls 0.18 0.10 0.14 0.08 4,000 0.07 ASC–3: Share of patients suffering a “wrong” event 0.03 1,400 0.03 0.03 0.45 0.51 ASC–4: Share of patients transferred to a hospital 21,000 0.43 0.42 74 75 77 N/A ASC–8: Share of ASC staff receiving an influenza vaccination ASC–9: Share of average risk patients with appropriate N/A endoscopy/polyp surveillance 77 80 81 ASC–10: Share of patients with polyp history with appropriate 79 79 endoscopy/polyp surveillance 80 N/A ASC–11: Share of patients with vision improvement 90 days after cataract surgery 96 96 N/A Note: ASC (ambulatory surgery center), N/A (not applicable). *The number of events was estimated using the average reported rate of occurrence and the total number of ASC claims in 2016 (4.9 million). The estimated number of events is not calculated for measures that do not pertain to adverse events. Source: Medicare Hospital Compare data for ASCs, 2013–2016. In addition to the adverse event measures, other ASCQR on its decisions to discontinue three process measures measures demonstrated improvement. For example, from in 2018 and for adding the two claims-based unplanned 2013 to 2016, the share of ASCs reporting their staff hospitalization measures for 2022. However, the received influenza vaccinations (ASC–8) increased from Commission maintains concern about four issues related 74 percent to 77 percent. Improvement and generally high to the ASCQR Program: levels of performance were also observed for measures The program does not include enough claims-based • of the surveillance and follow-up of patients treated for measures assessing clinical outcomes that apply to the certain gastroenterology or cataract surgeries. While room various specialties practiced at ASCs. For example, if for improvement exists for three of these other measures no further changes are made to the ASCQR measure (ASC–8, ASC–9, and ASC–10), these data appear to be set before 2022, the measure set will include two 10 trending in a positive direction. measures for ASCs conducting colonoscopies, one measure for ASCs conducting cataract surgeries, one ASC quality measures should continue to be measure for ASCs conducting orthopedic procedures, refined and one measure for ASCs conducting urology The Commission asserts CMS should continue to procedures. This potential measure set appears to improve the ASCQR Program by moving toward more exclude many services provided at ASCs. CMS-calculated claims-based outcome measures that apply to all ASCs. In addition, the Commission asserts • CMS’s delay of the Consumer Assessment of ® ® ASCQR measures should be synchronized with measures Healthcare Providers and Systems (CAHPS ) patient included in the Hospital Outpatient Quality Reporting experience survey quality data excludes an important 11 (OQR) Program to facilitate comparisons between part of assessing quality of care. Among the ASCs and HOPDs. The Commission commends CMS March 2019 | Report to the Congress: Medicare Payment Policy 139

170 Creating a value-based purchasing program for ambulatory surgical centers An ASC VBP should give rewards based on clear, n 2012, the Commission recommended that the absolute, and prospectively set performance targets (as Congress authorize and CMS implement a value- opposed to “tournament models,” which require that based purchasing (VBP) program for ambulatory I some providers gain while others lose). The Medicare surgical centers (ASCs). A VBP program would program should take into account, as necessary, reward high-performing providers and penalize low- differences in a provider’s population, including social performing providers (Medicare Payment Advisory 12 risk factors. Because adjusting results for social risk Commission 2012). factors can mask disparities in clinical performance, CMS established a quality reporting program for ASCs Medicare should account for social risk factors by in 2012. However, Medicare payments to ASCs are directly adjusting payment through peer grouping, not adjusted based on how they perform on quality where benchmarks for achievement are group specific, measures, only on whether they report the measures. and each provider is compared with its peers, defined The Commission believes that high-performing ASCs as providers that have similar patient populations in should be rewarded and low-performing facilities terms of social risk factors. In addition, funding for should be penalized through the payment system. VBP incentive payments should come from existing Medicare spending for ASC services. Initially, funding Consistent with the Commission’s overall position on for the incentive payments should be set at 1 percent to Medicare quality measurement, an ASC VBP program 2 percent of aggregate ASC payments. The size of this should incorporate measures that are patient-oriented, pool should be expanded gradually as more measures encourage coordination across providers and time, and are developed and ASCs become more familiar with promote change in the delivery system. The ASC VBP the program. (Our March 2016 report to the Congress program should include outcome, patient experience, provides more detail about our recommendation to and value measures (a value measure would address CMS about an ASC VBP program (Medicare Payment services that are costly but of low value). Also, quality Advisory Commission 2016). ■ measurement should not be burdensome for providers. ASCs can choose to use more granular measures to manage their own quality improvement. Commission’s quality measurement principles is that the ASCQR Program and OQR Program, relying quality programs include patient experience (Medicare either on measures of general surgical procedures or Payment Advisory Commission 2018b). CAHPS is measures of specific surgical procedures common the only survey in the ASCQR Program that queries to both settings. For example, CMS could consider patients about their experience. implementing OQR measure OP–36 (the number of hospital visits after any outpatient surgery) within the • ASCQR measures should be further synchronized ASCQR Program or implementing ASCQR measures with OQR measures to facilitate comparison across ASC–17 and ASC–18 (the number of hospital ASCs and HOPDs. For 2019 and 2020, the ASCQR visits following orthopedic and urology procedures, Program and the OQR Program possess five common respectively) within the OQR Program. In addition, quality measures that pertain to cataract procedures, the aforementioned delay in implementing the colonoscopy procedures, and rates of influenza CAHPS patient experience measures affects both the vaccination among health care personnel. CMS ASCQR Program and OQR Program and impedes the should consider further expanding the overlap of comparison of ASCs and HOPDs. Ambulatory surgical center services: Assessing payment adequacy and updating payments 140

171 TABLE Share of ASC cases with subsequent hospital visits, 2014 and 2017 5–8 Subsequent hospital visit within 7 days after discharge from ASC 2014 2017 Share of cases Share of cases within type Number of within type Number of of ASC ASC cases of ASC Type of ASC ASC cases All ASCs 90,552 1.9% 98,714 2.0% 2.4 Multispecialty 38,562 43,582 2.2 4,871 Gastroenterology and ophthalmology 2.6 5,311 1.9 51,990 1.7 55,132 1.8 Single specialty 6,745 Pain management 2.4 2.2 7,266 Urology 4,068 3.7 4,814 4.1 633 7.9 Cardiology 235 7.2 ASC (ambulatory surgical center). “Subsequent hospital visit” includes inpatient admissions, observation services, and emergency department visits but excludes Note: cases related to trauma or mental health services. MedPAC analysis of Medicare physician, hospital outpatient, and hospital inpatient claims. Source: version of this type of measure that applies to all • All reported quality data should continue to be made specialties and procedures, similar to OQR measure publicly available. In prior years, CMS elected to OP–36 (the number of hospital visits after any allow ASCs to voluntarily and temporarily withhold outpatient surgery). We found that in 2017, 2.0 percent their quality data from public reporting (Medicare of ASC discharges were associated with a subsequent Payment Advisory Commission 2016). The hospital visit within seven days after discharge from Commission disagrees with this practice except in rare 13,14 an ASC (Table 5-8). From 2014 to 2017, the circumstances. measure of subsequent hospitalizations within seven Other quality measures: Some ASC specialties days was fairly consistent across all ASCs. However, show increases in hospitalizations subsequent to the share of subsequent hospital visits increased ASC discharge slightly (suggesting quality of care worsened) at Because of the concerns cited above and the potential multispecialty ASCs, such as those specializing in value of clinical outcome measures that apply to all ASCs, both gastroenterology and ophthalmology (from 1.9 we believe CMS could consider developing new ASC percent in 2014 to 2.6 percent in 2017), and some quality measures covering any or all of the three following types of single-specialty ASCs. Although our measure areas: is not risk adjusted, it should be if used in the ASCQR Program or used to compare the performance of ASCs CMS should more broadly develop a measure of the • with HOPDs. number of Medicare beneficiaries discharged from ASCs who have subsequent unplanned hospital visits. • CMS could consider developing a measure of surgical CMS has already begun to implement these measures site infections (SSIs) occurring at ASCs for the for certain specialties (e.g., ASC–12, ASC–17, and ASCQR Program. CMS could calculate this measure ASC–18), but CMS has not developed these measures from claims rather than require ASCs to report it. for specialty areas or individual procedures that are Researchers have found that lapses in infection common to ASCs. The Commission developed a control were common among a sample of ASCs March 2019 | Report to the Congress: Medicare Payment Policy 141

172 the individual identification of ASCs. AHRQ asserts in three states (Schaefer et al. 2010). The Hospital Inpatient Quality Reporting Program includes an that these data can be used by ASCs to improve their SSI measure that applies primarily to inpatient practices and by the public to inform decisions about where to receive care (Agency for Healthcare Research procedures. Although CMS has considered an SSI and Quality 2018). measure for ASCs in the past (Centers for Medicare & Medicaid Services 2011), it is not currently working ASCs’ access to capital: Growth in number to develop one (Centers for Medicare & Medicaid of ASCs suggests adequate access Services 2016). In general, an SSI measure could be used to track infection rates for ASCs and identify Owners of ASCs require capital to establish new facilities quality improvement opportunities for ambulatory and upgrade existing ones. The change in the number of surgeries conducted in HOPDs and ASCs. In addition, ASCs is the best available indicator of ASCs’ ability to measuring SSI rates could encourage providers to obtain capital. The number of ASCs increased in 2017 collaborate and better coordinate care for ambulatory by 2.4 percent, faster than in previous years (Table 5-1, surgery patients. p. 131). In addition, through the first three-quarters of 2018, a reported 106 new ASCs have opened in several CMS could consider developing new measures • different states (Dyrda 2018a, Dyrda 2018b). However, that rely on specialty-specific clinical guidelines Medicare accounts for a small share—perhaps 20 to assess the appropriateness of specific services percent—of ASCs’ overall revenue, so factors other than conducted at ASCs. While the ASCQR Program Medicare payments may have a larger effect on access currently includes two ASC-reported colonoscopy to capital for this sector (Medical Group Management measures that assess appropriate follow-up care, CMS Association 2009). could consider claims-based measures that assess appropriateness. For example, current American A series of ASC acquisitions in recent years suggests Cancer Society guidelines state that patients over ASCs are a highly valued asset for hospital systems, the age of 85 should no longer receive colorectal private equity firms, and insurers. In 2015, Tenet cancer screening (American Cancer Society 2018). Healthcare Corporation, traditionally a hospital company, Using these guidelines, a new measure could identify began incrementally acquiring progressively larger shares the ASC-level share of colonoscopy cases in which of ASC chain United Surgical Partners (USP) (Kutscher beneficiaries are over age 85. CMS could consider 2015). Throughout 2017 and 2018, Tenet increased its similar measures of appropriateness for certain investment in USP, and in mid-2018 Tenet purchased an procedures that have become more common in ASCs additional 15 percent of USP from a private equity firm in recent years or concerns about appropriate use have for $630 million (Kacik 2018). In 2018, USP was the been suggested, such as spinal injections or certain second largest ASC firm, accounting for more than 200 orthopedic procedures. ASCs. This 2018 purchase increased Tenet’s ownership of USP to 95 percent, with the remaining 5 percent Department of Health and Human Services will owned by the health system Baylor, Scott, and White. In publicly report ASC-specific patient safety data general, hospital systems are increasingly turning their In response to the expanding scope of ASC services investment attention away from the inpatient setting and and the desire of ASCs to compare their performance toward ASCs and other outpatient capacity (Barclays with other ASCs, the Department of Health and Human 2018, Japsen 2018). For example, ASCs in 2017 Services (HHS), through the Agency for Healthcare accounted for roughly 20 percent of Tenet’s earnings. Research and Quality (AHRQ), will collect and publicly Currently, Tenet owns over 300 ASCs and HCA owns report survey data on ASC-specific patient safety culture more than 120 ASCs. In addition, Tenet and HCA state (Agency for Healthcare Research and Quality 2018, in their 2018 financial reports that ASCs are a component Dickson 2018a, Dickson 2018b). AHRQ worked with of their business strategy moving forward (Morningstar the ASC industry to design this program. Similar to its Document Research 2018a, Morningstar Document hospital safety survey data, AHRQ will collect survey Research 2018b). From 2016 to 2017, Tenet reported a 10 data from ASC staff regarding their perceptions of safety percent increase in ASC cases and a 14 percent increase culture in their workplace. This information will be in operating revenues (Morningstar Document Research reported on the AHRQ website in a format permitting Ambulatory surgical center services: Assessing payment adequacy and updating payments 142

173 TABLE Medicare payments to ASCs grew, 2012–2017 5–9 Average annual change 2012 2016 2017 2012–2016 2016–2017 $4.6 Medicare payments (in billions of dollars) $3.6 $4.3 4.1% 7.4% 3.5 $136 $126 $110 Medicare payments per FFS beneficiary 7.7 Note: ASC (ambulatory surgical center), FFS (fee-for-service). “Medicare payments” includes program spending and beneficiary cost sharing for ASC facility services. Payments include spending for new-technology intraocular lenses. Source: MedPAC analysis of data from the Office of the Actuary at CMS and data from physician/supplier standard analytic files. 2018a). Financial analysts assert that these hospital Although the various entities noted above appear to systems are acquiring ASCs or partnering with entities have adequate access to capital, we caution that these that own ASCs to better acclimate to a value-based care companies have ownership in a small share of the more environment that will require providing surgeries in lower than 5,000 ASCs. Consequently, the experience of these entities collectively may not reflect that of the entire ASC cost settings (Barclays 2018). sector. In addition, in October 2018, private equity firm Kohlberg, Kravis, Roberts, and Company completed the purchase Medicare payments: Payments have steadily of Envision Healthcare for $9.9 billion (Bannow 2018). increased Envision Healthcare owns over 250 ASCs as a part of its In 2017, ASCs received $4.6 billion in Medicare payments 2017 purchase of AmSurg Corporation. In January 2017, and beneficiaries’ cost sharing (Table 5-9). We estimate Surgical Care Associates—which owned approximately that spending by the Medicare program was $3.7 billion 200 ASCs in 33 states—was acquired by insurer and beneficiary cost sharing was $900 million (data not UnitedHealth Group’s Optum for $2.3 billion (Mathews shown). 2017). This acquisition is part of a larger stated effort by the insurer to provide primary care and ambulatory Spending per FFS beneficiary increased by an average services. annual rate of 3.5 percent from 2012 through 2016 and by 7.7 percent in 2017 (Table 5-9). The increase in 2017 Strong financial positions of this magnitude suggest that reflects a 1.9 percent increase in the ASC conversion ASCs are attractive to investors. Security and Exchange factor, a 1.7 percent increase in per capita volume, a 3.8 Commission filings from Surgery Partners Inc. (SPI), percent increase in the average relative weight of ASC which is an operator of nearly 100 ASCs and is not services, and a 0.3 percentage point increase from higher affiliated with a hospital or insurer, reported increases use of separately payable drugs (data not shown). The in revenue per case (11 percent) and same-store volume growth in spending in 2017 is unusually large. Relative to (14 percent) from 2017 to 2018 (Surgery Partners 2016, the higher growth in 2017 reflects a higher increase 2018b). SPI also demonstrated the ability to access in the ASC conversion factor and a higher increase in per capital by announcing in October 2018 the acquisition capita volume. The strong growth in the average relative of a $180 million loan for use in merger and acquisition weight that occurred in 2016 continued in 2017. In both activity (Surgery Partners 2018a). Finally, data from the 2016 and 2017, this growth was driven by increased Pennsylvania Health Care Cost Containment Council’s volume for high-cost procedures, such as implantation annual analysis of the state’s ASCs show that ASCs in of spinal neurostimulators, which may have resulted in Pennsylvania had an average total margin of 25 percent lower volume for relatively low-cost injections for pain in 2017 (Pennsylvania Health Care Cost Containment management. 15 Council 2018). March 2019 | Report to the Congress: Medicare Payment Policy 143

174 limited resources for collecting cost data. However, such How should Medicare payments change businesses typically keep records of their costs for filing in 2020? taxes and other purposes, and other facility providers that are typically small, such as home health agencies and Our analysis indicates that the number of ASCs has hospices, furnish cost data to CMS. increased, as has beneficiaries’ use of ASCs, and access To minimize the burden on CMS and ASCs, CMS should to capital has been adequate. Certain measures of ASC create a streamlined process for ASCs to track and submit quality indicate improvement, although we have identified a limited amount of cost data. As it did in 1986 and 1994, areas for improvement in ASC quality measurement. Our CMS could annually conduct a survey of a random sample information for assessing payment adequacy, however, is of ASCs, with mandatory response. The Government limited because Medicare does not require ASCs to submit Accountability Office conducted a similar random sample cost data, unlike other types of facilities. Since 2010, the survey of ASC costs in 2004. CMS could also streamline Commission has recommended that the Congress require ASC cost reporting by annually collecting a set of cost that ASCs submit cost data (Medicare Payment Advisory variables from all ASCs that is more limited than what Commission 2010). is collected through formal cost reports, which would Cost data would enable the Commission to examine the require less time for ASCs to complete. Alternatively, growth of ASCs’ costs over time and analyze Medicare CMS could require ASCs to submit cost data from their payments relative to the costs of efficient providers, existing cost accounting systems, provided the definitions which would help inform our decisions about the ASC of their reported cost variables are consistent with CMS’s update. Cost data are also needed to examine whether definitions. The Commission does not believe that a an alternative input price index would be an appropriate streamlined process for collecting cost data would place a proxy for ASC costs. As discussed in the text box, the large burden on ASCs. After all, individual taxpayers are Commission has previously expressed concern that the able to complete and submit lengthy income tax forms. price index CMS used to update the ASC conversion factor Therefore, the Commission sees no reason ASCs cannot from 2010 through 2018 (the CPI–U) likely does not submit at least minimal cost data. reflect ASCs’ cost structure (Medicare Payment Advisory For the Commission to determine the relationship between Commission 2010). Also, the price index that CMS has Medicare payments and the costs of efficient ASCs, ASCs said it will use to update the ASC conversion factor from would optimally submit the following information: 2019 through 2023—the hospital MB—does not reflect ASCs’ cost structure. total costs for the facility; • CMS has concluded that it needs data on ASC input costs Medicare unallowable costs, such as entertainment, • (Centers for Medicare & Medicaid Services 2012). To promotion, and bad debt; date, CMS has not required ASCs to submit cost data. However, CMS requested public comment on whether the costs of clinical staff who bill Medicare • the agency should collect cost data from ASCs for use separately, such as anesthesiologists and clinical nurse in determining ASC payment rates. ASC representatives anesthetists (these costs would be excluded from commented that they oppose a requirement for ASCs to the facility’s costs because these clinicians are paid submit formal cost reports, but expressed willingness separately under Medicare); to complete surveys if doing so is not administratively • total charges across all payers and charges for burdensome (Centers for Medicare & Medicaid Services Medicare patients (CMS could allocate total facility 2017). costs to Medicare based on Medicare’s proportion of We believe it is feasible for ASCs to provide cost total charges); and information. All other facility providers submit cost data • total Medicare payments. to CMS. Indeed, ASCs in Pennsylvania submit cost and revenue data annually to a state agency that uses the data In addition, CMS would need to collect data on specific to estimate margins for those ASCs (Pennsylvania Health cost categories to determine an appropriate input Care Cost Containment Council 2018). We recognize price index for ASCs. For example, CMS would need that ASCs are generally small facilities that may have Ambulatory surgical center services: Assessing payment adequacy and updating payments 144

175 Revisiting the ASC market basket index ASC payment rates from 2019 through 2023 (Centers rom 2010 through 2018, CMS used the consumer for Medicare & Medicaid Services 2018). As we stated price index for all urban consumers (CPI–U) as above, our analysis of GAO cost data showed that the market basket (MB) to update the conversion F ASCs have a different cost structure than hospitals. factor in the ambulatory surgical center (ASC) payment Therefore, we do not believe the hospital MB is an system. Because of our concern that the CPI–U likely appropriate market basket for ASCs. does not reflect ASCs’ cost structure, the Commission examined in 2010 whether an alternative MB index The ASC cost data from GAO used in our comparative would better measure changes in ASCs’ input costs analysis are 15 years old and do not contain information (Medicare Payment Advisory Commission 2010). on several types of costs. Therefore, the Commission Using data from a Government Accountability Office has recommended several times that the Congress (GAO) survey of ASC costs in 2004, we compared require ASCs to submit new cost data to CMS (Medicare the distribution of ASC costs with the distribution of Payment Advisory Commission 2018c, Medicare hospital and physician practice costs. We found that Payment Advisory Commission 2015, Medicare Payment ASCs’ cost structure is different from that of hospitals Advisory Commission 2014, Medicare Payment and physician offices. ASCs have a much higher share Advisory Commission 2013b, Medicare Payment of expenses for medical supplies and drugs than the Advisory Commission 2012, Medicare Payment other two settings, a much smaller share of employee Advisory Commission 2011b, Medicare Payment compensation costs than hospitals, and a smaller share Advisory Commission 2010). In each of the last six of all other costs (such as rent and capital costs) than years, the Commission recommended eliminating the physician offices. For more detail about our methods update to the ASC conversion factor, meaning the ASC and findings, see Chapter 2C of our March 2010 conversion factor would not change from the previous report to the Congress (Medicare Payment Advisory year. CMS should use cost data to examine whether an Commission 2010). existing Medicare price index is an appropriate proxy for ASC costs or an ASC-specific market basket should Since our 2010 analysis, CMS has considered whether be developed. A new ASC MB could include the same the hospital MB or the practice expense component of types of costs that appear in the hospital MB or MEI but the Medicare Economic Index (MEI) is a better proxy with different cost weights that reflect ASCs’ unique cost for ASC costs than the CPI–U (Centers for Medicare structure. ■ & Medicaid Services 2012). Most recently, CMS has decided to use the hospital MB as the basis for updating data on the share of ASCs’ costs related to employee 2015, 0.3 percent in 2016, 1.9 percent in 2017, and 1.2 compensation, medical supplies, medical equipment, percent in 2018. However, CMS has indicated that the building expenses, and other professional expenses (such CPI–U does not reflect the input costs of ASCs. as legal, accounting, and billing services). CMS could use CMS has made a significant regulatory change and this information to examine the cost structure of ASCs and decided to use the hospital MB as the basis for updating determine whether an existing Medicare price index is an the ASC conversion factor for a five-year period—2019 appropriate proxy for ASC costs or an ASC-specific MB through 2023. CMS based its decision to use the hospital should be developed. MB in place of the CPI–U on concerns that the differences CMS used the CPI–U to update the ASC conversion in payment rates between the ASC payment system and factor from 2010 through 2018. Using the CPI–U, CMS the OPPS has caused a shift of care from ASCs to HOPDs. increased the ASC conversion factor by 1.4 percent in March 2019 | Report to the Congress: Medicare Payment Policy 145

176 CMS believes that using the same update mechanism for of services continue to grow. Therefore, we believe it is both ASCs and HOPDs could “encourage the migration of unnecessary for CMS to spend five years assessing the services from the hospital setting to the ASC setting and feasibility of collecting cost data from ASCs. increase the presence of ASCs in health care markets or Recommendation geographic areas where previously there were none or few, thus promoting better beneficiary access to care” (Centers In evaluating a need for an update to the ASC conversion for Medicare & Medicaid Services 2018). The update factor for 2020, the Commission balanced the following to the ASC conversion factor for 2019 is 2.1 percent, objectives: which is based on a projected 2.9 percent increase in the maintain beneficiaries’ access to ASC services; • hospital MB minus a 0.8 percent reduction for multifactor productivity growth, as mandated by the Patient Protection • pay providers adequately; and Affordable Care Act of 2010. maintain the sustainability of the Medicare program • During the five-year period of using the hospital MB, by appropriately restraining spending on ASC CMS states that it will: services; assess whether there is a migration of services from • keep providers under financial pressure to constrain • hospitals to ASCs and costs; and assess the possibility of working with stakeholders • • require ASCs to submit cost data. to collect cost data from ASCs in a minimally burdensome manner and could propose a plan to collect cost data (Centers for Medicare & Medicaid In balancing these goals, the Commission concludes that Services 2018). the ASC update for 2020 should be eliminated and that the Secretary should collect cost data from ASCs. Beginning with the Commission’s March 2010 report to the Congress, the Commission has stated for several RECOMMENDATION 5-1 years in comment letters and in published reports that the CPI–U does not likely reflect the current input costs of The Congress should eliminate the calendar year 2020 ASCs (Medicare Payment Advisory Commission 2010). update to the Medicare conversion factor for ambulatory However, the Commission does not support using the surgical centers. hospital MB index as an interim method for updating the RECOMMENDATION 5-2 ASC conversion factor because evidence indicates that the hospital MB index does not accurately reflect the The Secretary should require ambulatory surgical centers costs of ASCs (Medicare Payment Advisory Commission to report cost data. 2018a). CMS acknowledges that the ASC cost structure is RATIONALE 5-1 AND 5-2 not identical to that of hospitals because ASCs tend to be single specialty and for profit, and they are not required On the basis of our payment adequacy indicators and the to comply with the Emergency Medical Treatment and importance of maintaining financial pressure on providers Labor Act of 1986. The Commission concurs with these to constrain costs, we believe that the ASC conversion observations and adds that, relative to hospitals, ASCs are factor should not be increased for 2020. That is, the 2020 more urban, serve a different mix of patients, have a much conversion factor in the ASC payment system should be higher share of expenses related to medical supplies and the same as the conversion factor in 2019. Though we drugs, and have a smaller share of employee compensation do not have cost data and we have reservations about the costs. measures used within the ASCQR Program, the indicators of payment adequacy for which we have information are The Commission asserts that CMS should forgo the positive: The volume of ASC services per beneficiary five-year period to assess the feasibility of ASC cost increased in 2017, the complexity of ASC services reporting and instead use its authority and resources to act provided increased, and the number of ASCs increased. quickly in gathering ASC cost data. ASCs are profitable Also, ASCs appear to have adequate access to capital, organizations, and the number of ASCs and the volume Ambulatory surgical center services: Assessing payment adequacy and updating payments 146

177 IMPLICATIONS 5-1 AND 5-2 ASC quality of care data have trended positive, and Medicare payments to ASCs have continued to grow. Spending The Commission has persistently recommended that the The Secretary has the authority to update the ASC • Secretary collect cost data from ASCs. Cost data would conversion factor and has decided to use the hospital enable CMS and the Commission to examine the growth MB index as the basis for updating the conversion of ASCs’ costs over time and evaluate Medicare payments factor from 2019 through 2023 (Centers for Medicare relative to the costs of an efficient provider, which & Medicaid Services 2018). The Patient Protection would help inform decisions about the ASC payment and Affordability Act of 2010 requires that the update update. Cost data are also needed to evaluate whether factor be reduced by a multifactor productivity an alternative input price index would be an appropriate measure. The currently projected hospital MB index proxy for ASC costs. increase for 2020 is 3.2 percent, and the forecast of productivity growth for 2020 is 0.6 percent, resulting We see no reason why ASCs should not be able to in a projected update of 2.6 percent to the conversion submit cost data. CMS collects cost data from all other factor for 2020. Relative to current Medicare law, our institutional providers participating in the Medicare recommendation would decrease federal spending by program. To date, the ASC industry has asserted that between $50 million and $250 million in the first year ASCs are small operations that lack the capacity and and by less than $1 billion over five years. accounting expertise to enable them to complete cost Beneficiary and provider reports. However, some of the sectors from which CMS Because of the growth in the number of ASCs and • collects cost data are predominantly small providers. the increase in ASCs’ revenue from Medicare, we do Therefore, any ASC should be able to compile and submit not anticipate that this recommendation will diminish a minimum set of cost data. Also, while the majority beneficiaries’ access to ASC services or providers’ of the ASC industry consists of freestanding facilities, willingness or ability to provide those services. hospital corporations and other large health care entities have entered the ASC industry in recent years and have • ASCs may incur some minimal administrative costs the capacity and expertise to complete cost reports. CMS to track and submit cost data, but we believe cost could limit the scope of the cost reporting system to accounting is standard practice in the ASC industry, minimize administrative burden on ASCs and the program. and ASCs should be able to draw cost data from that In addition, to implement this change, CMS should make source. ■ cost reporting a condition of ASC participation in the Medicare program. March 2019 | Report to the Congress: Medicare Payment Policy 147

178 Endnotes 8 Cost sharing is lower under the ASC payment system for 96.8 The payment rates in the ASC system are determined 1 percent of HCPCS codes that are covered under the ASC independently from the payment rates in the PFS. Therefore, payment system. it is possible for an office-based procedure to have its payment rate based on the standard method in one year and based on Rather than a full discontinuation of measures ASC–1 through 9 the PFS nonfacility rate the next year, or vice versa. ASC–4, CMS has decided to suspend these four measures. Suspension means that ASCs are no longer required to GAO surveyed a random sample of 600 ASCs to obtain cost 2 report data on these measures, but CMS will retain them data from 2004. They received reliable cost data from 290 in the ASCQR Program for possible future use. Patient facilities. experience will be assessed using the Consumer Assessment ® ® Munnich and Parente (2014) also found that the highest risk 3 ) survey (CAHPS of Healthcare Providers and Systems patients that underwent the five highest volume outpatient measures, but implementation of CAHPS measures has been procedures were less likely to have a subsequent visit to an delayed. emergency department or a hospital inpatient stay when they 10 We did not include data for ASC–6 (safe-surgery checklist) received the outpatient procedure in an ASC rather than a because ASC response rates were low, which we assume to be hospital. related to CMS discontinuing the measure for 2018. 4 For services that CMS has defined as device intensive (at least ® is a registered trademark of the Agency for 11 CAHPS 30 percent of the cost of the service is attributable to a device), Healthcare Research and Quality, a U.S. government agency. the differences in the payment rates between HOPDs and ASCs are smaller than 94 percent because the reimbursement The Commission also described its principles for a VBP 12 for the applicable device is the same in ASCs and HOPDs. program for ASCs in a letter to the Congress commenting on the Secretary’s report to the Congress on a VBP program for 5 State certificate of need (CON) laws for ASCs appear to affect ASCs (Medicare Payment Advisory Commission 2011a). the number of ASCs in the state. Twenty-seven states and the District of Columbia have CON laws for ASCs. Nine of the Subsequent hospital visits include emergency department 13 10 states with the fewest ASCs per capita have a CON law in services, outpatient observation services, and inpatient place, while only 4 of the 10 states that have the most ASCs services. per capita have CON laws. Among these four states, Maryland and Georgia have exceptions in their CON requirements that Among the approximately 100,000 ASC discharges associated 14 make it easier to establish new ASCs. with subsequent hospital stays within 7 days, roughly two- thirds had subsequent inpatient hospital stays and one-third ASCs as those with more than 67 single-specialty We define 6 had subsequent visits to an emergency department (data not percent of their Medicare claims in one clinical specialty. shown). We define multispecialty ASCs as those with more than 67 percent of their Medicare claims in more than one clinical 15 The margins for ASCs have important differences from the specialty. margins in other sectors such as hospitals. In particular, the cost data used to determine margins for most ASCs do not By statute, coinsurance for a service paid under the OPPS 7 include compensation for physician owners or the taxes paid cannot exceed the hospital inpatient deductible ($1,364 in on that compensation. 2019). The ASC payment system does not have the same limitation on coinsurance; for a small share of HCPCS codes covered under the ASC payment system, the ASC coinsurance exceeds the inpatient deductible. In these instances, the ASC coinsurance exceeds the OPPS coinsurance. Ambulatory surgical center services: Assessing payment adequacy and updating payments 148

179 References Centers for Medicare & Medicaid Services, Department of Agency for Healthcare Research and Quality. 2018. Ambulatory Health and Human Services. 2016. Medicare program: Hospital surgery center survey on patient safety culture database. https:// www.ahrq.gov/sops/quality-patient-safety/patientsafetyculture/ outpatient prospective payment and ambulatory surgical center payment systems and quality reporting programs; organ asc/index.html. procurement organization reporting and communication; Ambulatory Surgery Center Association. 2017. Benefits transplant outcome measures and documentation requirements; of physician ownership. http://www.ascassociation.org/ electronic health record (EHR) incentive programs; payment to advancingsurgicalcare/asc/benefitsofphysicianownership. nonexcepted off-campus provider-based department of a hospital; hospital value-based purchasing (VBP) program; establishment Ambulatory Surgery Center Association. 2012. Letter from of payment rates under the Medicare physician fee schedule William Prentice, Chief Executive Officer, to Marilyn Tavenner, for nonexcepted items and services furnished by an off-campus Acting Administrator of the Centers for Medicare & Medicaid provider-based department of a hospital. Final rule. Federal Services, regarding CMS–1589–P: Medicare program; proposed Register 81, no. 219 (November 14): 79562–79892. changes to the ambulatory surgical center payment system and CY 2013 payment rates. September 4. Centers for Medicare & Medicaid Services, Department of Health and Human Services. 2012. Medicare and Medicaid programs: American Cancer Society. 2018. American Cancer Society Hospital outpatient prospective payment and ambulatory guideline for colorectal cancer screening. https://www.cancer. surgical center payment systems and quality reporting programs; org/cancer/colon-rectal-cancer/detection-diagnosis-staging/acs- electronic reporting pilot; inpatient rehabilitation facilities quality recommendations.html#written_by. reporting program; revision to quality improvement organization Federal Register 77, no. 221 (November regulations. Final rule. American Medical Association. 2017. Updated data on physician 15): 68210–68565. practice arrangements: Physician ownership drops below 50 . Chicago, IL: AMA. https://www.ama-assn.org/sites/ percent Centers for Medicare & Medicaid Services, Department of Health default/files/media-browser/public/health-policy/PRP-2016- and Human Services. 2011. Medicare and Medicaid programs: physician-benchmark-survey.pdf. Hospital outpatient prospective payment; ambulatory surgical center payment; hospital value-based purchasing program; Bannow, T. 2018. KKR completes Envision acquisition. Modern physician self-referral; and patient notification requirements in Healthcare , October 11. provider agreements. Final rule. 76, no. 230 Federal Register (November 30): 74122–74584. Health care services: Initiating coverage of Barclays. 2018. August 14. hospital sector. Dickson, V. 2018a. HHS proposes new ambulatory surgery center , July 11. Modern Healthcare safety database. Berenson, R. A., P. B. Ginsburg, J. B. Christianson, et al. 2012. The growing power of some providers to win steep payment Dickson, V. 2018b. White House OKs ambulatory surgery center increases from insurers suggests policy remedies may be needed. safety database. Modern Healthcare , September 17. Health Affairs 31, no. 5 (May): 973–981. Dyrda, L. 2018a. 39 new ASCs were opened or announced in Q3. Carey, K. 2015. Price increases were much lower in ambulatory , October 11. Becker’s ASC Review surgery centers than hospital outpatient departments in 2007–12. 34, no. 10 (October 1): 1738–1744. Health Affairs Dyrda, L. 2018b. Where the 67 new ASCs are being built and , July 16. Becker’s ASC Review opened in the 1st half of 2018. Centers for Medicare & Medicaid Services, Department of Health and Human Services. 2018. Medicare program: Changes to Medicare: Payment Government Accountability Office. 2006. hospital outpatient prospective payment and ambulatory surgical for ambulatory surgical centers should be based on the hospital center payment systems and quality reporting programs. Final . Washington, DC: GAO. outpatient payment system 83, no. 225 (November 21): 58818–59179. Federal Register rule. Hair, B., P. Hussey, and B. Wynn. 2012. A comparison of Centers for Medicare & Medicaid Services, Department of ambulatory perioperative times in hospitals and freestanding Health and Human Services. 2017. Medicare program: Hospital centers. American Journal of Surgery 204, no. 1 (July): 23–27. outpatient prospective payment and ambulatory surgical center payment systems and quality reporting programs. Final rule. 82, no. 217 (November 13): 52356–52637. Federal Register March 2019 Report to the Congress: Medicare Payment Policy | 149

180 Hollenbeck, B. K., R. L. Dunn, A. M. Suskind, et al. 2014. Medicare Payment Advisory Commission. 2018b. Report to Ambulatory surgery centers and outpatient procedure use among the Congress: Medicare and the health care delivery system . Medicare beneficiaries. Medical Care 52, no. 10 (October): Washington, DC: MedPAC. 926–931. Medicare Payment Advisory Commission. 2018c. Report to the Hollingsworth, J. M., S. L. Krein, Z. Ye, et al. 2011. Opening of . Washington, DC: MedPAC. Congress: Medicare payment policy ambulatory surgery centers and procedure use in elderly patients: Medicare Payment Advisory Commission. 2016. Report to the Data from Florida. 146, no. 2 (February): Archives of Surgery . Washington, DC: MedPAC. Congress: Medicare payment policy 187–193. Medicare Payment Advisory Commission. 2015. Report to the Hollingsworth, J. M., Z. Ye, S. A. Strope, et al. 2010. Physician- Congress: Medicare payment policy . Washington, DC: MedPAC. ownership of ambulatory surgery centers linked to higher volume 29, no. 4 (April): 683–689. of surgeries. Health Affairs Report to the Medicare Payment Advisory Commission. 2014. Congress: Medicare payment policy . Washington, DC: MedPAC. Jacobson, G. 2014. Tenet CEO Trevor Fetter sees big growth in outpatient centers. Dallas News , November 4. Medicare Payment Advisory Commission. 2013a. Report to the Congress: Medicare and the health care delivery system . Japsen, B. 2018. Tenet Healthcare signals outpatient acquisitions Washington, DC: MedPAC. , February Modern Healthcare ahead amid value-based care push. 27. Report to the Medicare Payment Advisory Commission. 2013b. . Washington, DC: MedPAC. Congress: Medicare payment policy Kacik, A. 2018. Tenet ups stake in ambulatory surgery center , April 26. Modern Healthcare chain. Report to the Medicare Payment Advisory Commission. 2012. . Washington, DC: MedPAC. Congress: Medicare payment policy Kochman, B. 2014. Montefiore Medical Center opens 12-story , New York Daily News bedless outpatient center in the Bronx. Medicare Payment Advisory Commission. 2011a. MedPAC November 10. comment letter on the Department of Health and Human Services’ Report to the Congress: Medicare Ambulatory Surgical Koenig, L., and Q. Gu. 2013. Growth of ambulatory surgical , August 30. Center Value-Based Purchasing Implementation Plan American centers, surgery volume, and savings to Medicare. 108: 10–15. Journal of Gastroenterology Report to the Medicare Payment Advisory Commission. 2011b. . Washington, DC: MedPAC. Congress: Medicare payment policy Kutscher, B. 2015. Tenet makes big ambulatory play with deal Modern Healthcare , for majority United Surgical Partners stake. Report to the Medicare Payment Advisory Commission. 2010. March 23. Congress: Medicare payment policy . Washington, DC: MedPAC. Levingston, C. 2014. Premier Health buys Middletown surgery Merritt Hawkins. 2014. Survey of 20,000 U.S. Physicians: 80% Journal-News center. , September 18. of doctors are over-extended or at full-capacity . Dallas, TX: Merritt Hawkins. http://www.merritthawkins.com/uploadedFiles/ Mathews, A. W. 2017. UnitedHealth’s Optum to acquire Surgical MerrittHawkings/Clients/2014_Physicians_Foundation_Merritt_ Wall Street Journal , January 9. Care Affiliates for $2.3 billion. Hawkins_Release%282%29.pdf. Mathews, A. W. 2012. Same doctor visit, double the cost: Insurers Mitchell, J. M. 2010. Effect of physician ownership of specialty say rates can surge after hospitals buy private physician practices; hospitals and ambulatory surgery centers on frequency of use Medicare spending rises, too. , August 27. Wall Street Journal 145, no. 8 Archives of Surgery of outpatient orthopedic surgery. Medical Group Management Association. 2009. Ambulatory (August): 732–738. surgery center performance survey: 2009 report based on 2008 Moody’s Investors Service. 2018. Non-profit and public health . Washington, DC: MGMA. data care: Medians–Operating pressures persist as growth in expenses Medicare Payment Advisory Commission. 2018a. Comment exceeds revenue. August 28. letter on CMS’s 2019 proposed rule for the hospital outpatient Moody, E. 2014. Beaufort Memorial Hospital becomes full owner and ambulatory surgical center payment systems, September of Surgery Center of Beaufort. , December 2. The State 21. http://www.medpac.gov/docs/default-source/comment- letters/09212018_opps_asc_2019_medpac_comment_v2_sec. pdf?sfvrsn=0. Ambulatory surgical center services: Assessing payment adequacy and updating payments 150

181 Morningstar Document Research. 2018a. Form 10–K: Tenet Schaefer, M. K., M. Jhung, M. Dahl, et al. 2010. Infection Healthcare Corporation. February 26. Journal of the control assessment of ambulatory surgical centers. American Medical Association 303, no. 22 (June 9): 2273–2279. Morningstar Document Research. 2018b. HCA Holdings, for 10–K. February 23. Sowa, T. 2014. Providence medical center in Spokane Valley The Spokesman-Review , April 12. nearly ready for patients. Munnich, E. L., and S. T. Parente. 2014. Procedures take less time at ambulatory surgery centers, keeping costs down and ability to Strope, S. A., S. Daignault, J. M. Hollingsworth, et al. 2009. Health Affairs 33, no. 5 (May): 764–769. meet demand up. Physician ownership of ambulatory surgery centers and practice patterns for urological surgery: Evidence from the state of Pennsylvania Health Care Cost Containment Council. 2018. 47, no. 4 (April): 403–410. Medical Care Florida. . Financial analysis 2017: Volume 2, ambulatory surgery centers Harrisburg, PA: PHC4. Surgery Partners Inc. 2018a. Surgery Partners, Inc. announces closing of new $180 million incremental term loan. Press Physicians Advocacy Institute. 2018. Updated physician release. October 24. https://globenewswire.com/news- acquisition study: National and regional changes release/2018/10/24/1626236/0/en/Surgery-Partners-Inc- in physician employment, 2012–2016. http://www. Announces-Closing-of-New-180-Million-Incremental-Term- physiciansadvocacyinstitute.org/Portals/0/assets/docs/2016-PAI- Loan.html. Physician-Employment-Study-Final.pdf. Surgery Partners Inc. 2018b. Surgery Partners, Inc. announces Pinto, R. Z., C. G. Maher, M. L. Ferreira, et al. 2012. Epidural third quarter 2018 results. Press release. November 7. http:// corticosteroid injections in the management of sciatica: A ir.surgerypartners.com/news-releases/news-release-details/ systematic review and meta-analysis. Annals of Internal Medicine surgery-partners-inc-announces-third-quarter-2018-results. 157, no. 12 (December 18): 865–877. Trentman, T. L., J. T. Mueller, R. J. Gray, et al. 2010. Outpatient Robinson, J. C., T. Brown, and C. Whaley. 2015. Reference- surgery performed in an ambulatory surgery center versus a based benefit design changes consumers’ choices and employers’ hospital: Comparison of perioperative time intervals. American 34, no. 3 (March payments for ambulatory surgery. Health Affairs 200, no. 1 (July): 64–67. Journal of Surgery 1): 415–422. March 2019 | Report to the Congress: Medicare Payment Policy 151

182

183 CHAPTER 6 Outpatient dialysis services

184 RECOMMENDATION 6 For calendar year (CY) 2020, the Congress should update the CY 2019 Medicare end-stage renal disease prospective payment system base rate by the amount determined in current law. COMMISSIONER VOTES: YES 17 • NO 0 • NOT VOTING 0 • ABSENT 0

185 CHAPTER 6 Outpatient dialysis services Chapter summary In this chapter Outpatient dialysis services are used to treat the majority of individuals with • Are Medicare payments end-stage renal disease (ESRD). In 2017, nearly 395,000 beneficiaries with adequate in 2019? ESRD on dialysis were covered under fee-for-service (FFS) Medicare and How should Medicare • received dialysis from approximately 7,000 dialysis facilities. Since 2011, payments change in 2020? Medicare has paid for outpatient dialysis services using a prospective payment system (PPS) that is based on a bundle of services. The bundle includes dialysis drugs and ESRD-related clinical laboratory tests that were previously paid separately. In 2017, Medicare expenditures for outpatient dialysis services were $11.4 billion, a 0.4 percent increase over 2016 expenditures. Assessment of payment adequacy Our payment adequacy indicators for outpatient dialysis services are generally positive. Beneficiaries’ access to care— Measures of the capacity and supply of providers, beneficiaries’ ability to obtain care, and changes in the volume of services suggest payments are adequate. • Dialysis facilities appear to have the Capacity and supply of providers— capacity to meet demand. Between 2016 and 2017, the number of dialysis treatment stations grew faster than the number of FFS dialysis beneficiaries. Volume of services— Between 2016 and 2017, growth in the number of • FFS dialysis beneficiaries and total number of treatments was relatively March 2019 | Report to the Congress: Medicare Payment Policy 155

186 flat. At the same time, dialysis drug use (including erythropoiesis-stimulating agents (ESAs), which are used in anemia management) continued to decline, but at a slower rate than during the initial years of the dialysis PPS (2011 and 2012). The dialysis PPS created an incentive for providers to be more judicious about their provision of dialysis drugs. • The marginal profit— Medicare payments exceeded marginal costs by about 17 percent in 2017, suggesting that providers have an incentive to treat Medicare beneficiaries. Quality of care— Between 2012 and 2017, there were declines in mortality, hospitalization, and 30-day readmission rates, though the proportion of FFS dialysis beneficiaries using the emergency department increased. With regard to anemia management, negative cardiovascular outcomes associated with high ESA use generally declined, and blood transfusion use, which initially increased under the PPS, has trended downward since 2013. Between 2012 and 2017, beneficiaries’ use of home dialysis, which is associated with improved patient satisfaction and quality of life, increased from 9.5 percent to 11.0 percent of dialysis beneficiaries. Since 2014, a shortage of dialysis solutions needed for the predominant home method, peritoneal dialysis, has slowed this modality’s growth. The first-year results of the ESRD Seamless Care Organizations (ESCOs), modeled like accountable care organizations, were positive; for example, there were fewer inpatient admissions for beneficiaries, and all 13 ESCOs produced savings relative to the benchmarks. It is not clear whether this trend will continue since the results for 2017 and 2018 are not yet available. Providers’ access to capital— Information from investment analysts suggests that access to capital for dialysis providers continues to be strong. The number of facilities, particularly for-profit facilities, continues to increase. Under the dialysis PPS, the two largest dialysis organizations have grown through acquisitions and mergers with midsized dialysis organizations. Medicare payments and providers’ costs— Our analysis of Medicare payments and costs is based on 2016 and 2017 claims and cost report data submitted to CMS by freestanding dialysis facilities. During this period, cost per treatment increased by 2 percent, while Medicare payment per treatment increased by 0.6 percent. We estimate that the aggregate Medicare margin was –1.1 percent in 2017, and the 2019 Medicare margin is projected at –0.4 percent. The Commission’s recommendation for 2020 is that the Congress update the ESRD PPS base rate by the amount determined under current law. ■ Outpatient dialysis services: Assessing payment adequacy and updating payments 156

187 Dialysis treatment choices independently in the patient’s home or workplace five ialysis replaces the filtering function of the to seven days a week. During treatments, a cleansing kidneys when they fail. The two types of fluid (dialysate) is infused into the patient’s abdomen dialysis—hemodialysis and peritoneal dialysis D through a catheter. This infusion process (an exchange) (PD)—remove waste products from the bloodstream is done either manually (continuous ambulatory differently. Patients may select various protocols for peritoneal dialysis) or using a machine (automated each of these two dialysis types. peritoneal dialysis). Most dialysis patients travel to a treatment facility to Each dialysis method has advantages and undergo hemodialysis three times per week, although disadvantages—no one method is best for everyone. patients can also undergo hemodialysis at home. People choose a particular dialysis method for many Hemodialysis uses an artificial membrane encased in a reasons, including quality of life, patients’ awareness of dialyzer to filter the patient’s blood. Because of recent different treatment methods and personal preferences, clinical findings, there is increased interest in more and physician training and recommendations. The use frequent hemodialysis, administered five or more times of home dialysis has grown since 2009, a trend that per week while the patient sleeps, and short (two to has continued under the dialysis prospective payment three hours per treatment) daily dialysis administered system. Some patients switch methods when their during the day. Research also has increased interest in conditions or needs change. Although most patients the use of “every-other-day” hemodialysis; reducing the still undergo in-center dialysis, home dialysis remains two-day gap in thrice-weekly hemodialysis could be a viable option for many patients because of such linked to improved outcomes. advantages as increased patient satisfaction, better PD, the most common form of home dialysis, uses health-related quality of life, and fewer transportation the lining of the abdomen (peritoneum) as a filter to challenges compared with in-center dialysis. ■ clear wastes and extra fluid and is usually performed 1 received dialysis from about 7,000 dialysis facilities. Background Since 2011, Medicare has been paying facilities using a prospective payment system (PPS) payment bundle that End-stage renal disease (ESRD) is the last stage of includes dialysis drugs (for which facilities previously chronic kidney disease and is characterized by permanent received separate payments) and services for which irreversible kidney failure. Patients with ESRD include other Medicare providers (such as clinical laboratories) those who are treated with dialysis—a process that previously received separate payments. In 2017, Medicare removes wastes and fluid from the body—and those who Part B spending for outpatient dialysis services included have a functioning kidney transplant. Because of the in the payment bundle was $11.4 billion. In addition, Part limited number of kidneys available for transplantation D payments for dialysis drugs that are not yet included in and the variation in patients’ suitability for transplantation, the PPS payment bundle—a calcimimetic and multiple about 70 percent of ESRD patients undergo maintenance phosphate binders—totaled nearly $2.3 billion in 2016 dialysis (see text box on dialysis treatment choices). (the most recent data available). Patients receive additional items and services related to their dialysis treatments, including dialysis drugs to treat Characteristics of fee-for-service dialysis conditions such as anemia and bone disease resulting from beneficiaries, 2017 the loss of kidney function. The 1972 amendments to the Social Security Act extended Medicare benefits to people with ESRD, including In 2017, nearly 395,000 ESRD beneficiaries on dialysis those under age 65. For an individual with ESRD to were covered under fee-for-service (FFS) Medicare and March 2019 | Report to the Congress: Medicare Payment Policy 157

188 they are diagnosed. In addition, Medicare permits MA TABLE enrollment of ESRD beneficiaries with a functioning FFS dialysis beneficiaries are 6–1 disproportionately younger, male, kidney transplant. In 2017, about 19 percent of ESRD and African American compared with beneficiaries were enrolled in MA plans; by comparison, all Medicare FFS beneficiaries, 2017 just over 31 percent of all Medicare beneficiaries were enrolled in MA plans. In 2000, the Commission Percent of FFS: recommended that the Congress lift the prohibition on All Dialysis ESRD beneficiaries enrolling in MA (Medicare Payment beneficiaries beneficiaries Advisory Commission 2000). The 21st Century Cures Act of 2016 lifts the prohibition on ESRD beneficiaries Age 11% 4% Under 45 years enrolling in MA beginning in 2021. 12 38 45–64 years In 2017, most FFS dialysis beneficiaries (about 90 28 65–74 years 50 percent) were enrolled in Part D or had other sources of 75–84 years 23 18 creditable drug coverage. About 10 percent of FFS dialysis 85+ years 6 11 beneficiaries in 2017 had either no Part D coverage or Sex coverage less generous than Part D’s standard benefit. 56 Male 47 About 70 percent of FFS dialysis beneficiaries with Part D 53 Female 44 coverage received the low-income subsidy in 2017. Race Compared with all Medicare FFS beneficiaries, FFS 48 81 White dialysis beneficiaries are disproportionately young, male, 36 African American 10 and African American (Table 6-1). In 2017, 77 percent of All others 9 16 FFS dialysis beneficiaries were less than 75 years old, 56 Residence, by type of county percent were male, and 36 percent were African American. 80 Urban 84 By comparison, of all FFS Medicare beneficiaries, 66 11 Micropolitan 10 percent were less than 75 years old, 47 percent were male, Rural, adjacent to urban 5 5 and 10 percent were African American. A greater share Rural, not adjacent to urban 3 2 of dialysis beneficiaries resided in urban areas compared 1 1 Frontier with all FFS beneficiaries (84 percent vs. 80 percent, respectively). FFS dialysis beneficiaries were more likely Note: FFS (fee-for-service). Beneficiary location reflects the beneficiary’s county of residence in one of four categories (urban, micropolitan, rural adjacent to be dually eligible for Medicaid and Medicare compared to urban, and rural nonadjacent to urban) based on an aggregation of with all Medicare FFS beneficiaries (48 percent vs. 18 the urban influence codes. Frontier counties have six or fewer people per square mile. Totals may not sum to 100 percent due to rounding. percent, respectively; data not shown). Source: Data compiled by MedPAC from enrollment data and claims submitted by The adjusted rate (or incidence) of new ESRD cases dialysis facilities to CMS. (which includes patients of all types of health coverage who initiate dialysis or receive a kidney transplant) rose sharply in the 1980s and 1990s, leveled off in the early 2000s, and has declined slightly since its peak in 2006. Between 2006 and 2016 (most recent year available), the adjusted incidence rate decreased by 1 percent per year, qualify for Medicare, he or she must be fully or currently from 399 per million people to 355 per million people insured under the Social Security or Railroad Retirement 3 (United States Renal Data System 2018). We estimate program or be the spouse or dependent child of an eligible 2 that in 2017, about 83,000 FFS beneficiaries were new to beneficiary. dialysis, and nearly half (45 percent) were under age 65 Most dialysis beneficiaries have FFS coverage. The statute and thus entitled to Medicare based on ESRD (with or 4 currently prohibits enrollment of individuals with ESRD in without disability). Medicare Advantage (MA) plans. However, beneficiaries Better primary care management of the risk factors for who were enrolled in a managed care plan before chronic kidney disease (CKD)—particularly hypertension receiving an ESRD diagnosis can remain in the plan after Outpatient dialysis services: Assessing payment adequacy and updating payments 158

189 could provide incentives for the efficient delivery of and diabetes, which together are the primary causes of quality care by broadening the payment bundle existing roughly 7 of 10 new ESRD cases—can help prevent or at the time (to include commonly furnished drugs and delay the illness’s onset. Payers and dialysis providers services that providers formerly billed separately) and by are testing interventions among CKD patients to improve linking payment to quality. The PPS is designed to create their clinical outcomes (e.g., by reducing hospitalizations); incentives for facilities to provide services more efficiently prevent or slow kidney disease progression; and increase by reducing previous incentives, inherent in the former their preparedness for ESRD (e.g., by educating patients payment method, to overuse drugs. about treatment alternatives, including transplantation 5 and home dialysis). The Commission has long argued Under the outpatient dialysis PPS, the unit of payment is that primary care services are undervalued in Medicare’s a single dialysis treatment. For adult dialysis beneficiaries fee schedule and has made recommendations to support (18 years or older), the base payment rate does not differ primary care, which in turn could support better by type of dialysis (i.e., hemodialysis vs. peritoneal management of kidney disease risk factors. dialysis), but rather by patient-level characteristics (age, body measurement characteristics, onset of dialysis, and Since 2011, Medicare has paid for dialysis selected acute and chronic comorbidities) and facility-level services under the dialysis PPS factors (low treatment volume, rural location, and local To treat ESRD, dialysis beneficiaries receive care from 7 input prices). Medicare pays facilities furnishing dialysis two principal providers: (1) the clinicians (typically treatments in the facility or in a patient’s home for up to nephrologists) who prescribe and manage the provision three treatments per week, unless there is documented of dialysis and establish the beneficiary’s plan of care and medical justification for more than three weekly (2) facilities that provide dialysis treatments in a dialysis treatments. In addition, the ESRD Quality Incentive center or support and supervise the care of beneficiaries Program holds facilities responsible for the quality of on home dialysis. Medicare uses different methods to pay care they provide; in 2018, the program used 11 clinical for ESRD clinician and facility services. Clinicians receive measures and 5 reporting measures. Up to 2 percent of a a monthly capitated payment established in the Part B facility’s payment is linked to these quality measures. The physician fee schedule for outpatient dialysis–related Payment Basics Commission’s provides more information management services, which varies based on the number about Medicare’s method of paying for outpatient of visits per month, the beneficiary’s age, and whether dialysis services (available at http://www.medpac.gov/ 6 the beneficiary receives dialysis in a facility or at home. docs/default-source/payment-basics/medpac_payment_ While our work in this report focuses on Medicare’s basics_18_dialysis_final_sec.pdf?sfvrsn=0). payments to facilities, it is important to recognize that facilities and clinicians collaborate to care for dialysis Since it was implemented in 2011, the outpatient dialysis beneficiaries. One acknowledgment of the need for PPS has undergone several significant changes. In 2014, collaboration is Medicare’s Comprehensive ESRD Care CMS rebased the base payment rate, and in 2016, the Model, a shared savings program that began in 2015, agency recalibrated and redefined the payment adjusters. involving facilities and nephrologists. A text box on the dialysis PPS (p. 161) summarizes these two significant changes. To improve provider efficiency, in 2011, Medicare began a PPS for outpatient dialysis services that expanded the The most recent change to the dialysis PPS will occur in prospective payment bundle to include dialysis drugs, 2020, when CMS expands its transitional drug add-on laboratory tests, and other ESRD treatment items and payment adjustment (TDAPA); the agency will pay services that were previously billable separately. In providers separately for all dialysis drugs and biologics, addition, effective in 2012, outpatient dialysis payments are including biosimilars and generic drugs, that the Food and linked to the quality of care that dialysis facilities provide. Drug Administration approves on or after January 1, 2020. These changes, mandated by the Medicare Improvements Payment will equal the product’s average sales price. for Patients and Providers Act of 2008 (MIPPA), were There will be no reduction to the base rate even when based on the Commission’s recommendation to modernize a new dialysis product falls into 1 of the 11 functional the outpatient dialysis payment system (Medicare Payment categories of products that have been included in the Advisory Commission 2001). We contended that Medicare payment bundle since 2011. The TDAPA will apply to a March 2019 | Report to the Congress: Medicare Payment Policy 159

190 new product that fits into one of the existing functional The policy would undermine competition with • 8 categories for two years; existing drugs included in the ESRD bundle. The thereafter, the product will be Commission has documented the changes in drug use included in the PPS payment bundle without any change due to increased price competition with the vitamin to the base rate. For a product that does not fit into one of D and ESA therapeutic classes (Medicare Payment the existing functional categories, the TDAPA will apply Advisory Commission 2018c). until sufficient claims data for rate-setting analysis are available, but not for less than two years. Once sufficient In our comment letter, we asserted that if CMS decides to claims data are available, CMS will modify the base proceed with this proposed policy, at a minimum several rate, if appropriate, to account for the new product in the modifications to the proposal would be necessary: payment bundle. • CMS should require that the new product be an In our comment letter to CMS regarding the agency’s advance in medical technology that substantially TDAPA proposal, the Commission strongly urged CMS improves beneficiaries’ outcomes relative to not to proceed with its proposal to apply the policy to technologies in the PPS payment bundle. CMS could new renal dialysis drugs that fit into a functional category structure such a policy similar to the standard that (including composite rate drugs, which have never been the agency uses to pay for new technologies under paid separately by Medicare) and urged the agency to the inpatient PPS and devices under the outpatient withdraw the proposal (Medicare Payment Advisory PPS. CMS elected to not include this modification Commission 2018a). The Commission believes that to the final policy, stating that (1) its final policy will it is important to maintain the structure of the dialysis provide an opportunity for new drugs to compete with PPS and not create policies that would unbundle other similar drugs in the market, which could result services covered under the PPS or create incentives in lower prices for all drugs; and (2) the effectiveness that encourage high launch prices of new drugs and of drugs can depend on age, gender, race, genetic technologies. Access to new dialysis products is favorable predisposition, and comorbidities (Centers for under the dialysis PPS. For example, in 2015, nearly Medicare & Medicaid Services 2018b). one-quarter of all dialysis beneficiaries received epoetin beta, which was introduced to the U.S. market in that • CMS should not make duplicative payments for a new year. The Commission’s comment letter can be found at product (assigned to a functional category) by paying http://www.medpac.gov/docs/default-source/comment- and paying for its under the TDAPA for two years letters/08312018_esrd_cy2019_dme_medpac_comment_ functional category under the dialysis PPS base rate. v2_sec.pdf?sfvrsn=0. For example, the agency could reduce the TDAPA amount to reflect the amount already included in the In our comment letter, we objected to the TDAPA proposal base rate. In addition, CMS could consider paying a because: reduced share of the estimated incremental cost of the new drug as a way to share risk with dialysis providers Although new dialysis drugs could improve patient • and provide some disincentive for the establishment of outcomes, the proposal does not require that the new high launch prices. CMS elected to not include these drugs be more effective than current treatments to modifications to the final policy, stating that the policy qualify for the TDAPA. Under CMS’s policy, the only is temporary and not duplicative because, at the end proposed standard for paying the TDAPA is that a of the TDAPA two-year period, there is no additional drug is new. money added to the base rate for those drugs that fall • The policy duplicates payment that is already made as within an existing functional category (Centers for part of the dialysis PPS payment bundle. Beneficiaries Medicare & Medicaid Services 2018b). and taxpayers already pay for drugs in each functional • CMS should publish in the final rule an estimate of category because they are included in the payment the increase in beneficiaries’ and taxpayers’ spending bundle. Essentially, the TDAPA policy will make a due to the proposed policy change and the method second (duplicative) payment for new drugs that treat used to develop the estimate. According to the agency, the same clinical condition as drugs already included an estimate of expected spending changes was not in the payment bundle. Outpatient dialysis services: Assessing payment adequacy and updating payments 160

191 Significant changes to the outpatient dialysis PPS in 2014 and 2016 payment per treatment. These adjusters are applied ince its implementation in 2011, the dialysis to the base payment rate to account for factors prospective payment system (PPS) has undergone that can affect treatment costs. More information two significant changes, in 2014 and 2016. First, S about these payment changes can be found in the effective 2014, the base payment rate was rebased Commission’s March 2016 report to the Congress to account for the decline in dialysis drug use under (available at http://www.medpac.gov/docs/default- the dialysis PPS. Based on statutory and regulatory source/payment-basics/medpac_payment_basics_17_ changes, CMS set the 2014 base payment at $239.02. dialysis_finald8a311adfa9c665e80adff00009edf9c. The Commission’s March 2014 report to the Congress pdf?sfvrsn=0). The Commission’s methodological provides more information about the rebasing of concerns about these patient-level and facility-level the dialysis base payment rate (available at http:// refinements can be found in our comment letter to CMS medpac.gov/docs/default-source/reports/mar14_ch06. (available at http://medpac.gov/docs/default-source/ pdf?sfvrsn=0). comment-letters/medpac-comment-on-cms-s-proposed- Second, beginning in 2016, CMS uses recalibrated rule-on-the-end-stage-renal-disease-prospective- and redefined patient-level and facility-level payment payment-system-and-.pdf?sfvrsn=0). ■ adjustments to calculate each patient’s adjusted the volume of services, and the marginal profitability of included because the TDAPA policy addresses drugs Medicare dialysis beneficiaries under the PPS—shows that and biologics that have not been developed (Centers beneficiaries’ access to care remains favorable. for Medicare & Medicaid Services 2018b). Capacity has kept pace with patient demand Growth in the number of dialysis facilities and treatment stations alongside growth in the number of dialysis Are Medicare payments adequate in beneficiaries suggests that, between 2012 and 2016, 2019? provider capacity kept up with demand for care. During that period, the number of facilities increased annually To address whether payments for 2019 are adequate to by 5 percent; facilities’ capacity to provide care—as cover the costs that efficient providers incur and how much measured by dialysis treatment stations—grew 4 percent providers’ costs should change in the update year (2020), annually (Table 6-2, p. 162). By contrast, between 2012 we examine several indicators of payment adequacy. and 2016, the number of FFS dialysis beneficiaries grew We assess beneficiaries’ access to care by examining the 9 1 percent annually (data not shown). In the same period, capacity of dialysis facilities and changes over time in the capacity at facilities that were freestanding and for volume of services provided. We also examine quality profit each grew by 5 percent annually, while capacity at of care, providers’ access to capital, and the relationship facilities that were hospital based and nonprofit decreased between Medicare’s payments and facilities’ costs. Most annually (–5 percent and –1 percent, respectively). of our payment adequacy indicators for dialysis services Between 2012 and 2016, capacity at urban facilities grew are positive. 4 percent per year, while capacity at all rural facilities grew about 2 percent per year. Between 2016 and 2017, Beneficiaries’ access to care: Indicators continue to be favorable total dialysis capacity grew by 3 percent, while the number of FFS dialysis beneficiaries grew more slowly Our analysis of access indicators—including the capacity (by 0.4 percent; data not shown). of providers to meet beneficiary demand, changes in March 2019 | Report to the Congress: Medicare Payment Policy 161

192 TABLE Increasing number and capacity of freestanding, 6–2 for-profit, and largest dialysis organizations 2017 Average annual percent change Number of Number of Total facilities stations number Total Mean of FFS Total number number treatments number of of 2012– 2012– 2016– of 2016– (in millions) stations facilities 2017 stations 2017 2016 2016 3% 45.3 7,014 120,900 17 5% 4% 4% All Percent of total 4 95% 94% 95% 17 6 4 5 Freestanding –4 –5 –2 13 5 6 5 Hospital based –4 5 4 4 5 18 85 82 86 Urban Micropolitan 2 0.5 2 1 16 10 11 10 2 0.2 2 Rural, adjacent to urban 3 4 3 13 2 3 3 11 2 2 1 Rural, not adjacent to urban 0 –0.3 Frontier 3 –1 0.2 0.5 0.3 9 3 3 5 6 17 89 88 91 For profit 4 5 11 Nonprofit 16 –2 –1 –1 –2 9 12 Two largest dialysis organizations 76 5 5 6 6 73 18 74 24 All others –2 2 –1 2 16 26 27 FFS (fee-for-service). Provider location reflects the county where the provider is located in one of four categories (urban, micropolitan, rural adjacent to urban, and Note: rural nonadjacent to urban) based on an aggregation of the urban influence codes. Frontier counties have six or fewer people per square mile. Totals may not sum to 100 percent due to rounding. Source: Compiled by MedPAC from the Dialysis Compare database from CMS and claims submitted by dialysis facilities to CMS. Two large dialysis organizations (LDOs) dominate the Providers of outpatient dialysis services dialysis industry. In 2017, these LDOs accounted for In 2017, there were roughly 7,000 dialysis facilities in the about 73 percent of facilities and 76 percent of Medicare United States that furnished about 45.3 million Medicare- treatments. In addition to operating most dialysis paid treatments to FFS dialysis beneficiaries. Medicare facilities, the two LDOs are each vertically integrated. FFS accounted for about 62 percent of all treatments 10 Both organizations operate an ESRD-related laboratory, a furnished in 2017. According to CMS facility survey pharmacy, and one or more centers that provide vascular data, since the late 1980s, for-profit, freestanding facilities access services; they provide ESRD-related disease have provided the majority of dialysis treatments. In management services; and they operate dialysis facilities 2017, freestanding facilities furnished 95 percent of FFS internationally. One LDO manufactures and distributes treatments, and for-profit facilities furnished about 91 renal-related pharmaceutical products (e.g., phosphate percent (Table 6-2). In 2017, the capacity of facilities in binders), is the leading supplier of dialysis products (such urban and rural areas was generally consistent with where FFS dialysis beneficiaries lived. Outpatient dialysis services: Assessing payment adequacy and updating payments 162

193 as hemodialysis machines and dialyzers) to other dialysis Trends in number of dialysis treatments provided companies, and operates a Phase I–IV drug and device Between 2016 and 2017, there was little change in the clinical development company that focuses on the clinical number of FFS dialysis beneficiaries (0.4 percent) and development of new renal therapies. total Medicare-covered dialysis treatments (45.3 million 11 treatments in each year). The number of nonannualized Types of facilities that closed and their effect on dialysis treatments per beneficiary remained steady at 115. beneficiaries’ access to care Over the most recent five-year period (2012 to 2017), the Each year, we assess the types of facilities that closed and number of FFS dialysis beneficiaries and total dialysis whether certain groups of Medicare dialysis beneficiaries treatments each increased by 1 percent per year, while are disproportionately affected by facility closures. Using the number of nonannualized treatments per beneficiary facilities’ claims submitted to CMS and CMS’s Dialysis declined from 116 to 115. The slight decline in per Compare database and the Medicare Provider of Services beneficiary treatment growth may be associated with: file, we compared the characteristics of beneficiaries • CMS’s restatement (in the rule-making process) of treated at facilities that closed in 2016 with those at its policy for paying for dialysis furnished more than facilities that provided dialysis in 2016 and 2017, the most thrice weekly (Centers for Medicare & Medicaid current years for which complete data are available. Services 2014). The agency said that facilities must Between 2016 and 2017, the number of dialysis treatment provide medical justification to be paid for furnishing stations—a measure of providers’ capacity—increased more than three dialysis treatments per week and that by 3 percent. There was a net increase in the number of the choice of dialysis modalities that require more facilities that were freestanding, for profit, and located in than three treatments per week does not constitute both urban and rural areas. Compared with facilities that medical justification. treated beneficiaries in both years, facilities that closed in In 2015, CMS’s contractors issued local coverage • 2016 (about 40 facilities) were more likely to be hospital determinations (LCDs) that required certain based, nonprofit, and smaller (as measured by the number conditions, including heart failure, to be reported on of dialysis treatment stations), which is consistent with dialysis facility claims for Medicare to cover and long-term trends in the supply of dialysis providers pay for dialysis treatments exceeding thrice weekly (Table 6-2). (Centers for Medicare & Medicaid Services 2018b). According to our analysis, few dialysis FFS beneficiaries • In 2017, CMS’s contractors issued draft LCDs that (roughly 1,600 individuals) were affected by facility would have covered and paid for dialysis treatments closures in 2016. Our analysis found that beneficiary more than thrice weekly only for acute conditions groups who were disproportionately affected included outside the patient’s plan of care; these LCDs have yet beneficiaries who were African American and younger to be finalized. (ages 45 to 64). By contrast, findings from our prior three analyses found that groups disproportionately affected In 2017, there was one fewer dialysis treatment • by closures included beneficiaries who were White and day (based on a thrice weekly treatment schedule) older (Medicare Payment Advisory Commission 2018c, compared with 2012. Medicare Payment Advisory Commission 2017, Medicare Payment Advisory Commission 2016b). However, less Use of most dialysis drugs has declined under the than 1 percent of FFS beneficiaries in these two groups outpatient dialysis PPS were affected by facility closures. Our analysis of claims When CMS broadened the payment bundle in 2011 to data suggests that beneficiaries affected by these closures include separately billable dialysis-related drugs, the obtained care elsewhere. agency set the PPS payment rate based on a per treatment basis using claims data from 2007. In 2014, to account Volume of services for the decline in dialysis drug use under the dialysis PPS, To assess changes in the volume of dialysis services, the statute required that CMS rebase the PPS base rate we examined recent trends in the number of dialysis by comparing drug use in 2007 with such use in 2012. treatments provided to beneficiaries and in the use of Subsequently, we examined changes between 2007 and injectable drugs administered during dialysis. March 2019 | Report to the Congress: Medicare Payment Policy 163

194 FIGURE Medicare margins... X-X FIGURE Use of dialysis drugs has declined under the outpatient dialysis PPS 6–1 100 Other drugs 90 Vitamin D agents 80 Iron agents ESAs 70 60 50 40 (in 2018 dollars) 30 Estimated per treatment use 20 10 0 2017 2015 2014 2013 2016 2011 2010 2012 2008 2009 2007 PPS (prospective payment system), ESA (erythropoiesis-stimulating agent). Dollars per treatment calculated by multiplying drug units reported on claims by 2018 Note: average sales price. Drugs included are epoetin alfa, epoetin beta, darbepoetin (ESAs); iron sucrose, sodium ferric gluconate, ferumoxytol, ferric carboxymaltose (iron agents); calcitriol, doxercalciferol, paricalcitol (vitamin D agents); daptomycin, vancomycin, alteplase, levocarnitine (all other drugs). Source: MedPAC analysis of 100 percent claims submitted by dialysis facilities to CMS. Note: Note and Source are in InDesign. Source: 2017 (the most current year for which complete data are declined by 23 percent per year. Most of this decline was available) in the use per treatment of the leading dialysis due to declining ESA use, which also fell by 23 percent drugs and aggregated them into four therapeutic classes— per year during the same period. For ESAs, some of this erythropoiesis-stimulating agents (ESAs), iron agents, decline may also have stemmed from clinical evidence Notes about this graph: 12 vitamin D agents, and antibiotics. showing that higher doses of these drugs led to increased The dialysis PPS • Data is in the datasheet. Make updates in the datasheet. risk of morbidity and mortality, which resulted in the Food increased the incentive for providers to be more judicious • WATCH FOR GLITCHY RESETS WHEN YOU UPDATE DATA!!!! and Drug Administration (FDA) changing the ESA label in providing dialysis drugs included in the payment in 2011. bundle. Under the prior payment method, dialysis drugs • The column totals were added manually. were paid according to the number of units of the drug • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. Between 2016 and 2017, holding price constant, the use of administered: In other words, the more units of a drug all dialysis drugs declined by nearly 4 percent. During this • I can’t delete the legend, so I’ll just have to crop it out in InDesign. provided, the higher the Medicare payment. period, drug use declined for each of the four therapeutic • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph classes (ESAs, vitamin D agents, iron agents, and all other As shown in Figure 6-1, most of the decline in the per default when you change the data. drugs) (Figure 6-1). As shown in Table 6-3, per treatment treatment use of dialysis drugs—which was estimated by drug use increased for only three products—ESAs epoetin multiplying drug units per treatment reported on CMS • Use paragraph styles (and object styles) to format. beta and darbepoetin alfa and vitamin D agent calcitriol. claims by each drug’s 2018 average sales price (to hold • Data was from: R:\Groups\MGA\data book 2007\data book 2007 chp1 However, under the PPS (between 2010 and 2017), per price constant)—occurred in the early years of the PPS treatment use of calcitriol declined. (implemented in 2011). For example, between 2010 and 2012, use per treatment across all therapeutic classes Outpatient dialysis services: Assessing payment adequacy and updating payments 164

195 TABLE Use of dialysis drugs per treatment has declined under the outpatient dialysis PPS 6–3 Mean units per treatment* Aggregate percent change Dialysis drug 2010 2016 2017 2010–2016 2016–2017 ESAs 1,269 –8% –76% Epoetin alfa 1,383 5,214 1.26 2.14 2.17 72 1 Darbepoetin alfa Epoetin beta** N/A 3.02 3.16 N/A 5 Iron agents –14 –26 0.11 0.13 0.15 Sodium ferric gluconate 16.0 13.0 12.4 Iron sucrose –22 –4 Ferumoxytol 0.8 0.0092 0.0073 –99 –20 N/A N/A –74 Ferric carboxymaltose 0.00008 0.00031 Vitamin D agents 0.3 –5 Paricalcitol 2.3 0.3 –87 –10 0.9 1.5 1.3 54 Doxercalciferol –64 0.05 63 0.03 0.13 Calcitriol Antibiotics Daptomycin 0.22 0.11 0.09 –58 –20 Vancomycin 0.02 0.02 0.01 –42 –14 Other drugs –88 0.001 0.001 –18 0.010 Levocarnitine 0.002 –89 0.002 Alteplase 0.020 –9 Note: PPS (prospective payment system), ESA (erythropoiesis-stimulating agent), N/A (not applicable). Individual units per treatment are rounded; the aggregate percentage change is calculated using unrounded units per treatment. *Each drug is reported using its own drug units. **Epoetin beta was introduced to the U.S. market in 2015. MedPAC analysis of claims submitted by dialysis facilities to CMS. Source: 2018 (Seeking Alpha 2018). With the FDA approval of Prior Commission analysis showed that the outpatient a biosimilar for epoetin alfa in 2018, competition among dialysis PPS increased price competition within the ESA ESA products could increase (and ESA costs for facilities and vitamin D therapeutic classes. For example, our could drop further) in the future (Pfizer 2018). analysis of ESA utilization since 2013 shows that dialysis facilities and nephrologists switched beneficiaries from Dialysis marginal profitability suggests incentive epoetin alfa to darbepoetin alfa or epoetin beta. In at to serve Medicare beneficiaries least one situation, switching was an explicit goal: One of the LDOs announced its intent to have more than 70 Another measure of access is whether providers have a percent of the company’s ESA patients (110,000 patients) financial incentive to expand the number of Medicare switched to epoetin beta (from epoetin alfa) by the end beneficiaries they serve. In considering whether to treat of the first quarter of 2016 (Reuters 2016). According a patient, a provider with excess capacity compares to several sources, the LDO reduced its total ESA costs the marginal revenue it will receive (i.e., the Medicare by switching beneficiaries to epoetin beta (Reuters payment) with its marginal costs—that is, the costs that 2016, Seeking Alpha 2016). A midsized chain recently vary with volume. If Medicare payments are larger than announced that between 85 percent and 90 percent of its the marginal costs of treating an additional beneficiary, a facilities will have switched to epoetin beta by the end of provider has a financial incentive to increase its volume of March 2019 | Report to the Congress: Medicare Payment Policy 165

196 Medicare patients. In contrast, if payments do not cover beneficiary declined from 1.7 admissions per beneficiary to 1.5 admissions per beneficiary, respectively. This the marginal costs, the provider may have a disincentive to 13 finding is consistent with the trend of declining inpatient care for Medicare beneficiaries. admissions for all Medicare FFS beneficiaries during this For dialysis facilities, in 2017 Medicare payments exceed period. USRDS data show that hospital admission rates marginal costs by 17 percent, a positive indicator of patient fell for ESRD-related complications and comorbidities access because it means facilities with available capacity (cardiovascular, infection, and vascular access events) have an incentive to treat Medicare beneficiaries. during the most recent five-year period for which data are available (2011 to 2016) (United States Renal Data System Quality of care 14 2018). Between 2012 and 2017, 30-day readmission Our analysis focuses on changes in quality indicators— rates declined slightly (from 22 percent of admissions to including mortality and morbidity, process measures 21 percent of admissions), while the proportion of dialysis that assess dialysis adequacy and anemia management, beneficiaries who used the ED increased from an average and treatment utilization (home dialysis and kidney of 11 percent per month to about 12 percent per month. transplantation rates). The analysis, except where Between 2011 and 2016, adjusted annual rates of mortality indicated, is based on the Commission’s analysis of per 100 dialysis beneficiaries declined from 18 to 16 Medicare FFS enrollment and claims data and CMS’s (United States Renal Data System 2018). monthly monitoring data between 2012 and 2017 and of Beneficiaries’ fluid management is related to factors such U.S. Renal Data System (USRDS) data between 2011 and as the adequacy of the dialysis procedure and dietary 2016. management. According to the Commission’s analysis, For the most recent five-year period that data are available, between 2012 and 2017, from 97 percent to 98 percent of rates of mortality and of hospitalization and readmission hemodialysis beneficiaries and 91 percent to 93 percent declined, while emergency department (ED) use rose. of PD beneficiaries received adequate dialysis, defined Use of home dialysis, which is associated with improved as having enough waste removed from their blood. patient satisfaction and quality of life, increased. However, Between 2012 and 2017, the share of dialysis beneficiaries home dialysis growth slowed between 2014 and 2017, diagnosed with dehydration declined slightly, while the partly because of a shortage of the solutions needed for the share of beneficiaries diagnosed with fluid overload predominant home method, peritoneal dialysis (PD). The increased slightly (Centers for Medicare & Medicaid negative cardiovascular outcomes associated with high Services 2018a). ESA use have generally declined or remained constant, Process and health outcome measures reflect the change in and blood transfusion use, which initially increased under anemia management under the PPS. Anemia is measured the PPS, has declined since 2013. by a blood test to check the level of hemoglobin, the protein In assessing quality, we also examine the multiple factors that carries oxygen in red blood cells. Median hemoglobin that affect access to kidney transplantation. This procedure levels fell during the initial years of the dialysis PPS; since is widely regarded as a better ESRD treatment option 2014, levels have remained steady at 10.5 g/dL. Figure 6-2 than dialysis in terms of patients’ clinical and quality of shows that the proportion of dialysis beneficiaries with life outcomes and Medicare spending, and demand far higher hemoglobin levels declined, and the proportion with outstrips supply. We also discuss CMS’s payment model— lower hemoglobin levels increased (which is generally the Comprehensive ESRD Care Model—that aims to associated with lower ESA use). During the initial years of improve the health outcomes of dialysis beneficiaries the dialysis PPS, blood transfusion rates increased (from while lowering the total Medicare Part A and Part B per 2.7 percent per month in 2010 to 3.4 percent per month in capita spending on these beneficiaries. Last, we discuss 2012). However, since 2013, the proportion of beneficiaries CMS’s two renal quality measurement systems, the ESRD receiving a blood transfusion declined (from 3.3 percent per Quality Incentive Program (QIP) and the dialysis star month to 2.3 per month) (Centers for Medicare & Medicaid 15 rating system. Services 2018a). Stroke, acute myocardial infarction, and heart failure Quality under the PPS are cardiovascular outcomes associated with anemia Between 2012 and 2017, through the Commission’s management. Under the dialysis PPS, the cumulative analysis of claims data, mean all-cause hospital stays per Outpatient dialysis services: Assessing payment adequacy and updating payments 166

197 FIGURE Title here share of beneficiaries experiencing stroke declined, while FIGURE x-x the share experiencing acute myocardial infarction has Changes in hemoglobin levels 6–2 under the dialysis PPS remained relatively constant. Until 2015, the share of beneficiaries with heart failure decreased. However, there 80 has been an increasing trend between 2015 and 2017 Hemoglobin level at or exceeding 12 g/dL 16 (Centers for Medicare & Medicaid Services 2018a). 70 Hemoglobin level 10 g/dL- 12 g/dL As discussed in our June 2014 report, clinical process 60 measures can exacerbate the incentives in FFS to Hemoglobin level under 10 g/dL 50 overprovide and overuse services (Medicare Payment Hemoglobin level under 10 g/dL Hemoglobin level 10–12 g/dL Advisory Commission 2014b). For example, before 40 Hemoglobin level at or exceeding12 g/dL 2011, targeting higher hemoglobin levels was associated with higher ESA use among dialysis beneficiaries. 30 In addition, some clinical process measures may be experiencing outcome 20 only weakly correlated with better health outcomes. A Share of dialysis beneficiaries given hemoglobin level may reflect adequate anemia 10 management for one patient, whereas the same level may lead to a different response in a different patient. 0 January January January January January January January January Focusing on clinical outcomes, such as rates of stroke, is 2017 2015 2011 2014 2010 2013 2016 2012 a better indicator of anemia management in the dialysis population. The Commission recently stated that quality PPS (prospective payment system), g/dL (grams per deciliter). Data are Note: measurement should be patient oriented, encourage compiled on a monthly basis by CMS. coordination, and promote delivery system change and MedPAC analysis of Medicare claims submitted by dialysis facilities. Source: that Medicare quality incentive programs should use a small set of population-based measures (e.g., outcomes, patient experience, value) to assess quality of care across settings and populations (Medicare Payment Advisory Note: Note and Source in InDesign. Commission 2018b). at http://medpac.gov/docs/default-source/reports/mar18_ According to separate analyses by CMS and the medpac_ch6_sec.pdf?sfvrsn=0). Commission, between 2012 and 2017 the share of beneficiaries dialyzing at home steadily increased from a Since 2014, one nonclinical factor—the availability of monthly average of 9.5 percent to 11.0 percent (Centers solutions needed to perform peritoneal dialysis—may have Notes about this graph: for Medicare & Medicaid Services 2018a). While we are affected the growth in home dialysis. Beginning around • Data is in the datasheet. Make updates in the datasheet. encouraged by this modest increase, differences by race September 2014, the growth in PD, the predominant home • I had to force return the items on the x-axis. They will reflow if I update the data. persist: African Americans are less likely to use home method, slowed because of a shortage of solutions needed methods. According to the Commission’s analysis, African to perform this type of dialysis. Between 2014 and 2017, • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. Americans account for 26 percent of home dialysis the total number of home dialysis patients increased by • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph beneficiaries compared with about 36 percent of all 3 percent per year; by contrast, between 2012 and 2014, default when you change the data. dialysis beneficiaries. the total number of home patients increased by 7 percent • Use paragraph styles (and object styles) to format. per year. The supply shortage resulted from the product’s Researchers have identified many factors that affect leading manufacturer (Baxter) experiencing increased PD the use of home dialysis, including factors both clinical demand and limited manufacturing capacity (Baxter 2014, (patients’ other health problems and prior nephrology Neumann 2014). Because of the shortage, beginning in care) and nonclinical (e.g., patients’ social circumstances, August 2014, the manufacturer gave each dialysis provider physician’s training and preference, dialysis facility’s staff an allocation for how many new patients could be started experience). The dialysis PPS is associated with an overall on PD based on the provider’s history of growth during the increase in the use of home dialysis (Lin et al. 2017). The first six months of 2014 (Seaborg 2015). Although steps Commission’s recent discussions of these factors can be have been taken to increase the supply of PD solutions, a found in our March 2018 report to the Congress (located shortage of solutions continues to exist for one of the two March 2019 | Report to the Congress: Medicare Payment Policy 167

198 Americans were less likely than White patients to receive TABLE kidney transplants despite their fourfold greater likelihood Between 2012 and 2017, 6–4 the number of kidney transplants of developing ESRD; however, between 2012 and 2017, increased, and African Americans, the number of African Americans receiving a transplant Hispanics, and Asian Americans grew by 5 percent per year (to 5,276 individuals, data not accounted for an increasing share shown). According to Ephraim and colleagues, compared with other groups, the lower rates of kidney transplantation 2017 2012 for African Americans have been associated with multiple 19,849 16,487 Total transplants factors, including immunological incompatibility with deceased donor kidneys, lower rates of referral for Share of live donors 34% 29% transplantation, lower rates of cadaver kidney donation, Share of: and lack of knowledge and suboptimal discussions about 52 Whites 47 kidney transplantation among recipients, their families, African Americans 25 27 and health care providers (Ephraim et al. 2012). 16 Hispanics 18 7 Asians 6 A new kidney allocation system implemented in 2014 by Others 2 2 the United Network for Organ Sharing led to a narrowing of the disparities in national kidney transplant rates Totals may not sum to 100 percent due to rounding. Note: among Whites, African Americans, and Hispanics on the transplant waiting list, according to a new analysis Organ Procurement and Transplantation Network 2018. Source: (Melanson et al. 2017). Under the new system, the starting point for calculating waiting time was changed from the date the patient was put on the waiting list to the earlier of either that date or the date the patient started regular dialysis treatments. The new system led PD types (automated peritoneal dialysis) in 2018 (Food to a substantial increase in the kidney transplant rate for and Drug Administration 2018). African Americans and Hispanics in the months following implementation and a decrease in the rate of kidney Access to kidney transplantation transplantation for Whites. Before the new system, the average monthly transplantation rate was significantly Kidney transplantation is widely regarded as a better higher among Whites (1.07 percent) compared with ESRD treatment option than dialysis in terms of patients’ African Americans or Hispanics (0.80 percent and 0.79 clinical and quality of life outcomes. In addition, percent, respectively). After implementation of the system, transplantation results in lower Medicare spending; in the monthly rates changed significantly for all groups: 0.95 2016, average Medicare spending for patients who had percent for Whites, 0.96 percent for African Americans, a functioning kidney transplant was less than half the and 0.91 percent for Hispanics (Melanson et al. 2017). spending for dialysis patients ($25,942 vs. $89,367) (United States Renal Data System 2018). However, Education efforts directed at patients can be effective demand for kidney transplantation exceeds supply. in encouraging them to make an informed decision Factors that affect access to kidney transplantation besides about their treatment, including home dialysis, in-center donation rates include the clinical allocation process; dialysis, kidney transplantation, and conservative care. patients’ health literacy, clinical characteristics, and For example, a recent review of educational interventions preferences; the availability of education for patients; found a strong association between patient-targeted clinician referral for transplant evaluation at a transplant dialysis modality education and choosing and receiving center; and transplant center policies. PD (Devoe et al. 2016). An augmented nurse care management program that targeted persons with late- Between 2012 and 2017, according to the Organ stage chronic kidney disease resulted in a statistically Procurement and Transplantation Network, the significant reduction in the number of hospitalizations number of kidney transplants increased by 4 percent during the intervention period and, for those who per year to 19,849 (Table 6-4) (Organ Procurement and Transplantation Network 2018). In 2017, African Outpatient dialysis services: Assessing payment adequacy and updating payments 168

199 required renal replacement therapy, higher use of PD or a clinical and financial (Part A and Part B) outcomes of preemptive kidney transplant (Fishbane et al. 2017). prospectively matched dialysis beneficiaries. Of the 13 ESCOs participating in the first round, 12 are operated In 2010, to help inform beneficiaries diagnosed with by Dialysis Clinic Inc., DaVita, and Fresenius Medical Stage IV CKD (chronic kidney disease), the disease Care, all of which CMS designated as large because stage before ESRD, about their treatment options and each organization operates more than 200 dialysis managing the disease and related comorbidities, MIPPA facilities; 1 ESCO is operated by Rogosin Institute, established Medicare payment for up to six sessions of which CMS designated as small because the company kidney disease education (KDE) per beneficiary. Since operates fewer than 200 dialysis facilities. For the second its implementation, relatively few beneficiaries have been performance round, 24 additional ESCOs joined the provided KDE services. About 3,500 beneficiaries were model. Of the 37 participating ESCOs in the second provided such services in each year between 2015 and round, 33 are operated by large organizations, while 4 2017, compared with about 4,200 beneficiaries in 2012. In are operated by small organizations—Rogosin, Centers 17 2017, Medicare KDE spending was under $500,000. for Dialysis Care, Atlantic Dialysis, and Northwest Kidney Centers. Enrollment in the CEC Model increased According to the Government Accountability Office, from approximately 16,000 beneficiaries in the first payment limitations on the providers who can furnish performance year (October 2015 to December 2016) to KDE services and the beneficiaries who are eligible might roughly 55,000 beneficiaries in the second performance constrain the service’s use (Government Accountability year (Centers for Medicare & Medicaid Services 2016, Office 2015). MIPPA specified the categories of providers Kalantar-Zadeh 2018). who can furnish KDE services—physicians, physician assistants, nurse practitioners, clinical nurse specialists, In the CEC Model’s first round, Dialysis Clinic Inc., 18 and certain providers of services in rural areas. MIPPA DaVita, and Fresenius—the ESCOs that CMS considers also specified that beneficiaries with Stage IV CKD large—were held to two-sided risk-based payment, while are eligible for the benefit. Some stakeholders contend Rogosin Institute, a small dialysis organization, was held that other categories of beneficiaries, including those to one-sided risk-based payment. (Under two-sided risk, with Stage V CKD (i.e., ESRD) who have not started the provider is at financial risk if specified goals are not dialysis as well as individuals who have already initiated achieved but is rewarded if the goals are met. Under one- hemodialysis, might also benefit from Medicare KDE sided risk, the provider is not penalized financially if goals coverage. are not met, but it does share in the gains.) In the CEC Model’s second round, small dialysis organizations have The Comprehensive ESRD Care Model the option of one-sided or two-sided risk. The relatively high resource use by dialysis beneficiaries, In payment year 1 (PY1) of the CEC Model, all 13 ESCOs particularly rates of hospital admissions and hospital produced savings relative to their benchmarks, with 12 readmissions, suggests that further improvements in quality ESCOs producing enough savings to earn shared savings are needed and that some dialysis beneficiaries might payments (Centers for Medicare & Medicaid Services benefit from better care coordination. Under the authority 2017). The earned shared savings payments ranged from of the Center for Medicare & Medicaid Innovation, the $1 million to $12 million and totaled $51 million. Quality first round of the Comprehensive ESRD Care (CEC) measurement in PY1 was essentially pay for reporting; Model began October 1, 2015, and will continue through thus, all the ESCOs received a 100 percent score for December 31, 2020. The CEC Model is testing whether quality. In total, the demonstration saved 1.7 percent a new payment model implemented in FFS Medicare can relative to a spending benchmark. It is not clear whether improve the outcomes of dialysis beneficiaries as well as this trend will continue since the results for 2017 and 2018 lower their Medicare per capita spending. A second round are not yet available. of the CEC Model began on January 1, 2017. According to CMS’s contractor, in the ESCOs’ first Under this five-year initiative, ESRD Seamless Care year, there was a statistically significant decline of $153 Organizations (ESCOs)—which, like accountable in total Part A and Part B spending per beneficiary per care organizations (ACOs), are specific to the dialysis p < 0.10) (Marrufo et al. 2017). The month (PBPM) ( population—consist of at least one dialysis facility and contractor attributed this reduction to a statistically one nephrologist, and they are held accountable for the March 2019 | Report to the Congress: Medicare Payment Policy 169

200 significant decline in spending for acute inpatient Conversely, nearly 10 percent of facilities assigned 4 or 5 stars had some QIP payment reduction. The correlation services (–$102 PBPM, p < 0.01) and post-acute care coefficient between a facility’s star rating and QIP score services (–$59 PBPM, p < 0.05). was 0.36, which means there is a positive but somewhat The Commission has said that, if structured properly, weak correlation between the two quality programs. a shared savings program—in this case, for ESRD providers—could present an opportunity to correct some Providers’ access to capital: Growth trends of the undesirable incentives inherent in FFS payment and indicate access is adequate reward providers who are doing their part to control costs Providers need access to capital to improve their and improve quality. equipment and open new facilities so they can accommodate the growing number of patients requiring In addition to the CEC Model, dialysis beneficiaries in dialysis. The two LDOs as well as other renal companies selected geographic areas also have access to ESRD appear to have adequate access to capital. For example, in special needs plans (SNPs). Between October 2017 and 2017 and 2018: October 2018, enrollment increased and the number of ESRD SNPs remained steady. As of October 2018, about Fresenius Medical Care took a $150 million stake in • 5,600 dialysis beneficiaries were enrolled in 15 ESRD the tissue engineering firm Humacyte Inc. and will SNPs operated by 6 managed care organizations in 9 become the exclusive distributor of the company’s states (Arizona, California, Colorado, Illinois, Nevada, bioengineered blood vessels once the FDA approves the New Jersey, New York, North Carolina, and Texas). By product. These blood vessels are currently being tested comparison, as of October 2017, about 4,600 dialysis in the last of three phases that are typically required for beneficiaries were enrolled in 15 ESRD SNPs operated market approval in the United States and Europe. by 6 managed care organizations in the same states with ESRD SNPs in 2018. While the CEC Model and • Vifor Fresenius Medical Care Renal Pharma—a joint ESRD SNPs enroll only dialysis beneficiaries, other venture between Fresenius Medical Care and Vifor ACO models, such as those participating in the Medicare Pharma Group—acquired the international license to Shared Savings Program, might provide opportunities Cara Therapeutics’ investigational opioid analgesic for beneficiaries with earlier stages of kidney disease that treats pruritus (severe itching) associated with to receive better care coordination, particularly in the renal disease in hemodialysis patients. Vifor Fresenius management of kidney disease risk factors. Medical Care Renal Pharma paid Cara Therapeutics $50 million in advance and will invest an additional The ESRD QIP and the dialysis star rating system $20 million in Cara common stock to market the drug in countries outside the United States, Japan, and CMS measures quality for each dialysis facility using South Korea. Cara will solely promote the product in two measurement systems, the ESRD QIP, which was facilities not operated by Fresenius Medical Care in mandated by MIPPA and implemented in 2012, and the United States. Vifor Fresenius Medical Care Renal the dialysis star rating system, which CMS established Pharma Ltd. and Cara will promote the investigational through a subregulatory process in 2015. CMS assigns medicine to Fresenius Medical Care dialysis clinics from 1 to 5 stars; more stars mean that a dialysis facility under a profit-sharing arrangement. performs better on quality measures compared with the national average. In its comment letter to CMS, the • DaVita completed its acquisition of Renal Ventures, Commission questioned why CMS finds a second quality gaining 31 dialysis facilities and divesting 7 facilities system necessary for dialysis facilities (Medicare Payment (as required by the Federal Trade Commission), and Advisory Commission 2014a). We also raised concerns acquired Purity Dialysis, which operates 10 facilities that beneficiaries and their families might be confused if in Wisconsin. In 2017, DaVita sold its subsidiary, a facility’s star rating and QIP score diverge, which could DaVita Medical Group, to Optum for $4.9 billion. occur because the measurement systems use different methods and measures to calculate a facility’s performance Baxter, a manufacturer of renal products including • score. For example, a Commission analysis found that peritoneal dialysis machines, and the Mayo Clinic in 2017, 30 percent of facilities assigned only 1 star did announced the development of a new renal care center not have a QIP payment reduction in that payment year. Outpatient dialysis services: Assessing payment adequacy and updating payments 170

201 of excellence that will be located at the Mayo Clinic’s Medicare payments and providers’ costs dialysis center in Jacksonville, FL. Each year, we examine the relationship between Medicare’s payments and providers’ costs as part of Dialyze Direct LLC, a provider of staff-assisted home • our assessment of payment adequacy. To make this hemodialysis services in skilled nursing facilities assessment, we reviewed Medicare expenditures for (SNFs), signed a definitive agreement to acquire outpatient dialysis services in 2017 and examined trends Affiliated Dialysis Centers LLC (a dialysis provider in in spending under the PPS. We also reviewed evidence the Midwest that furnishes outpatient clinic dialysis, regarding providers’ costs under the PPS. home dialysis, and SNF dialysis services through its associated entities, and currently serves more than 400 Medicare payments for outpatient dialysis services patients). In 2017, Medicare spending for outpatient dialysis services Cricket Health, a provider of integrated kidney care, • was $11.4 billion, an increase of 0.4 percent compared announced funding of $24 million that will be used with 2016. Per capita spending held steady at roughly partially to create new home and in-center dialysis $29,000 in 2016 and 2017. The trend in total and per programs. capita spending reflects two factors: (1) a statutory update (of 0.55 percent) to the base dialysis payment rate in 2017 • CVS Health announced an initiative that will focus on and (2) the number of dialysis treatments per beneficiary, the development of home dialysis technology. In 2018, which held steady in 2016 and 2017. the company plans to initiate a pivotal clinical trial to demonstrate the safety and efficacy of a new home Beginning in 2017, dialysis facilities are able to furnish hemodialysis device in support of a planned FDA dialysis to beneficiaries with acute kidney injury (AKI), submission to obtain market clearance. as mandated by the Trade Preferences Extension Act of 2015. In 2017, Medicare spending for outpatient dialysis • Outset Medical said it raised $132 million in services for beneficiaries with AKI was nearly $40 million. equity financing, with funds slated to accelerate the Medicare pays facilities the dialysis PPS base rate adjusted commercial expansion of its Tablo hemodialysis by the PPS wage index for the treatment of beneficiaries system. 20 with AKI. Medicare spending for treatment of AKI by dialysis facilities is not included in the Commission’s In public financial filings, the two LDOs (Fresenius analysis of Medicare’s payments and costs for dialysis Medical Care and DaVita) reported positive financial facilities. performance related to their dialysis business for 2018, including strong organic volume and revenue Part D spending for dialysis drugs growth—that is, growth achieved apart from mergers and Under the dialysis PPS, the use of dialysis drugs included acquisitions. In addition, since 2010, the two LDOs have in the PPS payment bundle declined. By contrast, during grown through large acquisitions and mergers of other this period, the use (as measured by Medicare spending) dialysis facilities and other health care organizations. For of Part D dialysis drugs that are not yet included in the example, during this period, both of the largest dialysis PPS payment bundle increased. In 2016 (the most recent organizations acquired midsized for-profit organizations: year for which data are available), Part D spending for two DaVita acquired Purity and Renal Ventures, and Fresenius categories of dialysis drugs (calcimimetics and phosphate Medical Care acquired Liberty Dialysis. binders) totaled $2.3 billion, an increase of 22 percent Another positive indicator of the dialysis sector’s strong per year compared since 2011. During this period, on a access to capital is its all-payer margin. Using cost report per treatment basis, Part D spending for all dialysis drugs data submitted to CMS by freestanding dialysis facilities, 21 increased by 20 percent per year. In addition, between we estimate that the 2017 all-payer margin was roughly 20 2011 and 2016, total Part D spending for dialysis drugs 19 percent. In their financial documents, dialysis providers grew more rapidly than spending for all other Part D reported that FFS Medicare payment rates are significantly drugs prescribed to dialysis beneficiaries (22 percent lower than commercial rates (DaVita 2018). per year vs. 11 percent per year). In 2016, spending for Part D dialysis drugs constituted about 60 percent of In general, current growth trends among dialysis providers dialysis beneficiaries’ gross Part D spending. Medicare indicate that the dialysis industry is attractive to for-profit facilities. March 2019 | Report to the Congress: Medicare Payment Policy 171

202 In February 2017, the FDA approved the first calcimimetic FIGURE FIGURE Reduction in drug... 6-5 injectable product (etelcalcetide) that is a counterpart Higher volume dialysis 6–3 facilities have lower cost per to oral cinacalcet (paid for under Part D in 2017). treatment, 2011–2017 Consequently, beginning January 2018, CMS pays for both the oral and intravenous calcimimetics under the 360 2017 dialysis PPS using a TDAPA until sufficient claims data 2011 340 (at least two years’ worth) for rate-setting analysis are 2014 2016 2016 available. (Additionally, Part D plans will no longer 320 2017 pay for oral cinacalcet for dialysis beneficiaries after 2014 2018). According to CMS, these products qualify for a 300 TDAPA because the base dialysis payment rate has not 2011 280 yet accounted for their costs. For these products, CMS is paying providers 106 percent of the drug’s average sales 260 price. 240 Including dialysis drugs covered under Part D in the dialysis PPS bundle may lead to better management of 220 drug therapy and improve beneficiaries’ access to these 200 medications since some beneficiaries lack Part D coverage Adjusted median cost per treatment (in dollars) or have coverage less generous than the Part D standard <3,000 ≥30,000 benefit. The efficiency of dialysis care may improve after 8,000–8,999 4,000–4,999 9,000–9,999 6,000–6,999 3,000–3,999 7,000–7,999 5,000–5,999 calcimimetics are included in the dialysis PPS payment 15,000–19,999 10,000–14,999 25,000–29,999 20,000–24,999 bundle. For example, based on the results of a multicenter, Number of dialysis treatments prospective, randomized, placebo-controlled trial, some clinicians concluded that the routine use of cinacalcet Cost per treatment is adjusted to remove differences in the cost of labor. Note: 22 Dialysis treatments include those paid for by all sources (not just Medicare- may not be warranted (Palmer et al. 2013). Between paid treatments). 2015 and 2016, Part D spending for cinacalcet increased Note: Note and Source in InDesign. Source: MedPAC analysis of cost reports submitted by freestanding dialysis 27 percent to roughly $875 million. Giving the Secretary facilities to CMS and the end-stage renal disease wage index files. the flexibility to rebase the payment bundle after oral- only dialysis drugs are included in the dialysis PPS payment bundle might lead to savings for beneficiaries and taxpayers. Notes about this graph: Providers’ costs for outpatient dialysis services spending for Part D dialysis drugs is not included in the • Data is in the datasheet. Make updates in the datasheet. under the outpatient dialysis PPS Commission’s analysis of Medicare’s payments and costs • I had to force return the items on the x-axis. They will reflow if I update the data. To assess the appropriateness of costs for dialysis services for dialysis facilities. paid for under the dialysis PPS, we examine whether • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. aggregate dialysis facility costs reflect costs that efficient In 2011, the Secretary included Part D oral-only dialysis • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph providers would incur in furnishing high-quality care. For drugs and biologics (calcimimetics and phosphate binders) default when you change the data. this analysis, we use 2016 and 2017 cost reports submitted in the expanded payment bundle but delayed paying to CMS by freestanding dialysis facilities. For those years, for them under the dialysis PPS until January 1, 2014 • Use paragraph styles (and object styles) to format. we look at the growth in the cost per treatment and how (to permit sufficient time to address data and pricing total treatment volume affects that cost. issues). The Stephen Beck Jr. Achieving a Better Life Experience Act of 2014 delayed bundling these drugs until Between 2016 and 2017, the Cost growth under the PPS 2025. However, if an injectable equivalent (or form of cost per treatment increased by 2 percent, from about $243 administration other than an oral form) of a dialysis oral- per treatment to nearly $248 per treatment. During this only drug is approved by the FDA before 2025, CMS will period, the cost per treatment for ESAs and other dialysis- include both the oral and non-oral dialysis drugs in the related drugs declined by 10 percent and 4 percent, PPS payment bundle (Centers for Medicare & Medicaid respectively. These cost categories accounted for 9 percent Services 2015). Outpatient dialysis services: Assessing payment adequacy and updating payments 172

203 and about 2 percent, respectively, of the total cost of for increased Medicare payment due to low volume had treatment in 2017. The decline in cost per treatment for substantially higher costs per treatment for capital and ESAs and other injectable drugs somewhat offset increases administrative and general services compared with all in the other cost categories: other facilities. Administrative and general expenses and capital costs, • Medicare margins for freestanding facilities in which accounted for 26 percent and 17 percent of the 2017 cost per treatment, respectively, increased by 5 percent The Commission assesses current payments and costs and 6 percent, respectively. for dialysis services for freestanding dialysis facilities by comparing Medicare’s payments with facilities’ • Labor costs, which accounted for about 33 percent of Medicare-allowable costs. The latest and most complete the cost per treatment, increased by 3 percent. data available on payments and costs are from 2017. We • Supply and lab costs, which accounted for 11 percent estimate that the aggregate Medicare margin in 2017 and 2 percent of the cost per treatment, respectively, was –1.1 percent (Table 6-5, p. 174). Margins decidedly increased by less than 1 percent and 2 percent, varied by treatment volume; facilities in the lowest volume respectively. quintile had margins at or below –21.3 percent, and facilities in the top volume quintile had margins of 5.4 Variation in cost growth across freestanding dialysis percent or more. facilities shows that some facilities were able to hold their cost growth well below that of others. For example, Urban facilities had higher margins than rural facilities between 2016 and 2017, per treatment costs decreased (–0.4 percent vs. –5.5 percent). Much of the difference in by 3 percent for facilities in the 25th percentile of cost margins between urban and rural facilities is accounted for growth and increased by 5 percent for facilities in the 75th by differences in total treatment volume. Urban dialysis percentile. facilities are larger on average than rural facilities in the number of treatment stations and total treatments provided. It is unknown to what extent some of the variation in costs In 2017, urban facilities averaged about 12,000 treatments, among facilities results from differences in the accuracy while rural facilities averaged about 7,800 treatments (data of facilities’ reported data. In 2016 and 2017, we found not shown). substantial variation in the level of selected cost categories reported by the five largest dialysis organizations. For The Commission is concerned about the gap in the example, the cost per treatment for administrative and Medicare margin between urban and rural facilities. general services and for capital services each differed by Although some rural facilities have benefited from the roughly $30 per treatment among these organizations. We dialysis PPS’s 23.9 percent low-volume adjustment and anticipate that CMS’s audit of a representative sample of 0.8 percent rural adjustment, the Commission has stated facilities’ ESRD cost reports will examine their accuracy. that neither adjustment targets low-volume, geographically Consistent with our 2014 recommendation, the Protecting isolated facilities that are critical to beneficiary access Access to Medicare Act of 2014 (PAMA) funded CMS to (Medicare Payment Advisory Commission 2016a, audit a representative sample of ESRD facility cost reports Medicare Payment Advisory Commission 2015, Medicare beginning in 2012. Payment Advisory Commission 2014a). In addition, the design of the low-volume adjustment provides facilities Cost per treatment is correlated with facility service with an adverse incentive to restrict their service provision volume Cost per treatment is correlated with the total to avoid reaching 4,000 treatments, the threshold that number of treatments a facility provides. For this CMS defines as a low-volume facility (Government analysis, we adjusted the cost per treatment to remove Accountability Office 2013). The text box (p. 175) differences in the cost of labor across areas and included provides more information about the low-volume and all treatments regardless of payer. Our analysis showed, rural payment adjustments used in the dialysis PPS. The in each year from 2011 through 2017, a statistically Commission intends to continue to monitor the adequacy significant relationship between total treatments and cost of Medicare’s payments for rural and urban facilities in per treatment (correlation coefficient equaled –0.5) (Figure the upcoming years. In addition, we intend to consider 6-3). That is, the greater the facility’s service volume, alternative approaches that would better target low- the lower its costs per treatment. Facilities that qualified volume, geographically isolated facilities. March 2019 | Report to the Congress: Medicare Payment Policy 173

204 TABLE Medicare margins in 2017 varied by type of freestanding dialysis facility 6–5 Percent of Percent of freestanding freestanding Medicare dialysis facility treatments dialysis facilities margin Provider type All –1.1% 100% 100% Urban –0.4 88 82 –5.5 Rural 12 18 Treatment volume (quintile) –21.3 20 7 Lowest –10.6 12 20 Second Third –3.4 20 17 24 Fourth 0.8 20 5.4 39 20 Highest Totals may not sum to 100 percent due to rounding. Note: Source: Compiled by MedPAC from cost reports and outpatient claims submitted by facilities to CMS and the Dialysis Compare database. Projecting the Medicare margin for 2019 How should Medicare payments change The aggregate Medicare margin for 2019 is projected to in 2020? be –0.4 percent, slightly greater than the 2017 Medicare margin (–1.1 percent). This projection considers providers’ PAMA sets the update to the outpatient dialysis payment historical cost growth and the following policy changes base rate equal to the ESRD market basket index, less that went into effect between 2017 (the year of our most an adjustment for productivity (currently estimated at recent margin estimates) and 2019: 0.5 percent). Based on CMS’s latest forecast of changes in the ESRD market basket costs for calendar year 2020 PAMA set the update to the dialysis base payment rate • (2.4 percent), the update to the 2020 payment rate would in 2018 to account for the reduced drug utilization be 1.9 percent. In addition to this statutory provision, the under the dialysis PPS. This rebasing adjustment ESRD QIP is expected to decrease total payments by 0.35 reduced the statutory update (based on the ESRD percent in 2020. And beginning in 2020, Medicare will market basket offset by a productivity adjustment) by pay dialysis facilities separately for all new drugs and 1.0 percent in 2018. The net payment update was 0.3 biologics based on the product’s average sales price for percent in 2018. at least a two-year period. This policy will likely increase In 2019, the statutory dialysis base payment rate • Medicare payments to facilities because CMS will not (based on the ESRD market basket offset by a offset the dialysis PPS base rate (even for new drugs that productivity adjustment) increased by 1.3 percent. fall into 1 of the 11 functional categories that are already included in the payment bundle). • For 2018 and 2019, CMS estimates that payments will be reduced by 0.14 percent and 0.15 percent, Recommendation respectively, due to the ESRD QIP. The evidence on payment adequacy suggests that outpatient • Other regulatory changes implemented by CMS are dialysis payments are adequate. It appears that facilities expected to result in payments increased by about 0.2 have become more efficient under the PPS, as measured by percent in 2018 and 0.3 percent in 2019. declining use of most injectable dialysis drugs. Outpatient dialysis services: Assessing payment adequacy and updating payments 174

205 The low-volume and rural payment adjustments should focus on protecting only facilities critical to beneficiary access Since 2016, all rural facilities, irrespective of their he 23.9 percent low-volume and 0.8 percent treatment volume or proximity to other dialysis rural payment adjustments under the dialysis facilities, receive an adjustment of 0.8 percent. prospective payment system (PPS) are not T Before 2016, the dialysis PPS did not include such an targeting facilities that are critical to beneficiary adjustment. The Commission is concerned that neither access. CMS defines a low-volume facility as one that the low-volume adjustment nor the rural adjustment are provides fewer than 4,000 treatments (Medicare and targeting facilities that are critical to beneficiary access. non-Medicare) in each of the three years before the A prior Commission analysis that used facility and payment year and has not opened, closed, or received a claims data from 2013 found that: new provider number because of a change in ownership during the three-year period. For payment year 2016, • About 47 percent of the facilities that receive the CMS revised the distance requirement used to determine low-volume adjustment are within five miles of the eligibility for this payment adjustment by (1) including, next closest facility. The median distance between for the purposes of determining a facility’s eligibility, the facility that would receive the proposed treatments furnished by the facility in question and other adjustment and the next closest facility is six miles. facilities under common ownership that are within five road miles of the facility in question; and (2) applying About 28 percent of all rural facilities are within • the five-mile distance criterion to all facilities, regardless five miles of the next closest facility, and nearly 20 of when the facility was certified. Before payment year percent of facilities located in rural areas are high 2016, the dialysis PPS used a 25-mile distance criterion volume (Medicare Payment Advisory Commission and applied that criterion to only facilities certified on or 2015). ■ after January 1, 2011. RECOMMENDATION 6 IMPLICATIONS 6 For calendar year (CY) 2020, the Congress should update Spending the CY 2019 Medicare end-stage renal disease prospective • In 2020, the statute sets the payment update at the payment system base rate by the amount determined in market basket, net of the productivity adjustment. The current law. Commission’s recommendation would have no effect on federal program spending relative to the statutory RATIONALE 6 update. Most of our indicators of payment adequacy are positive, Beneficiary and provider including beneficiaries’ access to care, the supply and We do not anticipate any negative effects on • capacity of providers, volume of services, quality of beneficiary access to care. This recommendation care, and access to capital. Providers have become more is expected to have a minimal effect on providers’ efficient in the use of dialysis drugs under the PPS. willingness and ability to care for Medicare The Medicare margin was –1.1 percent in 2017 and is beneficiaries. ■ projected to be –0.4 percent in 2019. The 17 percent marginal profit is a positive indicator of beneficiary access. March 2019 | Report to the Congress: Medicare Payment Policy 175

206 Endnotes 8 Currently, drugs and biologics reported on dialysis facility 1 beneficiaries In this chapter, the term refers to individuals claims are categorized into 1 of the following 11 functional covered by Medicare, and patients refers to all individuals categories: access management, anemia management, who have ESRD. bone and mineral metabolism, cellular management, Generally, individuals are fully insured under Social Security 2 antiemetic, anti-infective, antipruritic, anxiolytic, excess fluid if they have 40 credits of covered employment (i.e., the management, fluid and electrolyte management, and pain individual is employed in a job that pays Social Security management. taxes). Individuals are currently insured under Social Security 9 Over a five-year period ending in 2016 (the most recent data if they have a minimum of six credits of covered employment available), the number of dialysis patients with any type of in the three years before ESRD diagnosis. insurance coverage grew by 4 percent per year (United States Incidence data are adjusted for age, sex, and primary ESRD 3 Renal Data System 2018). diagnosis. 10 These figures are based on the Commission’s analysis of 4 For individuals entitled to Medicare based on ESRD, Medicare and total treatments reported by freestanding Medicare coverage does not begin until the fourth month facilities on cost reports submitted to CMS. after the start of dialysis, unless the individual had a kidney 11 Analysis of treatment growth is based on Medicare-covered transplant or began training for self-care, including dialyzing treatments in each year. An analysis of both Medicare-covered at home. and noncovered treatments finds that total treatments declined For example, the Center for Medicare & Medicaid Innovation 5 by 1 percent and the nonannualized dialysis treatments per awarded a three-year cooperative agreement in 2014 to beneficiary declined from 118 to 116 between 2016 and 2017. Northwell Health to implement the Healthy Transitions These drug classes accounted for nearly all dialysis drug 12 program for adults with late-stage CKD (with an estimated spending (about 97 percent) in 2010, the year before the start glomerular filtration rate of less than 30 ml/min) that aimed to of the new payment method. (1) better prepare patients for ESRD care by improving patient education and shared decision making; (2) increase the share If we approximate marginal cost as total Medicare costs 13 of patients who select home dialysis or a preemptive kidney minus fixed building and equipment costs, then marginal transplant; (3) increase the rate of arteriovenous fistulas; profit can be calculated as follows: Marginal profit = (4) increase patients’ quality of life scores; and (5) generate (payments for Medicare services – (total Medicare costs – savings to Medicare (e.g., by reducing hospitalizations and fixed building and equipment costs)) / Medicare payments. emergency department visits). CMS’s contractor concluded This comparison is a lower bound on the marginal profit that the health system was successful in implementing its because we do not consider any potential labor costs that are program (e.g., effectively delivered the intervention by using fixed. nurse case managers) (Schneider and Lines 2018). Due to too few treatment beneficiaries, the contractor does not anticipate 14 Between 2011 and 2016, adjusted hospitalization rates being able to conduct a rigorous impact analysis of this (per patient year) for hemodialysis patients fell from 0.49 program (Schneider and Lines 2018). Other providers have to 0.45 admissions for cardiovascular events, from 0.48 to developed similar interventions that emphasize early patient 0.44 for infection events, and from 0.21 to 0.13 admissions education and shared decision making (Dialysis Clinic Inc. for vascular access events. Adjusted admission rates for PD 2019, Kaiser Permanente 2017). patients also declined for these ESRD-related complications and comorbidities during this period (United States Renal Under the Bipartisan Budget Act of 2018, beginning January 6 Data System 2018). 2019, clinicians who manage home dialysis beneficiaries can furnish their visits through telehealth (rather than in person). 15 Blood transfusions are of concern to patients because they (1) Beneficiaries are required to receive at least a monthly face- carry a small risk of transmitting blood-borne infections to to-face visit for the first three months of home dialysis and the patient, (2) may cause some patients to develop a reaction, once every three months thereafter. and (3) are costly and inconvenient for patients. Blood transfusions are of particular concern for patients seeking For pediatric dialysis beneficiaries (less than 18 years of age), 7 kidney transplantation because they increase a patient’s the base rate is adjusted for age and type of dialysis. alloantigen sensitization, which can require a patient to wait to receive a transplant. Outpatient dialysis services: Assessing payment adequacy and updating payments 176

207 16 According to CMS, the increasing cumulative share of In addition, for beneficiaries with AKI, Medicare pays dialysis 20 beneficiaries with heart failure beginning in 2015 could be facilities separately for drugs, biologics, and laboratory associated with the issuance of local coverage determinations services that are not renal dialysis services. in that year by CMS’s contractors that required certain 21 Part D spending per dialysis treatment is calculated by conditions, including heart failure, to be reported on dialysis dividing total Part D spending for dialysis drugs by the total facility claims for Medicare to cover dialysis treatments number of Part B dialysis treatments furnished by dialysis exceeding thrice weekly (Centers for Medicare & Medicaid facilities to Medicare beneficiaries with and without Part D. Services 2018b). 22 The Evaluation of Cinacalcet Hydrochloride Therapy 17 This analysis used 100 percent of 2012 through 2017 carrier to Lower Cardiovascular Events trial—a multicenter, and outpatient claims submitted for KDE services. prospective, randomized, placebo-controlled trial—found 18 MIPPA does not permit other providers (such as registered that cinacalcet did not significantly reduce the risk of death nurses, social workers, and dieticians) or dialysis facilities to or major cardiovascular events in patients with moderate to bill for KDE services. severe secondary hyperparathyroidism undergoing dialysis (Chertow et al. 2012). In 2018, both LDOs and several midsized organizations 19 contributed more than $100 million to defeat a public referendum in California that would have capped payments at 15 percent above patient care costs for dialysis patients with commercial coverage. March 2019 | Report to the Congress: Medicare Payment Policy 177

208 References DaVita. 2018. Annual report pursuant to section 13 or 15(d) of Baxter. 2014. Letter from Richard Marritt, Region Head the Securities Exchange Act of 1934 for the fiscal year ended and General Manager, US and Canada. July 28. http://www. December 31, 2017. February 23. homebybaxter.com/Documents/USMP%20MG2%2014-0093i_ Customer%20Letter%20(Update)_FINAL.pdf. Devoe, D. J., B. Wong, M. T. James, et al. 2016. Patient education and peritoneal dialysis modality selection: A systematic review Centers for Medicare & Medicaid Services, Department of Health 68, no. American Journal of Kidney Diseases and meta-analysis. and Human Services. 2018a. ESRD prospective payment system 3 (September): 422–433. (ESRD PPS) overview of 2011–2017 claims-based monitoring system. Q4 2017 Public Use File. https://www.cms.gov/Medicare/ Dialysis Clinic Inc. 2019. Comprehensive end-stage renal disease Medicare-Fee-for-Service-Payment/ESRDpayment/ESRD- care model and future payment models. https://www.dciinc.org/ Claims-Based-Monitoring.html. newsroom/comprehensive-end-stage-renal-disease-care-model- and-future-payment-models/. Centers for Medicare & Medicaid Services, Department of Health and Human Services. 2018b. Medicare program; end-stage renal Ephraim, P. L., N. R. Powe, H. Rabb, et al. 2012. The providing disease prospective payment system, payment for renal dialysis resources to enhance African American Patients’ Readiness to services furnished to individuals with acute kidney injury, End- Make Decisions about Kidney Disease (PREPARED) study: Stage Renal Disease Quality Incentive Program, Durable Medical BMC Nephrology Protocol of a randomized controlled trial. 13: Equipment, Prosthetics, Orthotics and Supplies (DMEPOS) 135. Competitive Bidding Program (CBP) and fee schedule amounts, and technical amendments to correct existing regulations related Fishbane, S., S. Agoritsas, A. Bellucci, et al. 2017. Augmented 83, to the CBP for certain DMEPOS. Final rule. Federal Register nurse care management in CKD Stages 4 to 5: A randomized no. 220 (November 14): 56922–57073. trial. American Journal of Kidney Diseases 70, no. 4 (October): 498–505. Centers for Medicare & Medicaid Services, Department of Health and Human Services. 2017. Medicare Comprehensive End Stage Food and Drug Administration. 2018. FDA drug shortages. Renal Disease Care Model performance year 1 (Oct 2015–Dec https://www.accessdata.fda.gov/scripts/drugshortages/dsp_ 2016) results. https://innovation.cms.gov/Files/x/cec-fncl-py1.pdf. ActiveIngredientDetails.cfm?AI=Peritoneal%20Dialysis%20 Solutions&st=c. Centers for Medicare & Medicaid Services, Department of Health and Human Services. 2016. Comprehensive ESRD (CEC) Model. Government Accountability Office. 2015. End-stage renal https://innovation.cms.gov/Files/slides/cec-appkickoff-slides.pdf. disease: Medicare payment refinements could promote increased Washington, DC: GAO. use of home dialysis. Centers for Medicare & Medicaid Services, Department of Health and Human Services. 2015. Medicare program; end-stage End-stage renal Government Accountability Office. 2013. renal disease prospective payment system, and Quality Incentive disease: CMS should improve design and strengthen monitoring 80, no. 215 (November 16): Federal Register Program. Final rule. of low-volume adjustment. Report 13–287. Washington, DC: 68967–69077. GAO. Centers for Medicare & Medicaid Services, Department of Health Kaiser Permanente. 2017. Regaining control when facing kidney and Human Services. 2014. Medicare program; end-stage renal failure. https://permanente.org/regaining-control-when-facing- disease prospective payment system, Quality Incentive Program, kidney-failure/. and durable medical equipment, prosthetics, orthotics, and 79, no. 215 (November 6): supplies. Final rule. Federal Register Kalantar-Zadeh, K. 2018. Is ESCO the future of dialysis care? 66120–66265. Renal & Urology News , October 12. Chertow, G. M., G. A. Block, R. Correa-Rotter, et al. 2012. Effect Lin, E., X. S. Cheng, K. K. Chin, et al. 2017. Home dialysis in the of cinacalcet on cardiovascular disease in patients undergoing Journal of the American Society prospective payment system era. New England Journal of Medicine dialysis. 367, no. 26 28, no. 10 (October): 2993–3004. of Nephrology (December 27): 2482–2494. Outpatient dialysis services: Assessing payment adequacy and updating payments 178

209 Neumann, M. 2014. Baxter, FMC sign supply agreement; ISPD Marrufo, G., B. Negrusa, D. Ullman, et al. 2017. Nephrology News & ‘deeply concerned’ about PD fluid shortage. Comprehensive EndStage Renal Disease Care (CEC) Model: , September 3. Issues Report prepared Performance Year 1 annual evaluation report. by the Lewin Group for the Centers for Medicare & Medicaid Organ Procurement and Transplantation Network. 2018. National Services. Baltimore, MD: CMS. data. https://optn.transplant.hrsa.gov/data/view-data-reports/ national-data/. Medicare Payment Advisory Commission. 2018a. Comment letter to CMS on the end-stage renal disease prospective payment Palmer, S. C., I. Nistor, J. C. Craig, et al. 2013. Cinacalcet in system and Quality Incentive Program proposed rule, August 31. patients with chronic kidney disease: A cumulative meta-analysis 10, no. 4: of randomized controlled trials. PLoS Medicine Medicare Payment Advisory Commission. 2018b. Report to e1001436. the Congress: Medicare and the health care delivery system. Washington, DC: MedPAC. Pfizer. 2018. Pfizer’s biosimilar Retacrit® (epoetin alfa-epbx) approved by U.S. FDA. Press release. https://www.pfizer.com/ Medicare Payment Advisory Commission. 2018c. Report to the news/press-release/press-release-detail/pfizer_s_biosimilar_ Congress: Medicare payment policy. Washington, DC: MedPAC. retacrit_epoetin_alfa_epbx_approved_by_u_s_fda. Medicare Payment Advisory Commission. 2017. Report to the Reuters. 2016. FMC aims for 110,000 U.S. patients to be on Congress: Medicare payment policy. Washington, DC: MedPAC. Mircera in first-quarter. February 24. http://www.reuters.com/ Medicare Payment Advisory Commission. 2016a. Comment article/us-fresenius-medi-results-mircera-idUSKCN0VX11F. letter on CMS’s proposed rule on the ESRD prospective payment Schneider, C., and L. Lines, National Institute of Diabetes system, July 29. and Digestive and Kidney Diseases. 2018. HCIA round two Medicare Payment Advisory Commission. 2016b. Report to the . Cambridge, MA: Mathematica evaluation: Northwell Health Washington, DC: MedPAC. Congress: Medicare payment policy. Policy Research. Medicare Payment Advisory Commission. 2015. Comment letter Seaborg, E. 2015. Peritoneal dialysis fluid shortage disrupted to CMS on the ESRD prospective payment system and the ESRD Kidney News , March. growth of popular therapy. Quality Incentive Program, August 6. Seeking Alpha. 2018. American Renal Associates (ARA) CEO, Medicare Payment Advisory Commission. 2014a. Comment Joe Carlucci on Q3 2018 results—Earnings call transcript. letter to CMS on the end-stage renal disease prospective payment https://seekingalpha.com/article/4220467-american-renal- system and Quality Incentive Program proposed rule, August 15. associates-ara-ceo-joe-carlucci-q3-2018-results-earnings-call- transcript?part=single. Medicare Payment Advisory Commission. 2014b. Report to the Congress: Medicare and the health care delivery system. Seeking Alpha. 2016. Fresenius Medical Care AG & Co. KGaA Washington, DC: MedPAC. (FMS) Rice Powell on Q4 2015 results—Earnings call transcript. February 24. https://seekingalpha.com/article/3926506-fresenius- Report to the Medicare Payment Advisory Commission. 2001. medical-care-ag-and-co-kgaa-fms-rice-powell-q4-2015-results- Washington, DC: MedPAC. Congress: Medicare payment policy. earnings-call. Report to the Medicare Payment Advisory Commission. 2000. United States Renal Data System, National Institute of Diabetes Congress: Medicare payment policy. Washington, DC: MedPAC. USRDS 2018 annual and Digestive and Kidney Diseases. 2018. data report. Bethesda, MD: NIDDK. Melanson, T. A., J. M. Hockenberry, L. Plantinga, et al. 2017. New kidney allocation system associated with increased rates of Health Affairs transplants among Black and Hispanic patients. 36, no. 6 (June 1): 1078–1085. March 2019 | Report to the Congress: Medicare Payment Policy 179

210

211 CHAPTER 7 Cross-cutting issues in post-acute care

212

213 CHAPTER 7 Cross-cutting issues in post-acute care Chapter summary In this chapter Post-acute care (PAC) providers offer important recuperation and • Medicare’s payments rehabilitation services to Medicare beneficiaries, about half of whom had a remain high, and revisions prior hospital stay. PAC providers include skilled nursing facilities (SNFs), to the SNF and HHA home health agencies (HHAs), inpatient rehabilitation facilities (IRFs), and payment systems need to be implemented long-term care hospitals (LTCHs). In 2017, fee-for-service (FFS) program spending on PAC services totaled $58.5 billion. • Quality measures should focus on claims-based The Commission has previously discussed the challenges to increasing outcome measures the accuracy of Medicare’s payments and overcoming the shortcomings of the separate FFS payment systems for PAC (Medicare Payment Advisory • Conclusion Commission 2018, Medicare Payment Advisory Commission 2017, Medicare Payment Advisory Commission 2015, Medicare Payment Advisory Commission 2014). Over more than a decade, the Commission has worked extensively on PAC payment reform, pushing for closer alignment of costs and payments and more equitable payments across different types of patients. Despite some actions by the Secretary and the Congress, Medicare’s payments remain too high relative to the costs of treating beneficiaries in three of the four settings (SNF, HHA, and IRF). After years of research and recommendations by the Commission, the Secretary is poised to make substantial changes to the designs of the prospective payment systems March 2019 | Report to the Congress: Medicare Payment Policy 183

214 (PPSs) Medicare uses to pay HHAs and SNFs. These changes are overdue and are consistent with longstanding recommendations made by the Commission. The Commission has two goals in making payment recommendations. The update recommendations aim to ensure that aggregate payments are adequate so that beneficiary access is preserved while taxpayers and the long-run sustainability of the program are protected. The recommendations to revise the payment systems aim to align program payments with the costs of treating patients with different care needs. Such targeting increases the equity of the program’s payments, thereby minimizing the financial incentive for providers to treat some beneficiaries over others. A uniform payment system for all PAC would increase the equity of payments across patients and providers in all settings, but its implementation is on a longer timetable. Until a unified PAC PPS is in place, Medicare must continue to improve its setting-specific payment systems. FFS Medicare continues to overpay for PAC services; moreover, the current HHA and SNF payment systems also create inequities across patients with different care needs and the providers that treat them. Furthermore, the overpayments and misalignments affect the benchmarks for Medicare Advantage plans and alternative payment models. On the quality front, there has been progress on defining common outcome measures across PAC providers and establishing value-based purchasing policies for HHAs (on a demonstration basis) and for SNFs. However, the Commission is increasingly concerned that trends in some provider-reported quality measures raise questions about the accuracy and reliability of this information. The Commission has work underway to examine the accuracy of the patient assessment–based quality measures. ■ Cross-cutting issues in post-acute care 184

215 and medically complex care, the Commission has Medicare’s payments remain high, and recommended redesigns of the HHA and SNF payment revisions to the SNF and HHA payment systems (in 2011 and 2008, respectively), which together pay for almost 80 percent of Medicare PAC stays. systems need to be implemented Medicare margins for three of the PAC settings (HHA, Post-acute care (PAC) providers offer important SNF, and IRF) have been above 10 percent for most of recuperation and rehabilitation services to Medicare the past 10 years (Figure 7-1). In each setting, Medicare beneficiaries, about half of whom had a prior hospital stay. margins increased substantially soon after a prospective PAC providers include skilled nursing facilities (SNFs), payment system (PPS) was implemented, indicating that home health agencies (HHAs), inpatient rehabilitation the initial base rates for each setting were too high and that facilities (IRFs), and long-term care hospitals (LTCHs). providers rapidly adjusted to the new payment rules. In 2017, fee-for-service (FFS) program spending on PAC services totaled $58.5 billion. Medicare margins for HHAs and SNFs have been especially high, even after rebasing and productivity and Since 2008, the Commission has made recommendations other payment adjustments mandated by the Congress. to lower the level of program spending in each of the Over the last decade, Medicare margins in HHAs and PAC settings by eliminating annual updates to payment SNFs averaged over 15 percent. Close behind, IRF rates, lowering payments below current levels, or both. margins averaged 11.1 percent. The average margin for To redistribute payments more equitably between therapy all LTCHs has been considerably lower, though higher for FIGURE Title here... FIGURE 1-XX Medicare margins have remained high for most post-acute care providers 7–1 25 20 15 10 5 HHA SNF Medicare margin (in percent) IRF 0 LTCH LTCH with ≥85% qualifying cases –5 2009 2010 2006 2008 2007 2017 2016 2015 2014 2013 2012 2011 Note: HHA (home health agency), SNF (skilled nursing facility), IRF (inpatient rehabilitation facility), LTCH (long-term care hospital). Medicare margin is calculated as (Medicare payments – Medicare costs)/Medicare payments. The Pathway to SGR Reform Act of 2013 established separate payment methodologies in cases that qualify as LTCH discharges and cases that do not. To qualify as an LTCH discharge, the stay either must have been immediately preceded by an acute care hospital stay that included at least three days in an intensive care unit or have had an LTCH principal diagnosis indicating prolonged mechanical ventilation. We did not calculate margins for LTCH-qualifying discharges before 2012. Source: MedPAC analysis of Medicare cost reports 2006–2017. Note: Note and Source are in InDesign. Report to the Congress: Medicare Payment Policy | March 2019 185 Source: Notes about this graph: • Data is in the datasheet. Make updates in the datasheet. • I deleted the years from the x-axis and put in my own. • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. • The dashed line looked ok here, so I didn’t hand draw it. • I can’t delete the legend, so I’ll just have to crop it out in InDesign. • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph default when you change the data. • Use paragraph styles (and object styles) to format.

216 TABLE Commission’s payment recommendations since 2008 7–1 Years the Commission made the recommendation IRF SNF HHA Recommended action LTCH No update (0 percent update) 2008–2011; 2008–2016 2008–2016 2009–2018 2016–2018 2017–2018 2009–2018 2012–2015 Lower payments 2011–2018 Revise the payment system design 2008–2018 Note: SNF (skilled nursing facility), HHA (home health agency), IRF (inpatient rehabilitation facility), LTCH (long-term care hospital). The table shows the years the recommendation was made by the Commission. A year can appear in the 0 percent update and lower payment categories because a recommendation covered multiple years, with a 0 percent update in one year and reductions in one or more subsequent years. providers with at least 85 percent of stays that meet the in calendar year 2020, and CMS plans to overhaul the new criteria to qualify to receive LTCH PPS payments. SNF PPS in fiscal year 2020. Both redesigns are consistent with the Commission’s recommended changes and would Because the level of program payments has been rebalance payments between therapy cases and medically high relative to the cost of treating beneficiaries, the complex cases. By increasing the equity of program Commission has recommended lowering and/or freezing payments, providers will have less financial incentive to Medicare’s payment rates for PAC for many years (Table favor admitting beneficiaries with certain care needs over 7-1). For HHAs, SNFs, and IRFs, the Commission other beneficiaries. The Commission urges the Secretary recommended no updates (0 percent updates) or lower to proceed with these planned reforms. payments each year since 2008 and for LTCHs since 2009. In some years, the Commission made a multiyear recommendation that included no update to payment rates in one year and reductions in subsequent years. Yet during Quality measures should focus on this period, without congressional action, SNF, IRF, and claims-based outcome measures LTCH payments were increased by statutory updates. For HHAs, although the Patient Protection and Affordable Since 1999, the Commission has called for a variety of Care Act of 2010 calls for annual rebasing of payments, quality initiatives, including the collection of uniform the mandated reductions have been offset by updates to patient assessment information, the reporting of outcome- payment rates and consequently have not gone nearly far based quality measures that focus on the key goals of enough in realigning payments to costs. PAC, and the implementation of value-based purchasing policies. The Congress and CMS have acted on many The Commission also recommended revising the of the Commission’s recommendations, including the payment systems for HHAs (in 2011) and SNFs (in development and collection of uniform patient assessment 2008) to increase the equity of program payments. The items, outcome-based quality measures, and value- Commission is pleased that the Secretary is poised based purchasing for HHAs and SNFs. To meet the to implement changes to the HHA and SNF PPSs requirements in the Improving Medicare Post-Acute that will base payments on the clinical and functional Care Transformation Act of 2014, CMS has undertaken characteristics of patients, not on the amount of therapy the development of measures of function and cognition, furnished. The Bipartisan Budget Act of 2018 requires skin integrity, Medicare spending per beneficiary, CMS to implement major changes to the home health PPS Cross-cutting issues in post-acute care 186

217 with other IRFs for like patients. The Commission is discharge to community, hospital readmissions, medication reconciliation, and incidence of major falls. concerned that when provider-reported patient assessment information affects a provider’s payments, providers The Commission has raised concerns that not all of the measures are outcome based or uniformly defined across respond inappropriately to these financial incentives. the settings, though such refinements may be made in the Given these disturbing trends, the Commission is future. increasingly wary of the accuracy of the provider-reported patient assessment information. The Commission has work Because the maintenance of and improvement in function are key goals of PAC, the Commission recommended underway to assess these data. Although these data are the development of uniform patient assessment items important for measuring patient outcomes and establishing across the four PAC settings. Information on a patient’s care plans, they may not be key to establishing accurate functional status, cognitive status, and changes in function payments. Our initial work on a unified PAC PPS found that payments could be accurate without measures of are used to establish care plans for patients, risk adjust payments, and measure quality of care. The HHA, SNF, patient function. The Commission will continue its work and IRF PPSs use patient assessment data to define the on design elements of a PAC PPS, including whether function is a necessary component of a case-mix system. case-mix groups that establish payments for most of the patient groups cared for. In addition, the HHA value-based purchasing demonstration uses measures of function to calculate provider performance. Conclusion Because patient assessment information affects payments As evidenced by years of high Medicare margins, the and quality results, it is important that it consistently and program is paying more for services than is warranted. accurately reflects patients’ levels of function. However, Further, its payment systems unfairly advantage some the use of this information to set payments and measure providers and encourage the admission of patients with and reward quality creates incentives for providers to certain care needs over others. Because FFS payment rates report it in ways that boost payments. Over time, we have form the basis of Medicare Advantage benchmarks and become increasingly concerned about the validity and accountable care organization targets, the overpayments utility of provider-reported patient assessment information. also affect non-FFS payment models and their success. Our recent analyses of provider-reported measures From the taxpayers’ perspective, unnecessarily high calculated from patient assessment information have raised payments contribute to the projected insolvency of the concerns that information gathered from these sources Hospital Insurance Trust Fund (2026). The Secretary plans may not be accurate. For example, on average, HHAs to implement long overdue changes to the SNF (in fiscal have reported considerable improvement over the course year 2020) and HHA (in calendar year 2020) PPSs. The of an episode in patients’ abilities to conduct activities of Commission urges the Secretary to follow through with daily living (such as walking and transferring). Yet, during these plans. the same time period, there was little or no improvement in claims-based measures (such as hospitalization and Until the implementation of a unified PAC PPS, Medicare 1 emergency room use). These divergent trends raise must continue to improve its setting-specific payment questions about the accuracy of the provider-reported systems so that is does not overpay for services and create information. In IRFs, where lower function at admission inequities that can affect beneficiaries’ access to care. ■ translates into higher payments, we found that high-margin IRFs appear to record lower patient function compared March 2019 | Report to the Congress: Medicare Payment Policy 187

218 Endnotes 1 We would expect similar trends in the provider-reported and claims-based measures. Studies have found that functional status is related to hospitalization rates and the use of emergency departments (Laudisio et al. 2015, Middleton et al. 2018, Slocum et al. 2015, Soley-Bori et al. 2015). Cross-cutting issues in post-acute care 188

219 References Middleton, A., J. E. Graham, and K. J. Ottenbacher. 2018. Laudisio, A., E. Marzetti, F. Franceschi, et al. 2015. Disability Functional status is associated with 30-day potentially preventable is associated with emergency room visits in the elderly: A hospital readmissions after inpatient rehabilitation among aged population-based study. Aging Clinical and Experimental Medicare fee-for-service beneficiaries. Archives of Physical Research 27, no. 5 (October): 663–671. 99, no. 6 (June): 1067–1076. Medicine and Rehabilitation Medicare Payment Advisory Commission. 2018. Report to the Slocum, C., P. Gerrard, R. Black-Schaffer, et al. 2015. . Washington, DC: MedPAC. Congress: Medicare payment policy Functional status predicts acute care readmissions from inpatient Medicare Payment Advisory Commission. 2017. Report to the 10, no. 11: PLoS One rehabilitation in the stroke population. Congress: Medicare payment policy . Washington, DC: MedPAC. e0142180. Medicare Payment Advisory Commission. 2015. Report to the Soley-Bori, M., R. Soria-Saucedo, C. M. Ryan, et al. 2015. Congress: Medicare payment policy . Washington, DC: MedPAC. Functional status and hospital readmissions using the Medical Expenditure Panel Survey. Journal of General Internal Medicine Medicare Payment Advisory Commission. 2014. Report to the 30, no. 7 (July): 965–972. Washington, DC: MedPAC. Congress: Medicare payment policy. March 2019 | Report to the Congress: Medicare Payment Policy 189

220

221 CHAPTER 8 Skilled nursing facility services

222 RECOMMENDATIONS The Secretary should proceed to revise the skilled nursing facility prospective payment 8-1 system in fiscal year 2020 and should annually recalibrate the relative weights of the case- mix groups to maintain alignment of payments and costs. COMMISSIONER VOTES: YES 17 • NO 0 • NOT VOTING 0 • ABSENT 0 ... 8-2 The Congress should eliminate the fiscal year 2020 update to the Medicare base payment rates for skilled nursing facilities. COMMISSIONER VOTES: YES 17 • NO 0 • NOT VOTING 0 • ABSENT 0

223 CHAPTER 8 Skilled nursing facility services In this chapter Chapter summary Skilled nursing facilities (SNFs) provide short-term skilled nursing and • Are Medicare payments rehabilitation services to beneficiaries after a stay in an acute care hospital. In adequate in 2019? 2018, about 15,000 SNFs furnished 2.3 million Medicare-covered stays to 1.6 How should Medicare • million fee-for-service (FFS) beneficiaries. Medicare FFS spending on SNF payments change in 2020? services was $28.4 billion in 2017, about 1 percent less than in 2016. Just over 4 percent of beneficiaries used SNF services. Medicaid trends • Assessment of payment adequacy To examine the adequacy of Medicare’s payments, we analyze beneficiaries’ access to care (including the supply of providers and volume of services), quality of care, provider access to capital, and Medicare payments in relation to providers’ costs to treat Medicare FFS beneficiaries. Most indicators of the adequacy of Medicare’s payments are positive. Beneficiaries’ access to care— Access to SNF services remains adequate for most beneficiaries. • The number of SNFs participating Capacity and supply of providers— in the Medicare program has been stable. The vast majority (89 percent) of beneficiaries live in a county with three or more SNFs or swing bed facilities (rural hospitals with beds that can serve as either SNF beds or acute care beds), and less than 1 percent live in a county without one. March 2019 | Report to the Congress: Medicare Payment Policy 193

224 Between 2016 and 2017, the median occupancy rate declined slightly but remained high (about 85 percent). • Volume of services— Medicare-covered admissions per FFS beneficiary decreased 2 percent between 2016 and 2017, consistent with a decrease in the number of admissions for hospital stays that last at least three days (required for Medicare coverage). Lengths of stay also declined by 2 percent. Both contributed to fewer covered days in 2017 compared with 2016. Lower SNF use reflects the growing presence of alternative payment models, not the adequacy of Medicare’s payments. • Marginal profit— An indicator of whether freestanding SNFs have an incentive to treat more Medicare beneficiaries—marginal profit—averaged 19.1 percent for freestanding facilities in 2017. Quality of care— Since 2011, SNF quality measures have shown mixed performance. The average rate of discharge to the community increased and the average rate of readmission during the SNF stay improved, the average rate of readmissions after the SNF stay worsened, and the measures of mobility remained the same. Changes in the measures between 2016 and 2017 were similarly mixed. Because most SNFs are part of nursing homes, we Providers’ access to capital— examine nursing homes’ access to capital. Despite relatively low total margins (a measure of the total financial performance across all payers and lines of business), lending and investment activities remain robust. Access to capital was adequate in 2018 and is expected to remain so in 2019. Lending wariness reflects broad changes in post-acute care, not the adequacy of Medicare’s payments. Medicare is regarded as a preferred payer of SNF services. Medicare payments and providers’ costs— Medicare’s spending in 2017 decreased 1 percent to $28.4 billion. In 2017, the average Medicare margin for freestanding SNFs was 11.2 percent—the 18th year in a row that the average was above 10 percent. Margins varied greatly across facilities, reflecting differences in costs and shortcomings in the SNF prospective payment system (PPS) that favor treating rehabilitation patients over medically complex patients. Revisions to the PPS are still needed to improve the accuracy and equity of Medicare’s payments across different types of patients. CMS plans to revise the SNF PPS beginning in fiscal year 2020. The redesign will increase payments for medically complex patients and patients with high costs for nontherapy ancillary items (such as drugs), consistent with the Commission’s previously recommended designs for the SNF PPS and a unified post-acute care PPS. Skilled nursing facility services: Assessing payment adequacy and updating payments 194

225 The Commission recommends that the Secretary proceed with revising the SNF PPS and annually recalibrate the relative weights of the case-mix groups to keep payments aligned with the costs of care. The implementation of a revised SNF PPS will increase the equity of Medicare’s payments across different conditions and narrow the disparities in financial performance across SNFs. The redesigned PPS is likely to alter the mix of cases treated in SNFs, providers’ cost structures, and the relative costs of different types of stays. To keep costs and payments aligned across types of cases, CMS will need to regularly recalibrate the relative weights of the new case-mix groups. The level of payments continues to be well above the cost to treat Medicare beneficiaries. Several factors indicate that the aggregate level of Medicare’s payments remains too high. First, since 2000, the average Medicare margin has been above 10 percent; the marginal profit in 2017 was even higher (19 percent), suggesting that facilities with available beds have an incentive to admit Medicare patients. Medicare Advantage (managed care) payment rates to SNFs, considered attractive by many SNFs, are considerably lower than the program’s FFS payments. The small differences between beneficiaries enrolled in Medicare Advantage and FFS who used SNF services in 2017 would not explain the large difference in payments. Costs varied widely for reasons unrelated to case mix and wages. The very high Medicare margin (18 percent) for efficient SNFs—those providers with relatively low costs and high quality—is further evidence that Medicare continues to overpay for SNF care. Considering these factors, the Commission recommends that the Congress eliminate the fiscal year 2020 update to the Medicare base rates. While the level of payments indicates a reduction to payments is needed to more closely align aggregate payments and costs, the SNF industry is likely to undergo considerable changes as it adjusts to the redesigned PPS. Given the impending changes, the Commission will proceed cautiously in recommending reductions to payments. A zero update would begin to align payments with cost while exerting pressure on providers to keep their cost growth low. Medicaid trends As required by the Patient Protection and Affordable Care Act of 2010, we report on Medicaid use and spending and non-Medicare (private-payer and Medicaid) margins. Medicaid finances most long-term care services provided in nursing homes, but also covers the copayments on SNF care for low-income Medicare beneficiaries (known as dual-eligible beneficiaries) who stay more than 20 days in a SNF. The number of Medicaid-certified facilities has declined slightly since 2013, March 2019 | Report to the Congress: Medicare Payment Policy 195

226 by less than 1 percent, but remains close to 15,000. CMS reports total FFS spending on nursing home services declined 1.6 percent between 2016 and 2017 but projects small increases for 2019. In 2017, the average total margin—reflecting all payers (including managed care, Medicaid, Medicare, and private insurers) and all lines of business (such as hospice, ancillary services, home health care, and investment income)—was 0.5 percent, down from 2016 (0.7 percent). The average non-Medicare margin (which includes all payers and all lines of business except Medicare FFS SNF services) was –2.4 percent, the same as in 2016. ■ Skilled nursing facility services: Assessing payment adequacy and updating payments 196

227 The term skilled nursing facility refers to a provider that 3 Background meets Medicare requirements for Part A coverage. Most SNFs (more than 90 percent) are dually certified as SNFs Skilled nursing facilities (SNFs) provide short-term and nursing homes (which typically provide less intensive, skilled nursing care and rehabilitation services such as long-term care services). Thus, a facility that provides physical and occupational therapy and speech–language skilled care often also provides long-term care services pathology services. Examples of SNF patients include that Medicare does not cover. Medicaid pays for the beneficiaries recovering from surgical procedures such majority of nursing facility days. as hip and knee replacements or from medical conditions The mix of facilities where beneficiaries receive skilled such as stroke and pneumonia. In 2017, almost 1.6 million nursing care has shifted over time toward freestanding fee-for-service (FFS) beneficiaries (4.2 percent of Part and for-profit facilities. In 2017, almost all facilities were A FFS beneficiaries) used SNF services at least once; freestanding (96 percent), and they accounted for an even program spending on SNF services was $28.4 billion larger share of revenue (97 percent; Table 8-1) than other (about 7 percent of FFS spending) (Boards of Trustees types of facilities. Hospital-based SNFs made up a small 2018, Office of the Actuary 2018b). Medicare’s median share (4 percent or less) of facilities, stays, and spending. payment per day was $480, and its median payment per 1 For-profit facilities accounted for 71 percent of all SNFs stay was $18,121. In 2016, about one-fifth of hospitalized and 75 percent of revenues. beneficiaries were discharged to SNFs. Medicare FFS–covered SNF days typically account Medicare covers up to 100 days of SNF care per spell of for a small share of a facility’s total patient days but a illness after a medically necessary inpatient hospital stay 2 disproportionately larger share of the facility’s revenues. of at least 3 days. For beneficiaries who qualify for a In freestanding facilities in 2017, Medicare FFS covered stay, Medicare pays 100 percent of the payment beneficiary stays constituted 11 percent of total facility for the first 20 days of the spell of illness. Beginning with days but accounted for 19 percent of facility revenue, a day 21, beneficiaries are responsible for copayments for decline from 2010 when FFS Medicare accounted for 23 day 21 through day 100 of the covered stay. For fiscal year percent of facility revenue (data not shown). The decrease 2019, the copayment is $170.50 per day. in the FFS Medicare share of revenues reflects the growth in Medicare Advantage (MA) enrollment. TABLE Freestanding SNFs and for-profit SNFs accounted for the majority 8–1 of facilities, Medicare stays, and Medicare spending, 2017 Type of SNF Facilities Medicare-covered stays Medicare spending 15,090 $25.9 billion 2,266,301 Total number Freestanding 96% 97% 96% Hospital based 4 4 3 Urban 73 83 85 Rural 27 17 15 For profit 71 71 75 Nonprofit 23 24 21 6 4 4 Government SNF (skilled nursing facility). Totals may not sum to 100 percent due to rounding and missing values. The spending amount included here is lower than that reported by Note: the Office of the Actuary, and the count of SNFs is slightly lower than what is reported in CMS’s Survey and Certification Providing Data Quickly system. MedPAC analysis of the Provider of Services and Medicare Provider Analysis and Review files for 2017. Source: March 2019 | Report to the Congress: Medicare Payment Policy 197

228 (not service provision); remove payments for NTA The most common hospital conditions of patients services from the nursing component; establish a separate referred to SNFs for post-acute care are septicemia, joint component within the PPS that adjusts payments for replacement, heart failure and shock, hip and femur NTA services; and implement an outlier payment policy procedures (except major joint replacement), kidney and (Medicare Payment Advisory Commission 2008). The urinary tract infections, chronic obstructive pulmonary Commission’s recommended revisions to the PPS would disease, renal failure, and pneumonia. In 2017, the top increase the equity of Medicare’s payments and result in 10 diagnoses accounted for 43 percent of all SNF stays. considerable redistribution of payments, raising payments Compared with other beneficiaries, SNF users are older; for medically complex patients and decreasing them for more frail; and disproportionately female, disabled, living patients who receive intensive rehabilitation therapy that in an institution, and dually eligible for Medicare and appears unrelated to their clinical conditions (Medicare Medicaid (Medicare Payment Advisory Commission 2013). Payment Advisory Commission and The Urban Institute 7 SNF prospective payment system and its 2015). The revisions should increase access for patients shortcomings requiring complex medical care or costly drugs. Based on the mix of patients and therapy practices, payments Medicare uses a prospective payment system (PPS) to 4 would increase for hospital-based facilities and nonprofit Information gathered pay SNFs for each day of service. facilities and would decrease for freestanding facilities from a standardized patient assessment instrument—the and for-profit facilities. The effects on individual facilities Minimum Data Set—is used to classify patients into would depend on their mix of patients and current therapy case-mix categories called resource utilization groups 5 practices. (RUGs). Although the payment system is referred to as “prospective,” two features undermine how prospective Each year since 2008, the Commission has urged CMS it is: The system makes payments for each day of care to move forward with the much-needed reform and, since (rather than a set payment for the entire stay), and it bases 2012, recommended revising and rebasing the SNF PPS payments partly on the minutes of rehabilitation therapy to address both the distribution and level of payments furnished to a patient. Both features result in providers (Medicare Payment Advisory Commission 2012). The having some control over how much Medicare will pay Commission was not alone in calling for an overhaul of them for their services. the SNF PPS. OIG recommended that CMS evaluate the extent to which therapy payments should be reduced, Almost since its inception, the SNF PPS was criticized change the method for paying for therapy, adjust Medicare for encouraging the provision of excessive rehabilitation payments based on patient characteristics (not the amount therapy services and not accurately targeting payments of therapy furnished), and strengthen the oversight of SNF for nontherapy ancillary (NTA) items such as drugs billing (Office of Inspector General 2015). (Government Accountability Office 2002, Government Accountability Office 1999, White et al. 2002). Over CMS plans to revise the SNF PPS beginning time, the accuracy of Medicare’s payments has steadily October 1, 2019 eroded: Payments for NTA services are unrelated to the CMS’s work on alternative designs for the SNF PPS began cost of SNF care, and therapy payments have become less 6 13 years ago in response to a legislative requirement (the and less proportional to the costs of therapy services. As Medicare, Medicaid, and SCHIP Benefits Improvement a result, the PPS continues to advantage providers that and Protection Act of 2000) to conduct research on furnish therapy services unrelated to a patient’s condition potential refinements of the SNF PPS (Liu et al. 2007, and avoid patients with high NTA costs (Medicare Maxwell et al. 2003, Urban Institute 2004). In 2017, Payment Advisory Commission and The Urban Institute CMS issued an advanced notice of proposed rulemaking 2015). The Office of Inspector General (OIG) of the and sought comments on a redesign of the SNF PPS Department of Health and Human Services found that the that it planned to implement in fiscal year 2019 (Centers profitability of therapy services increased as the amount for Medicare & Medicaid Services 2017). Considering of therapy provided per day increased (Office of Inspector stakeholder comments, CMS revised the design and General 2015). delayed implementation until fiscal year 2020 (Centers for In 2008, the Commission recommended revising the Medicare & Medicaid Services 2018b). PPS to base therapy payments on patient characteristics Skilled nursing facility services: Assessing payment adequacy and updating payments 198

229 care, providers’ access to capital, Medicare FFS payments Consistent with the Commission’s recommended in relation to costs to treat Medicare beneficiaries, and design for the SNF PPS, CMS’s patient-driven payment changes in payments and costs. We also compare the model will base payments on patient characteristics, not performance of SNFs that have relatively high Medicare the amount of therapy services furnished to patients. margins and those with low Medicare margins, and we There will be five components—nursing, physical and compare relatively efficient SNFs with other SNFs. occupational therapy, speech–language pathology, NTA, and room and board—that will be summed to establish a Beneficiaries’ access to care: Access is stable daily payment. Except for the room and board component for most beneficiaries (which is uniform for every day of care), each component will have its own case-mix factors in which clinical We do not have direct measures of access in part because characteristics play a considerably larger role compared the need for SNF care, as opposed to the need for a with the current design. To reflect the declining average different PAC service or none at all, is not well defined. daily costs for physical and occupational therapy and NTA Instead, we consider the supply and capacity of providers services over the course of a stay, the daily payments for and evaluate changes in service volume. We also assess these components will be lower for days later in the stay. whether providers have a financial incentive to expand the So that individual therapy remains the dominant modality, number of Medicare beneficiaries they serve. group and concurrent therapy cannot make up more than The number of SNFs participating in the Medicare 25 percent of total therapy minutes. Given the clinical program in 2018 was stable at 15,326 (Centers for focus of the redesign, SNFs are likely to evaluate the Medicare & Medicaid Services 2018a). There was a clinical and coding expertise of its staff and the presence handful of new facilities (73, the majority of which were of physicians and medical directors and to reassess their for profit) and a number of terminations. There have been contracts with therapy vendors. 69 terminations as of November 2018, most of which were CMS estimates that the design will redistribute payments at the facilities’ initiative. This number is greater than at from patients assigned to the highest rehabilitation the same point in 2017, when there were 51 terminations. case-mix groups to medical patients, patients with high The SNF industry is highly fragmented and characterized NTA costs, and patients requiring tracheostomy or by independent providers and local and regional chains. ventilator services (Centers for Medicare & Medicaid Of the 50 largest operators, most are privately held. The Services 2018b). This redistribution is consistent with the 25 largest nursing home chains in the country operate Commission’s recommended designs for the SNF PPS 19 percent of all facilities (IQVIA Institute for Human and a unified post-acute care (PAC) PPS. CMS noted Data Science 2018). Single operators make up about 40 that the redesigned SNF PPS will bring the payment percent of the industry, small (often regional or religious) system closer to an eventual transition to a unified operators make up about one-quarter of facilities, with the PAC PPS (Centers for Medicare & Medicaid Services remaining third run by large chains (Ritchie and Johnson 2018b). Although intended to be budget neutral, provider 2017). The share of hospitals with financial links to SNFs responses to the new PPS, including changes in the has slowly increased as alternative payment models recording of patient diagnoses, will shape how spending encourage hospitals to lower spending and improve will change. Because case mix, service provision, and cost clinical outcomes for services furnished in post-acute care. structures are likely to change for many SNFs, CMS may In 2015, 18 percent of hospitals had a financial link to a need to recalibrate the relative weights of the case-mix SNF, up from 11 percent in 2005 (Fowler et al. 2017). One groups to keep payments aligned with the cost of care. study found that the integration of hospitals and SNFs increases Medicare payments for the hospital and PAC stays (combined) by extending the lengths of the SNF Are Medicare payments adequate in stays but also lowers rehospitalization rates (Konetzka et 2019? al. 2016). In 2017, 89 percent of beneficiaries lived in counties with To examine the adequacy of Medicare’s FFS payments, three or more SNFs or swing bed facilities (rural hospitals we analyze beneficiaries’ access to care (including the with beds that can serve as either SNF beds or acute supply of providers and volume of services), quality of care beds). Less than 1 percent of beneficiaries lived in a March 2019 | Report to the Congress: Medicare Payment Policy 199

230 TABLE SNF admissions and days continued to decline in 2017 8–2 Percent change Percent change 2010 2013 Volume measure 2016–2017 2017 2010–2017 2016 Covered admissions per 1,000 FFS beneficiaries 73.0 69.3 65.9 64.6 –2.0% –11.5% 1,872 1,972 Covered days per 1,000 FFS beneficiaries –17.7 –4.1 1,623 1,693 27.0 27.1 Covered days per admission –7.4 –2.3 25.1 25.7 SNF (skilled nursing facility), FFS (fee-for-service). “FFS beneficiaries” includes users and non-users of SNF services. Data include 50 states and the District of Note: Columbia. Source: Centers for Medicare & Medicaid Services 2018c. county without a SNF or swing bed facility, and another use by FFS beneficiaries declined over 11 percent, and 11 percent lived in counties with one or two SNFs or covered days per admission decreased almost 18 percent. swing bed facilities. The declines in SNF use reflects several trends, including Between 2016 and 2017, median occupancy rates for a growing presence of alternative payment models such freestanding SNFs declined slightly but remained high as accountable care organizations (ACOs) and bundled (84.7 percent). The lower occupancy rates reflect shorter payments that result in fewer beneficiaries referred to stays and fewer admissions. Occupancy rates at hospital- SNF care and shortened stays (Colla et al. 2016, Dummit based facilities were slightly lower (80.4 percent). There et al. 2016, McWilliams et al. 2017). Two studies of is wide variation in occupancy rates. In 2017, one-quarter CMS’s mandatory bundling initiative found participating of freestanding facilities had occupancy rates at or below hospitals had lower use of institutional PAC but similar 73 percent while another quarter had rates 91 percent or quality outcomes (Dummit et al. 2018, Finkelstein et al. higher. This variation indicates that some markets have 2018). The use of a narrower network of preferred SNFs the capacity to accommodate more admissions while has also resulted in shorter SNF stays (Dummit et al. other markets do not. The median occupancy rates for 2018, Huckfeldt et al. 2018). Hospitals participating in freestanding SNFs in rural areas and those in frontier the Comprehensive Care for Joint Replacement payment locations were lower than the average (77 percent and 71 model have adopted several strategies that could enhance percent, respectively). the care beneficiaries receive, including improved patient education; dedicated staff for coordinating care among the Between 2016 and 2017, SNF admissions hospital, physicians, and PAC providers; earlier initiation decreased and stays shortened of discharge planning; and wider use of standardized patient protocols (Dummit et al. 2018). In 2017, 4.2 percent of FFS beneficiaries used SNF services, the same share as in 2016. Between 2016 and Some SNFs report negative experiences of pressure from 2017, SNF admissions per 1,000 FFS beneficiaries ACOs and managed care organizations to shorten SNF decreased 2 percent (Table 8-2) (Centers for Medicare stays. A study of 25 SNFs participating in managed care & Medicaid Services 2018c). We examine service use and ACOs reported increased paperwork and time spent for only FFS beneficiaries because the CMS data on negotiating longer stays for patients, instances of declining users, days, and admissions do not include service use to admit patients who were likely to require long stays, by beneficiaries enrolled in Medicare Advantage (MA) and one instance of switching the attending physician to plans. Covered days per 1,000 FFS beneficiaries in 2017 remove the patient from an ACO (Tyler et al. 2018). A declined 2.3 percent to 25.1 days. The combination of survey of chief financial officers reported cumbersome fewer admissions and shorter stays resulted in 4.1 percent processes that they said made it more difficult for patients fewer days per 1,000 beneficiaries. Since 2010, SNF to receive the care they needed (Ziegler 2018). Skilled nursing facility services: Assessing payment adequacy and updating payments 200

231 home) focused on beneficiaries between 2000 and 2009 The decline in SNF admissions is also tied to the small who were recovering from hip fracture (Jung et al. 2016). decline in FFS per capita inpatient hospital stays that were It found that patients with more therapy were more likely three days or longer and therefore qualified beneficiaries to be discharged home, but the benefit of additional for Medicare coverage of SNF care. Although total per therapy decreased as the therapy intensity increased, and capita inpatient admissions increased, hospital admissions there was no additional benefit for patients in the highest for stays of at least three days decreased 0.6 percent. The case-mix groups. The large growth in days assigned to the expanded use of observation stays (during which a patient intensive therapy group raises the question of the value of is observed and treated but not admitted to the hospital) by these additional therapy services. hospitals is another contributing factor to lower SNF use (Mendelson et al. 2018). Because a three-day hospital stay Facilities differed in the amount of intensive therapy is required for Medicare coverage, some beneficiaries not they provided, though the differences by provider type meeting this requirement may continue to receive care that and ownership have narrowed over time as all providers is not covered by Medicare or be discharged home. assigned a larger share of days to intensive rehabilitation case-mix groups. In 2017, there was a 16 percentage Service mix reflects biases in PPS design point difference between freestanding and hospital- Since the PPS was implemented, providers have responded based facilities in the share of days assigned to intensive to the incentives to furnish enough therapy to classify therapy (83 percent in freestanding facilities, 67 percent days into rehabilitation case-mix groups and, within those in hospital-based facilities). There were smaller (2 groups, into the highest payment groups. Between 2002 percentage points) differences in case-mix between for- and 2017, the share of days classified into rehabilitation profit and nonprofit facilities (84 percent and 82 percent, case-mix groups in freestanding facilities increased from respectively). 78 percent to 95 percent; days assigned to special care, clinically complex, and extensive services made up the In 2017, the share of days assigned to medical case- other 5 percent of days. During the same period, the share mix groups or to extensive services case-mix groups of intensive therapy days (days assigned to the ultra-high was low (5 percent in 2017). Hospital-based units were and very high groups) as a share of total days rose from disproportionately represented in the group of SNFs with 27 percent to 83 percent. The share of days assigned to the highest shares (defined as the top quartile) of medically the highest rehabilitation case-mix groups (the ultra-high complex admissions. While making up 4 percent of group) increased from 7 percent to 58 percent. facilities, hospital-based SNFs made up 8 percent of the SNFs with the highest shares (the top quartile) of Changes in the frailty of beneficiaries at admission to a medically complex admissions. SNF do not explain the increases in therapy. Between 2012 and 2017, the average SNF user was the same age and In 2018, the Department of Justice continued to enforce had the same average risk score but by 2017 was slightly the False Claims Act by investigating fraud and abuse less able to perform activities of daily living (ADLs). The in SNFs’ therapy billings. It reached agreements in average Barthel index, a composite measure of a person’s four cases to settle allegations of improperly billing disability, was 5 percent lower, indicating less ability to for intensive therapy services that were not reasonably perform ADLs. For the 10 ADLs we examined, the shares or medically necessary (Department of Justice 2018a, of SNF users requiring the most help decreased for 7 Department of Justice 2018b, Department of Justice activities, remained the same for 1 activity, and increased 2018c, Department of Justice 2018d). The department 8 for 2 activities. Yet during this period, the amount of alleged that the defendant engaged in one or more of intensive therapy furnished to beneficiaries increased 15 the following strategies: falsely reporting the minutes of percent. OIG found that SNFs had increased their billing therapy delivered, furnishing services that were medically for the highest levels of therapy even though beneficiary unnecessary given the patient’s clinical care needs, characteristics—including age and the reasons for and discouraging therapists from providing services beyond severity levels of the preceding hospital stay—remained the minimum threshold minutes for a given case-mix unchanged (Office of Inspector General 2015). A study group, pressuring therapists and patients to complete examining whether additional therapy improved patient planned minutes of care even when patients were sick outcomes (in this case, the likelihood of being discharged March 2019 | Report to the Congress: Medicare Payment Policy 201

232 or declined to participate in therapy, or presumptively Quality of care: Measures indicate mixed assigning patients to the highest rehabilitation case- performance mix group regardless of each patient’s individual care The Commission tracks three broad categories of SNF needs. Since 2013, the Justice Department has settled quality indicators: risk-adjusted rates of discharge to the 16 cases involving allegations of improper provision of community, hospital readmission, and change in functional rehabilitation therapy services. status during the SNF stay (see text box on measures of SNF quality, pp. 204–205). We use these measures Medicare’s case-mix groups may have a broader impact because they reflect the goals of most beneficiaries: beyond Medicare-covered stays. One study of nursing to return home, avoid a readmission, and improve or homes in New York found that nursing home residents maintain function. Because of evidence that the function (whose care is not covered by Medicare) treated in for- information reported by inpatient rehabilitation facilities profit facilities in the last month of life were more likely (IRFs) and home health agencies (HHAs) may reflect to receive intensive therapy than low or medium levels of financial considerations, the Commission is concerned therapy (Temkin-Greener et al. 2018). New York Medicaid that the function information may not be reliable. The bases its payments on an older version of the same case- readmission rate during the SNF stay measures how mix groups that Medicare uses, which considers the well the SNF detects, monitors, and furnishes adequate amount of therapy in establishing payments. care to prevent readmissions. The postdischarge measure indicates how well facilities prepare beneficiaries and their Though access does not appear to be an issue in general, caregivers for safe and appropriate transitions to the next industry representatives and patient advocates report that health care setting (or home). some providers are reluctant to admit patients with high NTA costs (such as those who need expensive antibiotics). Changes in quality show mixed results: Some measures The design proposed by CMS should improve access for have improved since 2011 while others have not. these patients because payments will increase for patients The average rates of discharge to the community and with high NTA care needs by an estimated 27 percent readmission during SNF stays improved, the average rate (Centers for Medicare & Medicaid Services 2018b). of readmissions after discharge from the SNF worsened, Providers may avoid patients who are likely to require and two measures of change in function were essentially long stays and exhaust their Medicare benefits because a the same over this period. The most recent changes facility’s daily payments decline if the patient becomes (between 2016 and 2017) also indicate mixed progress. eligible for Medicaid or the stay results in bad debt. Rates of community discharge and readmissions Marginal profit: A measure of the attractiveness of show uneven progress Medicare patients Since 2011, SNF outcome-based measures show mixed Another measure of access is whether providers have a results; some measures improved while others worsened financial incentive to expand the number of Medicare slightly (Table 8-3). The average risk-adjusted rates of beneficiaries they serve. In considering whether to treat discharge to the community steadily improved and reached a patient, a provider with excess capacity compares 10 40 percent in 2017, up from 33.5 percent in 2011. the marginal revenue it will receive (i.e., the Medicare During the same period, the average risk-adjusted rate of payment) with its marginal costs—that is, the costs that potentially avoidable readmissions during the SNF stay vary with volume. If Medicare payments are larger than also improved, declining from 12.4 percent to 10.9 percent the marginal costs of treating an additional beneficiary, a in 2017. The average risk-adjusted rate of potentially provider has a financial incentive to increase its volume of avoidable readmissions during the 30 days after discharge Medicare patients. In contrast, if payments do not cover from the SNF worsened. Changes between 2016 and the marginal costs, the provider may have a disincentive 2017 exhibited a similar mixed pattern: an improved 9 to care for Medicare beneficiaries. For providers with discharge to community rate, a worse rate of readmissions available data, the marginal profit in 2017 was at least 19.1 after discharge from the SNF, and no change in the rate percent. Because Medicare payments far exceed facilities’ of readmissions during the SNF stay. There is a low marginal costs, facilities with available beds have an correlation between the during-stay readmission rates and incentive to admit Medicare patients, also signifying a the readmission rates during the 30 days after discharge positive indicator of patient access. from the SNF (0.16, which was statistically significant Skilled nursing facility services: Assessing payment adequacy and updating payments 202

233 TABLE Mean risk-adjusted rates of community discharge and 8–3 potentially avoidable readmissions, 2011–2017 2017 2011 2013 2015 2016 Measure 39.5% 40.0% Discharged to the community 33.5% 35.7% 38.8% Potentially avoidable readmissions: During SNF stay 12.4 11.2 10.9 10.9 10.4 5.5 5.9 During 30 days after discharge from SNF 6.1 5.8 5.0 Note: SNF (skilled nursing facility). Higher rates of discharge to the community indicate better quality. Higher readmission rates indicate worse quality. Rates are the average of facility rates calculated for all facilities with 25 or more stays, except the rate of potentially avoidable readmissions during the 30 days after discharge, which is reported for all facilities with 20 or more stays. Source: Analysis of fiscal year 2011 through fiscal year 2017 Minimum Data Set and inpatient acute hospital claims data for fee-for-service beneficiaries. given the sample sizes), confirming that the measures in October 2018. The VBP program withholds 2 percent capture different dimensions of quality. of payments; of the withheld amount, 60 percent will be returned to providers as incentive payments and 40 The general trend of lower readmission rates during the percent will be retained as program savings. On net, the SNF stay since 2011 in part reflects the increased attention program lowered net payments for the majority of SNFs from hospitals and ACOs to avoid readmission penalties (73 percent). These SNFs did not earn back some portion by partnering with SNFs that have low readmission of the amount withheld, and about one-fifth of SNFs rates. Some hospitals have established preferred provider did not earn back any portion of the 2 percent withhold. networks with higher quality SNFs, hoping to lower their For-profit facilities were overrepresented in the group of own readmission rates in exchange for increased referrals SNFs with the largest reduction. Net payments to over to SNFs. Two studies found that hospitals with more one-quarter of SNFs (27 percent) increased under the extensive collaboration efforts (such as transition care and VBP; they earned back more than the 2 percent withheld. visits by hospital staff to SNFs) had fewer readmissions The largest increase (a net gain of 1.6 percent) was earned (Rahman et al. 2018, Zhu et al. 2018). Another study by 11 percent of SNFs, and for-profit facilities were found that hospitals with a network of preferred SNFs underrepresented in this group. had lower readmission rates from their partnering SNFs (McHugh et al. 2017). While hospitals on average lowered In addition to the single VBP measure, CMS publicly their readmission rates between 2007 and 2013, hospitals reports SNF performance on six other measures. The affiliated with ACOs were quicker to lower them (Winblad three assessment-based measures are the share of patients et al. 2017). Because the ACO-affiliated hospitals were at with pressure ulcers that worsened, the share of patients greater financial risk, they may have had more effective experiencing one or more falls with major injury, and the discharge planning and information sharing with the SNFs share of patients with admission and discharge functional they used. In addition to partnering with hospitals, many assessments and a care plan that addresses function. The SNFs want to secure volume from MA plans, though there three claims-based measures are the rate of successful is some evidence that MA plans guide their enrollees to discharges to the community (i.e., discharged to the lower quality facilities (Meyers et al. 2018). community without deaths or unplanned readmissions within the 31 days after discharge), the rate of potentially As part of the Protecting Access to Medicare Act of 2014, preventable readmissions following discharge from the the Congress enacted a SNF value-based purchasing SNF, and Medicare spending per beneficiary. Since (VBP) policy that uses one measure—readmissions within October 2018, providers that do not submit the necessary 11 30 days of discharge from the preceding hospital stay. data to calculate the three assessment-based measures on The VBP program began adjusting payments to providers March 2019 | Report to the Congress: Medicare Payment Policy 203

234 Measures of skilled nursing facility quality musculoskeletal injuries, acute delirium, adverse drug o assess skilled nursing facility (SNF) quality, reactions, cellulitis/wound infection, pressure ulcers, the Commission examines risk-adjusted rates and blood pressure management. The count excludes of readmission to the hospital, discharge to the T readmissions that were likely to have been planned community, and change in functional status during (e.g., inpatient chemotherapy or radiation therapy) the SNF stay for beneficiaries in fee-for-service (FFS) and readmissions that signal a premature discharge Medicare. from the hospital. We separately measure readmissions The community discharge measure includes that occur during the SNF stay and those that occur beneficiaries discharged to a community setting within 30 days of discharge from the SNF because they (including assisted living) and excludes beneficiaries measure different aspects of care—care furnished by discharged to an inpatient setting (e.g., an acute care the SNF and the SNF handoff to the next setting (or hospital or nursing home) within one day of the SNF home). We do not use CMS’s measure (readmissions discharge. The measure also excludes beneficiaries who that occur within 30 days of discharge from the die within 1 day of the SNF discharge and beneficiaries hospital) because it conflates the two dimensions of who are readmitted to an acute care hospital within 30 care. days of admission to the SNF (Kramer et al. 2015). The observed readmission and community discharge Beneficiaries who are discharged to a nursing home are 12 rates were risk adjusted for medical comorbidity, not counted as community discharges. cognitive comorbidity, mental health comorbidity, The readmission measures count patients whose function, and clinical conditions (e.g., surgical wounds primary diagnosis for rehospitalization was considered and shortness of breath). The rates reported are the potentially avoidable; that is, the development of average risk-adjusted readmission rates for all facilities the conditions leading to the hospital admission with 25 or more stays (20 stays for the postdischarge typically could have been managed with appropriate readmission measure). Demographics (including race, care to avoid the hospitalization. The potentially gender, and age categories except younger than age avoidable conditions include congestive heart failure, 65 years) were not important in explaining differences electrolyte imbalance/dehydration, respiratory in readmission and community discharge rates after infection, septicemia, urinary tract or kidney controlling for beneficiaries’ comorbidities, mental 13 infection, hypoglycemia and diabetic complications, illness, and functional status (Kramer et al. 2014). anticoagulant complications, fractures and (continued next page) SNF stay, while others (such as those with chronic and at least 80 percent of assessments will have their update degenerative diseases) may expect, at best, to maintain for that year reduced by 2 percentage points. their function. We measure SNF performance on both Measures of changes in functional status were aspects of patient function—improvement and no decline essentially unchanged (see text box on SNF quality measures). The risk-adjusted rates consider the likelihood that a patient’s functionality Most SNF beneficiaries receive rehabilitation therapy, will change, given the functional ability at admission. and the amount of therapy furnished to them has steadily However, given the evidence in HHAs and IRFs that increased over time. Yet patients vary considerably in the reporting of functional status may be influenced by their expected improvement during the SNF stay. Some financial considerations, the Commission is increasingly patients are likely to improve in several ADLs during their Skilled nursing facility services: Assessing payment adequacy and updating payments 204

235 Measures of skilled nursing facility quality (cont.) to examine quality trends over time, unless providers Two risk-adjusted measures of functional change changed during the week the initial assessments were are used to gauge the share of a facility’s stays conducted. during which patients’ function improves (the rate of improvement in one, two, or three mobility measures— The initial assessment conducted during each stay is bed mobility, transfer, and ambulation) and the share used to assign the patient to 1 of 22 case-mix groups of stays during which patients’ functioning does not using 3 measures of mobility—bed mobility, transfer, decline (including stays with improvement and stays and ambulation (Kramer et al. 2014). This classification with no change), given the prognosis of the facility’s system acts as a form of risk adjustment, differentiating patients. Change is measured by comparing initial and patients based on their expected ability to perform the discharge assessments. For patients who go on to use three mobility-related activities of daily living (ADLs). long-term nursing home care, the assessment closest A patient’s prognosis is measured using the patient’s to the end of Medicare coverage is used as long as it ability to eat and dress because these two ADLs is within 30 days of the end of the SNF stay. Although encompass cognitive functioning and other dimensions the initial assessment often occurs toward the end of physical functioning that facilitate rehabilitation. of the first week of the stay, the Minimum Data Set information pertains to the number of times over the Risk-adjusted rates compare a facility’s observed rates past week that assistance was provided rather than the with its expected rates ((actual rate / expected rate) × recorded functional status at a single point in time. the national average rate) based on the mix of patients Therefore, measurement error due to the reliance on across functional outcome groups. Each facility-level an assessment conducted at the end of the first week measure combines the functional status information for of the stay is unlikely and would not affect our ability the three mobility measures. ■ found one-quarter of facilities in 2017 had risk-adjusted wary of the accuracy of the provider-reported functional community discharge rates at or below 31.9 percent, assessments because the data are generally obtained by whereas the best performing quarter of facilities had rates observing the patient and are somewhat subjective. The of 49.1 percent or higher (Table 8-5, p. 207). Similar Commission has work underway to examine the accuracy variation was seen in readmissions during the SNF stay: of these data. The worst performing quartile had rates at or above 13.6 The average risk-adjusted rates of functional change— percent, whereas the best quartile had rates at or below 7.8 rate of improvement in one, two, or three mobility ADLs percent. Finally, rates of readmission in the 30 days after (bed mobility, transfer, and ambulation) and the rate of no discharge from the SNF varied most—a twofold difference decline in mobility—were essentially unchanged between between the 25th percentile and the 75th percentile. The 2011 and 2017 (Table 8-4, p. 206). In 2017, 43.9 percent amount of variation across and within the groups suggests of stays had improvement in mobility, and 87 percent considerable room for improvement, all else being of stays had no decline in mobility. So, even though equal. There was less variation in the mobility measures, the program paid for more therapy over this period (the particularly the measure detecting no decline in mobility. share of days assigned to the highest rehabilitation case- The relatively high and fairly uniform rates could indicate mix groups increased), the average functional status of that most SNFs are able to prevent declines for most beneficiaries did not improve. beneficiaries. Large variation in quality measures indicates Consistent with prior years, in 2017, nonprofit SNFs considerable room for improvement had higher rates of community discharges and fewer Considerable variation exists across the industry in performance on the quality measures we track. We March 2019 | Report to the Congress: Medicare Payment Policy 205

236 TABLE Mean risk-adjusted functional outcomes in SNFs 8–4 were essentially unchanged between 2011 and 2017 2017 2011 2013 2015 2016 Composite measure 43.6% Rate of improvement in one or more mobility ADLs 43.6% 43.7% 43.6% 43.9% 87.1 87.0 87.2 87.1 87.2 Rate of no decline in mobility SNF (skilled nursing facility), ADL (activity of daily living). The three mobility ADLs include bed mobility, transfer, and ambulation. The rate of mobility improvement Note: refers to the average rates of improvement in bed mobility, transfer, and ambulation, weighted by the number of stays included in each measure. Stays with improvement in one, two, or three of these ADLs are counted in the improvement measure. The rate of stays with no decline in mobility is the share of stays with no decline in any of the three mobility ADLs. Rates are the mean of facility rates and are calculated for all facilities with 25 or more stays. Analysis of fiscal year 2011 through fiscal year 2017 Minimum Data Set data. Source: readmissions (that is, better rates) during the SNF stay Providers’ access to capital was adequate in than for-profit facilities. Nonprofit SNFs on average had 2018 community discharge rates that were 10 percent higher The vast majority of SNFs operate within nursing homes; and during-stay readmission rates that were 15 percent therefore, in assessing SNFs’ access to capital, we look lower than for-profit facilities. The rates for readmissions at the availability of capital for nursing homes. Medicare during the 30 days after discharge were similar on average. makes up a minority share of almost all facilities’ revenues. We also found differences in the performance of hospital- based and freestanding SNFs in 2017. Compared with Access to capital was adequate in 2018 and is expected to freestanding facilities, hospital-based SNFs had, on remain so in 2019 (Kaufman 2018). Many investors and average, higher rates of discharge to the community (18 lenders remain optimistic about this sector because of its percent higher) and lower during-stay readmission rates relatively low costs compared with other institutional PAC (27 percent lower). The average readmission rate during providers and the long-term demographics that will fuel the 30 days after discharge was higher for hospital-based demand. Capital markets are reported to be “robust,” with SNFs compared with freestanding facilities, indicating “tremendous investor demand,” even though facilities’ an opportunity for hospitals to improve their discharge total margins are low and occupancy rates have declined in planning, the handoffs of these beneficiaries to the next recent years (Connole 2018, Flynn 2018). Improved state setting or home, and the quality of the providers to which economies have also stabilized Medicaid payments for the they refer beneficiaries. long-term care portion of providers’ businesses. Medicare is increasingly focused on measuring the value The Department of Housing and Urban Development of the care it purchases. In addition to implementing a (HUD) continues to be an important lending source. VBP program in October 2018, CMS has improved the In fiscal year 2018, HUD financed 317 projects, with Nursing Home Compare website, a Medicare website the insured amount totaling $3.6 billion, a 6 percent that displays comparative information about SNFs and increase from 2017 (Department of Housing and Urban nursing homes to help beneficiaries select a provider. Development 2018). During fiscal year 2018, both the CMS expanded the number of short-stay quality number and size of the loans increased. Refinancing, measures reported in Nursing Home Compare to include rather than new construction or renovation, continues measures that reflect the key goals of this post-acute to make up most of HUD loans. HUD plays a smaller care. The short-stay measures include improvement in lending role than it has previously because low-cost function, readmissions, discharge to the community, borrowing and widely available capital sources have made patient experience with pain, presence of new or it only one of many alternative lenders. worse pressure ulcers, vaccination rates, and use of antipsychotic medications. Skilled nursing facility services: Assessing payment adequacy and updating payments 206

237 TABLE 8–5 SNF quality measures varied considerably across SNFs, 2017 Risk-adjusted rates Ratio of 75th to 75th 25th 25th percentile Mean Quality measure percentile percentile 40.0% Discharged to the community 1.5 31.9% 49.1% 43.9 35.8 52.0 1.5 Average mobility improvement across the three mobility ADLs during the SNF stay Rate of no decline in mobility during SNF stay 87.0 82.5 92.6 1.1 Potentially avoidable readmissions during SNF stay 10.9 7.8 13.6 1.7 3.9 6.1 7.8 Potentially avoidable readmissions within 30 days after discharge from SNF 2.0 Note: SNF (skilled nursing facility), ADL (activity of daily living). Higher rates of discharge to community indicate better quality. Higher readmission rates indicate worse quality. “Mobility improvement” is the average of the rates of improvement in bed mobility, transfer, and ambulation, weighted by the number of stays included in each measure. “No decline in mobility” is the share of stays with no decline in any of the three mobility ADLs. Rates are the average of facility rates and are calculated for all facilities with 25 or more stays, except the rates of potentially avoidable readmissions during the 30 days after discharge, which are reported for all facilities with 20 or more stays. Source: Analysis of fiscal year 2017 Minimum Data Set and inpatient acute hospital data. Yet some lenders and investors are wary of this setting. As some of the larger national players (such as Kindred Total margins for nursing homes across all lines of Healthcare and Sabra Healthcare REIT) exited or pared business and all patients are modest and have ranged back their investments in the sector, smaller regional between 0.6 percent and 3.8 percent since 2001. Because investors picked up the offerings, assisted by widely a “total margin” includes the mostly Medicaid-funded available capital. Although small regional operators are long-term care (the nursing home portion of the business), less able to spread their financial risks across diverse the overall financial performance of this setting is locations, they have greater familiarity with the local heavily influenced by state policies regarding the level of markets, referral patterns from hospitals, and individual Medicaid payments and the ease of entry into a market facility performance that may offer them a competitive (e.g., whether there is a requirement for a certificate of advantage (Spanko 2018). In contrast to reluctant lenders, need). The aggregate total margin for freestanding SNFs in these investors view the industry as remarkably stable, 2017 remained positive (0.5 percent), slightly lower than having the advantage of demographic trends and being a the total margin in 2016 (0.7 percent). lower cost alternative to other institutional PAC. Some investors eye the slim total margins, declining The nursing home industry is increasingly dividing occupancy rates, and increasing share of revenues into providers that can expand their service lines and from payers with lower rates and opt to pare back their successfully participate in alternative payment models investments or avoid the sector altogether. Reflecting these and providers that cannot. The transition from FFS to trends, the average price per bed decreased 18 percent alternative payment models (including ACOs and bundled between 2016 and 2017 (to $81,350), though it remains payments) and VBP will require SNFs to achieve good the third highest price ever (Irving Levin Associates outcomes and communicate that performance to potential Inc. 2018). However, reluctance to invest in this setting partners (hospitals and health systems) to secure volume. does not reflect the adequacy of Medicare’s FFS SNF Some facilities have started to develop and market their payments: Medicare remains a preferred payer. “niche” clinical capabilities to hospitals, aiming to care for patients with special care needs, such as patients March 2019 | Report to the Congress: Medicare Payment Policy 207

238 a Medicare-covered beneficiary was nine times higher than FIGURE the relative weight for a Medicaid-covered resident (White After declining in 2017, FFS program 8–1 FIGURE Overpayments... spending on SNF services is expected and Zheng 2018). The average nursing relative weight X-X to increase in 2018 and 2019 was 40 percent higher for a Medicare-covered beneficiary compared with a Medicaid-covered resident. $35 $900 900 $800 Medicare payments and providers’ costs: 800 $30 Medicare margins remained high in 2017 $700 700 $25 In 2017, the aggregate Medicare margin for freestanding $600 600 SNFs was 11.2 percent. Margins for individual facilities $20 continue to vary depending on the facility’s share of $500 500 intensive therapy days, size, and cost per day. High-margin $400 400 $15 SNFs had higher shares of intensive therapy days and 300 $300 lower average costs per day compared with low-margin $10 Average spending per SNFs. Differences by ownership were considerable, 200 $200 FFS beneficiary with for-profit facilities having much higher Medicare FFS program spending $5 Program spending (in billions) 100 $100 margins than nonprofit facilities. The 987 freestanding (in billions) Average spending per FFS beneficiary facilities defined as relatively efficient—providers with $0 0 $0 2016 2004 2012 2008 consistently low costs and higher quality care, in relative 2006 2018 2010 2014 terms—had Medicare margins of 18 percent, indicating Medicare overpays freestanding facilities for this care. Note: SNF (skilled nursing facility), FFS (fee-for-service). Fiscal year spending is Some MA plans’ payment rates were considerably lower shown. Data for 2018 and 2019 are estimates. than Medicare’s FFS payment rates, and the disparity is Source: Office of the Actuary 2018b. unlikely to be explained by differences in patient mix. Trends in FFS spending and cost growth In fiscal year 2017, Medicare FFS spending for SNF services was $28.4 billion, about 1 percent lower than Note: Note and Source in InDesign. on ventilators or dialysis, those requiring dementia or in 2016 (Figure 8-1) (Office of the Actuary 2018b). wound care, or those with respiratory or heart conditions. Between 2004 and 2010, the average increase in program The revised PPS that CMS plans to implement will also spending was over 8 percent a year. In 2011, program exert pressure on providers to develop skilled nursing spending was unusually high because rates for the new capabilities if they do not already have them. Some case-mix classification system included an adjustment observers note that some small solo operators will opt to Notes about this graph: that was too large for the mix of therapy modalities sell rather than transition to a new model of care, which • I did this all manually, since it has two axes. (i.e., individual versus group or concurrent) assumed in will likely result in more consolidation in the industry. setting the rates. The industry took advantage of the new Because Medicaid payments are lower than Medicare FFS policies by quickly shifting its mix of modalities, and 900 payments, some representatives in the industry argue that spending increased by over 14 percent in 2011. To correct 800 high Medicare payments are needed to subsidize losses on for the excessive payment, CMS revised the adjustment Medicaid residents. The Commission does not support this downward in 2012, and total payments declined almost 8 700 policy for several reasons (see text box on not subsidizing percent in 2012. Although there was no significant overall 600 other payments). It should be noted that while Medicare’s change in program spending, annual changes have been payments are higher than Medicaid’s, the programs pay highly variable, ranging from a 4.5 percent increase in 500 for different levels of care. Medicare pays for skilled 2015 to a 2.8 decrease in 2016. On a per FFS beneficiary 400 services after a hospitalization; Medicaid covers long-term basis, spending in 2017 ($743) was slightly lower (–0.4 300 care. Differences in the level of care are captured by the percent) than in 2016. The Office of the Actuary estimates relative weights for the average Medicare beneficiary and that FFS spending has increased in 2018 and will further 200 Medicaid resident. The average therapy relative weight for increase to $29.9 billion in 2019. 100 0 Skilled nursing facility services: Assessing payment adequacy and updating payments 208

239 Medicare’s skilled nursing facility payments should not subsidize payments from Medicaid or other payers In addition, Medicare’s subsidization does not edicare payments, which are financed by differentiate among states with relatively high and low taxpayer contributions to the Part A Trust Medicaid payments. If Medicare raises or maintains Fund, effectively subsidize payments from M its high payment levels, states could be encouraged other payers, most notably Medicaid. High Medicare to further reduce their Medicaid payments and, in payments may also subsidize payments from private turn, create pressure to raise Medicare rates even payers. Industry representatives contend that this more. Higher Medicare payments could also further subsidization should continue. The Commission encourage providers to select patients based on payer believes such cross-subsidization is poor policy for source or rehospitalize dual-eligible patients to qualify several reasons. First, it results in poorly targeted them for a Medicare-covered, higher payment stay. subsidies. Facilities with high shares of Medicare Finally, Medicare’s high payments represent a subsidy beneficiary days receive the most in subsidies from from trust fund dollars (and taxpayer support) of the higher Medicare payments, while facilities with low low payments made by states and private payers. If the shares of Medicare beneficiary days—presumably the Congress wishes to financially support certain nursing facilities with the greatest financial need—receive the facilities (such as those with high Medicaid shares) smallest subsidies. efficiently, it could do so through a separate, targeted policy. ■ Between 2016 and 2017, SNFs kept the growth in the freestanding SNFs was 11.2 percent, the 18th consecutive average cost per day below the market basket (2.3 percent year of Medicare margins above 10 percent (Figure 8-2, compared with the market update of 2.7 percent). Costs p. 210). Medicare margins declined slightly because, increased more quickly for nonprofit SNFs compared with although SNFs kept their cost growth below the update for-profit SNFs (3.0 percent compared with 2.2 percent, to payments, the sequester has lowered payments by 2 respectively). Cumulatively since 2012, the industry kept percent each year since April 2013. SNFs have countered the growth in the average cost per day below the market this reduction to the payment rate by keeping their cost basket (11.1 percent compared with the market basket of growth low and assigning days to higher payment case- 12.3 percent). Over the same period, nonprofit SNFs had mix groups. higher cost growth compared with for-profit SNFs (14.7 In 2017, hospital-based facilities (3 percent of program percent for nonprofit facilities compared with 10.1 percent spending on SNFs) continued to have extremely negative for for-profit SNFs). In addition to higher cost growth, Medicare margins (–68 percent), in part because of nonprofit facilities had average costs per day in 2017 that the higher cost per day reported by hospitals. Previous were about 10 percent higher than the cost per day in for- analysis by the Commission found that routine costs in profit facilities. Differences in the level of cost per day by hospital-based SNFs were higher, reflecting more staffing, ownership have grown over time. higher skilled staffing, and shorter stays (over which to allocate costs) (Medicare Payment Advisory Commission SNF Medicare margins remain high 2007). However, hospital administrators consider their The Medicare margin is a key measure of the adequacy of SNF units in the context of the hospital’s overall financial the program’s payments because it compares Medicare’s performance and mission. Hospitals with SNFs can lower FFS payments with providers’ costs to treat FFS their inpatient lengths of stay by transferring patients to beneficiaries. In 2017, the aggregate Medicare margin for their SNF beds, thus making inpatient beds available to March 2019 | Report to the Congress: Medicare Payment Policy 209

240 FIGURE Freestanding Medicare margins... FIGURE X-X Aggregate freestanding SNF Medicare margins have been above 10 percent since 2000 8–2 25 21.3 19.4 20 18.0 17.7 17.5 16.7 14.7 14.1 15 13.8 13.2 13.1 12.7 12.8 12.8 11.6 11.2 11.0 10.1 10 Medicare margin (in percent) 5 0 2015 2016 2000 2001 2017 2003 2002 2009 2011 2012 2010 2013 2014 2008 2007 2006 2005 2004 SNF (skilled nursing facility). Medicare margin is calculated as the sum of Medicare payments minus the sum of Medicare’s costs, divided by Medicare payments. Note: Source: MedPAC analysis of freestanding SNF cost reports, 2000–2017. Note: Note and Source are in InDesign. Source: treat additional inpatient admissions. As a result, hospital- that averaged 9 percentage points higher than facilities based SNFs can contribute to the bottom-line financial with low shares of these days (13.1 percent compared with performance of hospitals: Hospitals with SNFs had lower 4.1 percent, respectively). Notes about this graph: inpatient costs per case and higher inpatient Medicare Medicare margins also reflect the economies of scale that margins than hospitals without SNFs. • Data is in the datasheet. Make updates in the datasheet. larger SNFs are able to achieve. Small (25–50 beds) and • WATCH FOR GLITCHY RESETS WHEN YOU UPDATE DATA!!!! low-volume facilities (bottom quintile of total facility High and widely varying SNF Medicare margins days) had low average Medicare margins (–0.3 percent indicate PPS reforms are still needed • The column totals were added manually. and 0.6 percent, respectively) compared with large and The persistently high Medicare margins and their wide • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. high-volume facilities (12.6 percent and 13.4 percent, variation indicate that the PPS needs to be revised and • I can’t delete the legend, so I’ll just have to crop it out in InDesign. respectively). SNFs with the lowest cost per day (SNFs rebased so that payments more closely match patient in the bottom 25th percentile) had an average Medicare • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph characteristics, not the services provided to them. In margin of 22.8 percent compared with 0.3 percent 2017, one-quarter of freestanding SNFs had Medicare default when you change the data. for SNFs with the highest cost per day (the top 25th margins of 20.2 percent or higher, while another quarter • Use paragraph styles (and object styles) to format. percentile). of freestanding SNFs had margins of 0.8 percent or lower • Data was from: R:\Groups\MGA\data book 2007\data book 2007 chp1 (Table 8-6). Providers’ case mix played a key role in Since 2006, for-profit facilities’ Medicare margins have shaping Medicare margins. In 2017, facilities with high averaged about 10 percentage points higher than nonprofit shares of intensive therapy days had Medicare margins facilities’ margins. Nonprofit facilities had an average Skilled nursing facility services: Assessing payment adequacy and updating payments 210

241 Medicare margin of 1.7 percent, while the average margin TABLE for for-profit SNFs was 13.7 percent. The disparity reflects 8–6 Variation in freestanding SNF differences in facilities’ mix of patients, costs, size, and Medicare margins reflects the mix service provision. Nonprofit facilities tend to have higher of cases and cost per day, 2017 costs per day (about 10 percent higher) and, since 2011, Medicare have had higher cost growth compared with for-profit margin Provider group facilities. The higher costs for nonprofit facilities partly All providers 11.2% reflect their smaller size. In 2015, the median nonprofit facility had 85 beds compared with 103 beds for the For profit 13.7 median for-profit facility, suggesting that the nonprofits 1.7 Nonprofit may not be able to achieve the same economies of scale as Rural 9.7 larger facilities. As for revenues, nonprofits had somewhat Urban 11.5 lower shares of the more profitable ultra-high and very Frontier 3.8 high therapy days compared with for-profit facilities (82 percent compared with 84 percent, respectively) and 25th percentile of Medicare margins 0.8 shorter stays, both lowering revenue. 75th percentile of Medicare margins 20.2 Intensive therapy: High share of days 13.1 The highest margin freestanding SNFs (those in the top 4.1 Intensive therapy: Low share of days quartile of the distribution of Medicare margins) appear to pursue both cost and revenue strategies (Table 8-7, p. 212). Medically complex: High share of days 9.0 Compared with lower margin SNFs (those in the bottom 12.1 Medically complex: Low share of days quartile), high-margin SNFs had considerably lower Small (20–50 beds) –0.3 daily total, routine, and ancillary costs and lower cost per Large (100–199 beds) 12.6 discharge. Economies of scale play a role; high-margin SNFs were larger on average, with a higher occupancy 0.3 Cost per day: High rate, than lower margin facilities. Somewhat surprisingly, Cost per day: Low 22.8 high-margin facilities had larger shares of dual-eligible 9.6 Cost per discharge: High beneficiaries, minority beneficiaries, and Medicaid days. Cost per discharge: Low 12.2 It is possible that, given their larger Medicaid mix (and 13.4 Facility volume: Highest fifth the lower payments typically made by Medicaid), these 0.6 Facility volume: Lowest fifth facilities keep their costs lower, which contributes to their higher Medicare margins. Note: SNF (skilled nursing facility). The margins are aggregates for the facilities included in the group. “Intensive therapy” days are those classified in the On the revenue side, high-margin SNFs had revenues ultra-high and very high rehabilitation case-mix groups. “Low” is defined per day that were 15 percent higher, driven in part by as facilities in the lowest 25th percentile; “high” is defined as facilities in the highest 25th percentile. “Frontier” refers to SNFs located in counties having larger shares of intensive therapy days and, to a with six or fewer people per square mile. Facility volume includes all smaller extent, a lower mix of medically complex days. facility days. The differences in financial performance based on a Source: MedPAC analysis of 2017 freestanding SNF Medicare cost reports. provider’s case mix illustrate the need to revise the PPS, such as using the design proposed by CMS. Differences in payments per discharge between high- and low-margin SNFs were even larger (43 percent higher) because of the disproportionately urban, accounting for 79 percent of longer lengths of stay. this group even though they make up a smaller share of Ownership of low-margin and high-margin facilities did freestanding SNFs (73 percent). not mirror the industry mix. Although for-profit facilities Relatively efficient SNFs illustrate Medicare’s made up 71 percent of freestanding SNFs in 2017, they payments are too high constituted a smaller share (57 percent) of the low- margin facilities and a higher share (86 percent) of the The Commission is required by the Medicare Prescription high-margin group. Similarly, high-margin SNFs were Drug, Improvement, and Modernization Act of 2003 to consider the costs associated with efficient providers. The March 2019 | Report to the Congress: Medicare Payment Policy 211

242 TABLE 8–7 Cost and revenue differences explain variation in Medicare margins for freestanding SNFs in 2017 Ratio of SNFs in the SNFs in the SNFs in the top margin quartile bottom margin top margin to SNFs in the quartile quartile Characteristic bottom margin quartile Cost measures $399 0.68 Standardized cost per day $271 $117 $167 0.70 Standardized ancillary cost per day 0.68 $224 Standardized routine cost per day $152 0.80 $14,116 $11,285 Standardized cost per discharge 87 65 1.35 Average daily census (patients) 84% 86% Occupancy rate (in percent) 1.02 Revenue measures Medicare payment per day $522 $452 1.15 1.43 $15,714 $22,470 Medicare payment per discharge 35 42 Medicare length of stay (days) 1.21 80% 1.10 Share of days in intensive therapy 88% 0.75 Share of medically complex days 3% 4% 23% Medicare share of facility revenue 1.77 13% 1.16 57% 66% Medicaid share of days Patient characteristics 1.07 1.32 1.41 Case-mix index Share dual-eligible beneficiaries 39% 26% 1.50 2.80 5% 14% Share minority beneficiaries 0.86 30% Share very old beneficiaries 35% Facility mix 57% 86% Share for profit N/A 70% 79% N/A Share urban Note: = 3,284) were in the top 25 percent n SNF (skilled nursing facility), N/A (not applicable). Values shown are medians for the quartile. Top margin quartile SNFs ( = 3,283) were in the bottom 25 percent of the distribution of Medicare margins. n of the distribution of Medicare margins. Bottom margin quartile SNFs ( “Standardized cost” refers to Medicare costs adjusted for differences in area wages and the case mix (using the nursing component’s relative weights) of Medicare beneficiaries. “Intensive therapy” days are days classified in ultra-high and very high rehabilitation case-mix groups. “Medically complex” includes days assigned to clinically complex and special care case-mix groups. “Very old beneficiaries” are 85 years and older. Figures in the first two columns are rounded, but ratios were calculated on unrounded data. Source: MedPAC analysis of freestanding 2017 SNF cost reports and claims. analysis informs the Commission’s update discussion by Second, performance has to be consistent, meaning that examining the adequacy of payments for those providers the provider cannot have poor performance on any metric that perform relatively well on cost and quality measures. in any of three consecutive years preceding the year under evaluation. The Commission’s approach is to develop a The Commission follows two principles when selecting set of criteria and then examine how many providers meet a set of efficient providers. First, the providers must do them. It does not establish a set share (for example, 10 relatively well on both cost and quality metrics (see text percent) of providers to be considered efficient and then box on identifying relatively efficient SNFs, p. 214). define criteria to meet that pool size. Skilled nursing facility services: Assessing payment adequacy and updating payments 212

243 TABLE 8–8 Financial performance of relatively efficient freestanding SNFs is a combination of lower cost per day and higher revenues per day Type of SNF Ratio of relatively Relatively efficient Other SNFs Performance in 2017 efficient to other SNFs Community discharge rate 1.27 39.8% 50.3% Readmission rate 0.83 10.9% 9.0% $324 $297 0.92 Standardized cost per day Standardized cost per discharge $8,948 $12,310 0.73 Medicare revenue per day $526 $476 1.11 18.0% Medicare margin 10.5% 1.71 3.61 0.6% 2.3% Total margin Facility case-mix index 1.06 1.36 1.44 38 days Medicare average length of stay 30 days 0.79 85% 1.03 Occupancy rate 87% 1.27 100 Average daily census 79 1.21 55% Share ultra-high therapy days 66% Share medically complex days 4.2% 3.8% 1.09 58% Medicaid share of facility days 0.93 63% 84% Share urban 67% N/A Share for profit 68% N/A 79% 21% 15% Share nonprofit N/A SNF (skilled nursing facility), N/A (not applicable). The number of facilities included in the analysis was 11,462. SNFs were identified as “relatively efficient” Note: based on their cost per day and two quality measures (community discharge and readmission rates) between 2014 and 2016; their performance was evaluated in 2017 and is displayed in the table. Relatively efficient SNFs were those in the best third of the distribution for one measure and not in the worst third for any measure in each of three years and were not a facility under “special focus” by CMS. Costs per day and per discharge were standardized for differences in case Quality measures were rates of risk-adjusted community discharge and readmission during the SNF stay mix (using the nursing component relative weights) and wages. . Quality measures were calculated for all facilities with at least 25 stays. “Ultra-high therapy days” includes days for patients with potentially avoidable conditions assigned to ultra-high case-mix groups. “Medically complex days” includes days assigned to clinically complex and special care case-mix groups. Table shows the medians for the measure. Figures in the first two columns are rounded, but ratios were calculated on unrounded data. Source: MedPAC analysis of quality measures and Medicare cost report data for 2014–2017. To identify efficient SNFs, we examined the financial SNFs had community discharge rates that were 27 percent performance of freestanding SNFs with consistent cost and higher and readmission rates that were 17 percent lower quality performance. To measure costs, we looked at costs (Table 8-8). Standardized costs per day were 8 percent per day that were adjusted for differences in area wages lower than for other SNFs. The aggregate Medicare and case mix. The quality measures were risk-adjusted margin for efficient SNFs was high (18 percent), rates of community discharge and potentially avoidable indicating that although these providers were relatively readmissions during the SNF stay. low cost and achieved relatively high quality, the program could get better value for its purchase if its payments were Our analyses found that many SNFs (987) had relatively lower. The high margin for these providers underscores low costs and provided relatively good quality care. the need for the program to lower its payments to more Compared with other SNFs in 2017, relatively efficient closely align them with the costs of care. March 2019 | Report to the Congress: Medicare Payment Policy 213

244 Identifying relatively efficient skilled nursing facilities analysis) provided relatively low-cost, high-quality e defined relatively efficient skilled nursing care—17 more facilities than last year. Less than half facilities (SNFs) as those with relatively (44 percent) were identified as efficient last year. low costs per day and good quality of care W Relatively efficient facilities were more likely to be for three years in a row, 2014 through 2016. The cost urban and for profit. Efficient SNFs were located in 44 per day was calculated using cost report data and was states, including 2 in frontier locations. adjusted for differences in case mix (using the nursing component relative weights) and area wages. To assess The method we used to assess performance attempts quality, we examined risk-adjusted rates of community to limit incorrect conclusions about performance based discharge and potentially avoidable readmissions that on poor data. Using three years to categorize SNFs as occurred during the SNF stay. Only facilities with at efficient (rather than just one year) avoids categorizing least 25 stays were included in the quality measures. providers based on random variation or on one To be included in the relatively efficient group, a SNF “unusual” year. In addition, by first assigning a SNF to had to be in the best third of the distribution of at a group and then examining the group’s performance in least one measure and not in the bottom third on any the next year, we avoided having a facility’s poor data measure for three consecutive years. Another criterion affect both its own categorization and the assessment of was that SNFs not be part of CMS’s Special Focus the group’s performance. Thus, a SNF’s erroneous data Facility Initiative for any portion of time covered by the could result in its inaccurate assignment to a group, but definition (2014 through 2016), which excluded five because the group’s performance is assessed with data 14 facilities from the pool of efficient providers. from later years, these “bad” data would not directly affect the assessment of the group’s performance. ■ We found that almost 9 percent (987 of the 11,462 facilities that had all of the data items required for this Similar to high-margin SNFs, efficient SNFs appear payments. (We use “MA” as shorthand for all managed to pursue cost and revenue strategies. On the cost side, care payments since MA makes up the majority of rates efficient SNFs achieved greater economies of scale, reported as “managed care payments.”) We compared with a higher daily census compared with other facilities Medicare FFS and MA payments at three nursing home (100 compared with 79, respectively) and slightly higher companies for which such information was publicly occupancy rates. Since the efficient providers were also available. For these companies, Medicare’s FFS payments higher quality, their volume could reflect their success averaged 21 percent higher than MA rates (Table 8-9). in attracting admissions. On the revenue side, efficient We do not know whether the lower average daily payment providers had higher shares of the most intensive therapy reflects differences in service intensity (for example, fewer days that raised their daily Medicare payments relative intensive therapy days), lower payments for the same to all SNFs. They also had lower Medicaid shares, which service, or some combination. We also do not know how improved their total financial performance; efficient these rates compare with rates paid to smaller chains and providers’ total margin was 2.3 percent compared with independent facilities. It is possible that smaller companies 0.6 percent for other SNFs. Efficient facilities had more have less leverage and do not negotiate similarly low rates. complex case mixes (driven in part by higher therapy However, similar differences in payments were reported intensity) and much shorter stays. by the National Investment Center for Seniors Housing & Care, a nonprofit organization that supports access and FFS payments for SNF care are considerably choice for seniors’ housing and care, including nursing higher than MA payments for three publicly homes and assisted living. It found that for the 1,449 traded nursing home companies SNF properties included in its sample, FFS payments Another indicator that Medicare’s payments under the SNF PPS are too high is the comparison of FFS and MA Skilled nursing facility services: Assessing payment adequacy and updating payments 214

245 TABLE Comparison of Medicare fee-for-service and managed care 8–9 daily payments in 2018 to three companies Medicare payment Company Ratio of FFS to MA payment FFS Managed care (MA) $455 Diversicare $397 1.15 616 1.33 462 Ensign Group Genesis HealthCare 525 1.15 458 Note: FFS (fee-for-service), MA (Medicare Advantage). MA makes up the majority of managed care payments. The Genesis rate is reported as “insurance,” which includes managed care but excludes Medicaid managed care and private pay. Third quarter 10–Q 2018 reports available at each company’s website. Source: per day were 21 percent higher than MA rates (National percent, as required by the Balanced Budget Act of 2018. Investment Center for Seniors Housing & Care 2018). We also reduced 2019 payments by the portion of the VBP withhold that will be retained as program savings. We compared the patient characteristics of beneficiaries enrolled in FFS and MA plans in 2017 and found the The projected Medicare margin for 2019 is 10 percent. differences are unlikely to explain the magnitude of the The margin is expected to be lower than the 2017 margin differences between FFS payments and payments typically because of the MACRA-mandated update in 2018 and 15 made by MA plans. program savings from VBP that will lower revenues in Compared with FFS beneficiaries, 2019. MA enrollees were about the same age, had slightly higher Barthel scores (about two points, indicating slightly more independence), and had lower risk scores (5 percent lower, indicating fewer comorbidities). The considerably How should Medicare payments change lower MA payments indicate that some facilities accept in 2020? much lower payments to treat MA enrollees who may not be much different in terms of case mix from FFS In considering how payments should change for 2020, we beneficiaries. Some publicly traded firms report seeking note that costs are estimated to increase 3.1 percent that managed care patients as a business strategy, indicating year. The update to payments is estimated to be lower than that the MA rates are attractive. 3.1 percent because productivity adjustments will lower the update by an estimated 0.5 percent, for a net update of Payments and costs for 2019 2.6 percent. The change in Medicare margins will depend To project the aggregate Medicare margin for 2019, the on whether cost growth exceeds the growth in payments Commission considers the relationship between SNF on a case-mix-adjusted basis. costs and Medicare payments in 2017 as a starting point. To estimate costs for 2018 and 2019, we assumed a cost In fiscal year 2020, CMS plans to make substantial growth equal to the average for the past five years (slightly changes to the SNF PPS. The Commission has called for a below market basket) and no behavioral changes. To revised PPS since 2008 and urges the Secretary to proceed estimate 2018 payments, we assumed payments in 2018 with the redesign. While CMS estimated the redesign would increase by 1.0 percent, as required by the Medicare to be budget neutral, provider responses to the new PPS Access and CHIP Reauthorization Act of 2015 (MACRA). may alter program spending and facilities’ cost structures For 2019, we assumed payments would increase by 2.4 and mix of cases. Thus, behavioral responses will dictate March 2019 | Report to the Congress: Medicare Payment Policy 215

246 and should become part of CMS’s annual upkeep of the whether rebasing and recalibration will be needed to keep SNF PPS, just as it is part of the annual updates made to payments aligned with the cost of care. acute care hospitals. Regarding the level of payments, indicators of the As CMS noted in its final rule for updating rates for fiscal adequacy of Medicare’s payments are positive. The year 2019, the redesigned SNF PPS and the unified PAC aggregate Medicare margin for SNFs has been above 10 percent since 2000 and is projected to be 10 percent PPS establish similar incentives for providers. SNFs will gain valuable experience under the revised SNF PPS that in 2019. In 2017, the marginal profit was 19.1 percent, will ready them for an eventual transition to a PAC PPS. indicating facilities with an available bed have an incentive to admit Medicare patients. Relatively efficient SNFs IMPLICATIONS 8-1 had a median Medicare margin of 18 percent, further evidence that the level of payments is too high relative Spending to the cost of care. Furthermore, FFS payments were • Relative to current law, this recommendation would considerably higher than the MA payments made to some not change program spending. The recommendation is SNFs, suggesting that some facilities are willing to accept budget neutral to the current level of spending. much lower rates than FFS payments to treat Medicare Beneficiary and provider beneficiaries. These factors show that the PPS continues to exert too little pressure on providers. • A redesigned PPS and an annual recalibration of the relative weights would increase the equity of RECOMMENDATION 8-1 Medicare’s payments for all beneficiaries, thereby helping to ensure access for all beneficiaries, The Secretary should proceed to revise the skilled nursing including those with medically complex conditions facility prospective payment system in fiscal year 2020 and should annually recalibrate the relative weights of the and those with high NTA costs. We do not expect the case-mix groups to maintain alignment of payments and recommendation to affect providers’ willingness or costs. ability to care for Medicare beneficiaries. RECOMMENDATION 8-2 RATIONALE 8-1 The Congress should eliminate the fiscal year 2020 update After proposing to revise the SNF PPS and postponing to the Medicare base payment rates for skilled nursing the implementation, CMS refined its design and appears facilities. poised to implement a revised PPS in October 2019. The revisions will increase the equity in payments for RATIONALE 8-2 different types of stays, increasing payments for medically Current law will increase base payments by a projected complex stays and decreasing payments for stays that 2.6 percent (the market basket net of productivity) in fiscal include intensive therapy unrelated to a patient’s care year 2020. The aggregate Medicare margin in 2017 was needs. While the redesign would narrow the disparities 11.2 percent, indicating that the current level of Medicare’s in financial performance that result from the mix of cases payment rates is more than adequate to accommodate cost facilities treat and therapy practices, it would not, and growth and provide care to Medicare beneficiaries without should not, address disparities that result from providers’ an update to the base rate. inefficiencies. While the level of Medicare’s payments indicates that The recommendation also calls for the Secretary to a reduction to payments (i.e., not simply maintaining annually recalibrate the relative weights of the case-mix payment rates at current levels) is needed to align groups. The redesign may encourage many SNFs to aggregate payments to aggregate costs, we expect the change their mix of cases and their cost structures and SNF industry to undergo considerable changes as it thereby shift the relative costs of days assigned to the adjusts to the redesigned PPS. Given the impending case-mix groups. To keep payments aligned with the changes, the Commission will proceed cautiously in cost of care for all types of days, CMS should recalibrate recommending reductions to payments to more closely the relative weights of the case-mix groups on a regular align them to costs. A zero update would begin to align schedule. Recalibration is administratively straightforward Skilled nursing facility services: Assessing payment adequacy and updating payments 216

247 TABLE 8–10 The number of nursing homes treating Medicaid enrollees declined slightly from 2017 to 2018 Average annual percent change 2013–2017 2017–2018 2013 2015 2018 2017 –0.5% –0.1% 14,955 15,024 15,076 15,082 Number of facilities The 2018 number is through November of that year; it does not include data from the full calendar year. Note: Certification and Survey Provider Enhanced Reporting on CMS’s Survey and Certification Providing Data Quickly system, 2013–2018. Source: report nursing home spending trends for Medicaid and payments with costs while exerting pressure on providers financial performance for non-Medicare payers. Medicaid to keep their cost growth low and to engage in practice patterns encouraged by alternative payment models. The revenues and costs are not reported in the Medicare cost Commission will monitor beneficiary access, quality of reports. In a joint publication with the Medicaid and CHIP Payment Access Commission, we report on characteristics, care, and financial performance and may consider future recommendations based on industry responses to the new service use, and spending for dual-eligible beneficiaries payment system. (Medicare Payment Advisory Commission and the Medicaid and CHIP Payment and Access Commission IMPLICATIONS 8-2 2018). Spending Medicaid covers nursing home (long-term care) and Relative to current law, this recommendation would • skilled nursing care provided in nursing facilities. lower program spending by between $750 million Medicaid also pays for long-term care services that and $2 billion for fiscal year 2020 and by between $5 Medicare does not cover. For beneficiaries who are dually billion and $10 billion over five years. Savings occur eligible for Medicaid and Medicare, Medicaid pays the because current law requires market basket increases Medicare copayments required of beneficiaries beginning for 2020. on day 21 of a SNF stay. Beneficiary and provider Count of Medicaid-certified nursing homes We do not expect this recommendation to have • The number of nursing facilities certified as Medicaid adverse effects on beneficiaries’ access to care. Given providers has stayed relatively stable, with a small the current level of payments, we also do not expect decline between 2017 and 2018 (Table 8-10). The decline the recommendation to affect providers’ willingness may reflect the expansion in some states of home- and or ability to care for Medicare beneficiaries. community-based services (HCBS), which allow more beneficiaries to remain in their homes rather than an institution. State HCBS waivers and federal initiatives Medicaid trends have accelerated the trend toward HCBS. In fiscal year 2018, all 50 states and the District of Columbia expanded Section 2801 of the Patient Protection and Affordable the number of beneficiaries served by HCBS, an increase Care Act of 2010 requires the Commission to examine from 47 states in fiscal year 2017 and 46 states in fiscal spending, use, and financial performance trends in the years 2015 and 2016 (Gifford et al. 2017). The reduced Medicaid program for providers with a significant portion number of Medicaid providers may also reflect some of revenues or services associated with Medicaid. We facilities shifting their focus to the skilled care market. March 2019 | Report to the Congress: Medicare Payment Policy 217

248 FIGURE FIGURE Title here... 1-XX Total Medicaid fee-for-service spending on nursing home 8–3 services declined about 2 percent, 2001–2018 60 55 50 45 40 35 Total spending (in billions of dollars) 30 2017 2003 2015 2016 2005 2018 2004 2006 2007 2008 2009 2010 2011 2012 2002 2013 2001 2014 Spending does not include any managed care organization spending on nursing homes. Data for 2018 are projected. Note: Office of the Actuary 2018. Source: Columbia increased rates (Gifford et al. 2018). More states Spending increased rates to nursing homes in 2018 than in 2017 Note: Note and Source are in InDesign. Spending on Medicaid-funded nursing home services (only 34 states raised rates in 2017, and 15 states restricted (combined state and federal funds) totaled $43.3 billion Source: rates) (Gifford et al. 2017). Furthermore, the National in 2017 (Figure 8-3) (Office of the Actuary 2018a). CMS Investment Center for Seniors Housing & Care reported estimates that FFS Medicaid spending on nursing home that Medicaid revenue per day has been increasing steadily Notes about this graph: services decreased by 1.6 percent between 2017 and since 2011 (National Investment Center for Seniors 2018 and that spending will increase by 0.45 percent • Data is in the datasheet. Make updates in the datasheet. Housing & Care 2018). Rates will likely shift in 2019; in 2019. This trend of lower spending is in part due to • I deleted the years from the x-axis and put in my own. 40 states and the District of Columbia have indicated that an increased use of managed care organizations, whose they will increase nursing home rates. Eleven states plan to • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. spending is not included in these data. As of 2017, 24 restrict rates in 2019 (Gifford et al. 2018). states operated capitated managed long-term services • The dashed line looked ok here, so I didn’t hand draw it. and supports (Lewis et al. 2018). This number is a 50 States continue to use provider taxes to raise federal • I can’t delete the legend, so I’ll just have to crop it out in InDesign. percent increase from 2012, when only 16 states had such matching funds. In fiscal year 2018, 44 states and the • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph programs. Furthermore, total enrollment in these programs District of Columbia levied provider taxes on nursing default when you change the data. more than doubled between 2012 and 2017, from 800,000 homes to increase federal matching funds (Gifford et 16 beneficiaries to 1.8 million beneficiaries. • Use paragraph styles (and object styles) to format. al. 2018). States can use the augmented revenue to increase payments to providers or to mitigate reductions to Analysis of Medicaid rate-setting trends found that 17 payments for services to Medicaid patients (Kaiser Family states restricted (froze or reduced) rates paid to nursing Foundation 2017). homes in 2018, while 34 states and the District of Skilled nursing facility services: Assessing payment adequacy and updating payments 218

249 TABLE Over the past 10 years, total margins remained positive in 8–11 freestanding SNFs but non-Medicare margins were negative 2012 2010 2017 Type of margin 2014 2016 2008 0.5% Total margin 0.7% 2.2% 3.6% 1.8% 1.9% Non-Medicare margin –2.4 –1.5 –2.0 –1.5 –2.4 –2.4 SNF (skilled nursing facility). “Total margin” includes the revenues and costs associated with all payers and all lines of business. “Non-Medicare margin” includes Note: the revenues and costs associated with Medicaid and private payers for all lines of business. MedPAC analysis of Medicare freestanding SNF cost reports for 2008 to 2017. Source: payer mix, median facility payments increased 3.4 percent Total margins and non-Medicare margins in between 2016 and 2017 for nonprofit facilities compared nursing homes with 2.1 percent for for-profit facilities. Differences in Total margins reflect all payers (including Medicare, cost growth (median facility costs per day) did not explain Medicaid, private insurers, and managed care) across all the diverging performances. Consistent with the growth lines of business (for example, nursing home care, hospice in Medicare cost per day (see p. 209), nonprofit facilities care, ancillary services, home health care, and investment experienced higher cost growth compared with for-profit income). In 2017, total margins were positive (0.5 percent) facilities. (Table 8-11). The median total margin was 0.6 percent, with margins at the 25th and 75th percentiles ranging from The declines in the average total and non-Medicare –5.3 percent to 5.5 percent, respectively (data not shown). margins reflect the lower average occupancy rates (which Rural and urban freestanding SNFs had similar total raises the average cost per day) and the lower volume margins (0.6 percent and 0.5 percent, respectively). Total of high-payment Medicare FFS patients. Beneficiaries margins have declined since 2013 (when the total margin receiving skilled nursing services are increasingly was 1.9 percent), reflecting the impact of reductions to enrolled in alternative payment models (including bundled Medicare payments mandated by the Patient Protection payments and ACOs) and MA plans, which typically seek and Affordable Care Act of 2010 and the growing share to shorten stays or avoid this setting entirely. In addition, of managed care payments that are lower than Medicare’s payments from MA plans are generally lower than FFS payments. The aggregate non-Medicare margin (the Medicare’s FFS rates (see p. 214). ■ profitability of all services except Medicare FFS SNF) in 2017 was –2.4 percent, the same as in 2016. Total margins varied by ownership, though the differences were much smaller than the differences by ownership in Medicare margins. In 2017, the average nonprofit SNF total margin was 1.9 percent compared with 0.2 percent for the average for-profit facility. In 2016, the total margin was 0.8 percent for both nonprofit and for-profit providers. The diverging performances reflect differences in the changes in mixes of revenue and days by payer. Compared with for-profit providers, the share of Medicare revenue and days decreased less for nonprofit facilities and the share of non-Medicare revenues (the lower- payment revenue sources) grew at half the rate of for-profit facilities. (Note that Medicaid revenues are not reported on the Medicare cost report.) Reflecting differences in March 2019 | Report to the Congress: Medicare Payment Policy 219

250 Endnotes 1 refers to an individual 7 Throughout this chapter, We include patients who are assigned to the clinically beneficiary whose SNF stay coverage is paid for by Medicare (Part A). complex and special care case-mix groups in our definition of medically complex. Clinically complex patients have burns, Some beneficiaries who no longer qualify for Medicare surgical wounds, hemiplegia, or pneumonia, or they receive coverage remain in the facility to receive long-term care chemotherapy, oxygen therapy, intravenous medications, services, which are not covered by Medicare. During long-term care stays, beneficiaries may receive care such or transfusions while in a SNF. Special care patients are as physician services, outpatient therapy services, and comatose; have quadriplegia, chronic obstructive pulmonary prescription drugs that are paid for separately under the disease, septicemia, diabetes requiring daily injections, Part B and Part D benefits. Services furnished outside the fever with specific other conditions, cerebral palsy, multiple Part A–covered stay are not paid under the SNF prospective sclerosis, Parkinson’s disease, respiratory failure, a feeding payment system and are not considered in this chapter. tube, pressure ulcers of specific sizes, or foot infections; Except where specifically noted, this chapter examines FFS receive radiation therapy or dialysis while a resident; or require parenteral or intravenous feedings or respiratory Medicare spending and service use and excludes services therapy for seven days. Intensive therapy days are classified and spending for SNF services furnished to beneficiaries in the ultra-high and very high rehabilitation case-mix groups. enrolled in Medicare Advantage plans. Some beneficiaries also qualify for Medicaid and are referred to as “dual-eligible Rehabilitation groups are based on minutes of rehabilitation beneficiaries.” provided per week. “Ultra-high rehabilitation” includes patients who receive more than 720 minutes per week; “very 2 A spell of illness ends when there has been a period of high rehabilitation” includes patients who receive 500–719 60 consecutive days during which the beneficiary was an minutes per week. inpatient of neither a hospital nor a SNF. Coverage for another 100 days does not begin until a beneficiary has not had The share of SNF users requiring the most assistance 8 hospital care or skilled care in a SNF for 60 consecutive days. dropped for transferring, walking in corridor, eating, performing personal hygiene, toileting, dressing, and bed Observation days and emergency room stays do not count mobility; remained the same for always being incontinent; toward the three-day hospital stay requirement. and increased for help with bathing and always being bowel For services to be covered, the SNF must meet Medicare’s 3 incontinent. requirements of participation and agree to accept Medicare’s 9 payment rates. Medicare’s requirements relate to many If we approximate marginal cost as total Medicare costs aspects of staffing and care delivery, such as requiring a minus fixed building and equipment costs, then marginal registered nurse in the facility for 8 consecutive hours per profit can be calculated as follows: day and licensed nurse coverage 24 hours a day, providing Marginal profit = (payments for Medicare services – (total physical and occupational therapy services and speech– Medicare costs – fixed building and equipment costs)) / language pathology services as delineated in each patient’s Medicare payments plan of care, and providing or arranging for physician services 24 hours a day in case of an emergency. This comparison is a lower bound on the marginal profit because we do not consider any potential labor costs that are The program pays separately for some services, including 4 fixed. certain chemotherapy drugs; certain customized prosthetics; certain ambulance services; Part B–covered dialysis; The Commission’s measure of discharge to community 10 emergency services; and certain outpatient services provided captures a key goal of many beneficiaries: to go home. It in a hospital (such as computed tomography, MRI, radiation measures the share of beneficiaries discharged home from a therapy, and cardiac catheterizations). SNF. In contrast, CMS’s quality reporting measure gauges the share of beneficiaries who were discharged home, did not Payment Basics The SNF 5 is available at http://medpac.gov/ have an unplanned readmission within 31 days of discharge, docs/default-source/payment-basics/medpac_payment_ and remained alive. basics_18_snf_final_sec.pdf?sfvrsn=0. 11 CMS’s VBP readmission measure differs from the 6 Payments for NTA services are included in the nursing Commission’s measures that separately track readmissions component, even though NTA costs vary much more than during the SNF stay and readmissions that occur within 30 nursing care costs and are not correlated with them. Skilled nursing facility services: Assessing payment adequacy and updating payments 220

251 days after discharge. By including readmissions that occur 14 The Special Focus Facility Initiative is a program to stimulate within 30 days of discharge from the hospital, CMS’s measure improvements in the quality of care at nursing homes with can include readmissions that occur during the SNF stay and a history of serious quality problems. The initiative targets after discharge, depending on the length of the SNF stay. For homes with a pattern over three years of more frequent and short SNF stays, CMS’s measure includes readmissions after more serious problems (including harm or injury to residents) discharge from the SNF but still within 30 days of discharge detected in their annual facility surveys. Facilities that from the hospital stay. For long SNF stays, the measure improve and maintain those improvements can “graduate” includes only readmissions that occur within the first 30 days from the program. Providers that do not improve face civil of the SNF stay (assuming an immediate transfer from the monetary penalties (fines) and eventual termination from hospital) and misses readmissions that occur later in the SNF Medicare and Medicaid. stay. We compared the assessments conducted at the beginning of 15 Separate models (with their own covariates) are used to 12 stays (the “day 5” assessment). MA plans are not required estimate expected community discharge rates for different to submit these assessments, and we cannot determine discharge destinations (e.g., discharged home with home what share of plans submits them or the possible bias in the health care, discharged home without home health care, and assessments that are submitted. discharged to a nursing home). A provider tax works as follows: A state taxes all nursing 16 13 With inclusion of the other covariates, age categories were not homes and uses the collected amount to help finance the found to be significant in explaining variation in outcomes state’s share of Medicaid funds. The provider tax increases and were dropped from the models except for the model the state’s contribution, which, in turn, raises the amount of explaining differences in readmission during the 30 days after federal matching funds. The augmented federal funds more discharge for beneficiaries younger than 65 years residing in than cover the cost of the provider tax revenue, which is the community. returned to providers. The provider tax is limited to 6 percent of net patient revenues. March 2019 | Report to the Congress: Medicare Payment Policy 221

252 References Department of Justice. 2018b. Signature HealthCARE to pay Boards of Trustees, Federal Hospital Insurance and Federal more than $30 million to resolve False Claims Act allegations 2018 Supplementary Medical Insurance Trust Funds. 2018. related to rehabilitation therapy. News release. June 8. https:// annual report of the Boards of Trustees of the Federal Hospital www.justice.gov/opa/pr/signature-healthcare-pay-more-30- . Insurance and Federal Supplementary Insurance Trust Funds million-resolve-false-claims-act-allegations-related. Washington, DC: Boards of Trustees. Department of Justice. 2018c. Two consulting companies and Centers for Medicare & Medicaid Services, Department of nine affiliated skilled nursing facilities to pay $10 million to Health and Human Services. 2018a. Certification and Survey resolve False Claims Act allegations relating to medically Provider Enhanced Reporting (CASPER) on CMS’s survey and unnecessary rehabilitation therapy services. News release. July certification providing data quickly (PDQ) system. 18. https://www.justice.gov/opa/pr/two-consulting-companies- Centers for Medicare & Medicaid Services, Department of Health and-nine-affiliated-skilled-nursing-facilities-pay-10-million. and Human Services. 2018b. Medicare program; prospective Department of Justice. 2018d. United States reaches settlement payment system and consolidated billing for skilled nursing with four facilities and two medical companies to resolve facilities (SNF) final rule for FY 2019, SNF Value-Based allegations of fraudulent billing in skilled nursing facilities. Purchasing Program, and SNF Quality Reporting Program. News release. March 22. https://www.justice.gov/usao-md/pr/ Proposed rule. 83, no. 153 (August 8): 39162– Federal Register united-states-reaches-settlement-four-facilities-and-two-medical- 39290. companies-resolve. Centers for Medicare & Medicaid Services, Office of Information Dummit, L., K. Smathers, O. J. Bright, et al. 2018. CMS Products and Data Analytics, Department of Health and Comprehensive Care for Joint Replacement Model: Performance Human Services. 2018c. Personal communication with Maria year 1 evaluation report . Report prepared by the Lewin Group for Diacogiannis, November 1. the Centers for Medicare & Medicaid Services. Falls Church, VA: Centers for Medicare & Medicaid Services, Department of Health The Lewin Group. https://downloads.cms.gov/files/cmmi/bpci- and Human Services. 2017. Medicare program: Prospective models2-4yr3evalrpt.pdf. payment system and consolidated billing for skilled nursing Dummit, L. A., D. Kahvecioglu, G. Marrufo, et al. 2016. facilities: Revisions to case-mix methodology. Advance notice of Association between hospital participation in a Medicare bundled Federal Register proposed rulemaking with comment. 82, no. 85 payment initiative and payments and quality outcomes for lower (May 4): 20980–21012. Journal of the American extremity joint replacement episodes. Colla, C. H., V. A. Lewis, L. S. Kao, et al. 2016. Association 316, no. 12 (September 27): 1267–1278. Medical Association between Medicare accountable care organization implementation Finkelstein, A., Y. Ji, N. Mahoney, et al. 2018. Mandatory JAMA and spending among clinically vulnerable beneficiaries. Medicare bundled payment program for lower extremity joint Internal Medicine 176, no. 8 (August 1): 1167–1175. replacement and discharge to institutional post-acute care: Interim Connole, P. 2018. LT/PAC financing sources remain rock solid Journal analysis of the first year of a 5-year randomized trial. , September. Provider despite outside pressures. 320, no. 9 (September 4): of the American Medical Association 892–900. Department of Housing and Urban Development. 2018. FY 2018 summary statistics. https://www.hud.gov/federal_housing_ Flynn, M. 2018. REIT sell-offs, investor demand fuel strong administration/healthcare_facilities/residential_care. Skilled Nursing News , July 24. skilled nursing volume in Q2. Department of Justice. 2018a. Louisville skilled nursing facility Fowler, A. C., D. C. Grabowski, R. J. Gambrel, et al. 2017. to pay $5,191,470 to settle false claims allegations. News release. Corporate investors increased common ownership in hospitals March 29. https://www.justice.gov/usao-wdky/pr/louisville- and the postacute care and hospice sectors. 36, no. Health Affairs skilled-nursing-facility-pay-5191470-settle-false-claims- 9 (September 1): 1547–1555. allegations. Skilled nursing facility services: Assessing payment adequacy and updating payments 222

253 States focus Gifford, K., E. Ellis, B. C. Edwards, et al. 2018. Kramer, A., R. Fish, and M. Lin. 2014. Potentially avoidable on quality and outcomes amid waiver changes: Results from a readmission and functional outcome quality measure development 50-state Medicaid budget survey for state fiscal years 2018 and and 2012 rates for SNFs . Report prepared for the Medicare 2019 . Prepared by staff from Health Management Associates and Payment Advisory Commission. Washington, DC: MedPAC. the Kaiser Commission on Medicaid and the Uninsured/Kaiser Kramer, A., M. Lin, R. Fish, et al. 2015. Refinement of community Family Foundation. Menlo Park, CA: Kaiser Family Foundation. discharge, potentially avoidable readmission, and functional https://downloads.cms.gov/files/cmmi/bpci-models2-4yr3evalrpt. outcome SNF quality measures, for fiscal years 2011, 2012, pdf. and 2013 . Report prepared for the Medicare Payment Advisory Gifford, K., E. Ellis, B. C. Edwards, et al. 2017. Medicaid moving Commission. Washington, DC: MedPAC. ahead in uncertain times: Results from a 50-state Medicaid Lewis, E., S. Eiken, A. Amos, et al. 2018. The growth of managed budget survey for state fiscal years 2017 and 2018 . Prepared long-term services and supports (MLTSS) programs: A 2017 by staff from Health Management Associates and the Kaiser update . Ann Arbor, MI: Truven Health Analytics. Commission on Medicaid and the Uninsured/Kaiser Family Foundation. Menlo Park, CA: Kaiser Family Foundation. Final report Liu, K., B. Garrett, D. A. Wissoker, et al. 2007. to CMS: Options for improving Medicare payment for skilled Skilled nursing Government Accountability Office. 2002. nursing facilities . Report submitted by the Urban Institute to the facilities: Providers have responded to Medicare payment system Centers for Medicare & Medicaid Services. Washington, DC: The . GAO–02–841. Washington, DC: GAO. by changing practices Urban Institute. Skilled nursing Government Accountability Office. 1999. A review of issues in Maxwell, S., J. Wiener, and K. Liu. 2003. facilities: Medicare payments need to better account for the development and implementation of the skilled nursing facility nontherapy ancillary cost variation . GAO/HEHS–99–185. . Report to the Centers for Medicare prospective payment system Washington, DC: GAO. & Medicaid Services. Contract no. 500–00–0025/TO#2. Huckfeldt, P. J., L. Weissblum, J. J. Escarce, et al. 2018. Do Baltimore, MD: CMS. skilled nursing facilities selected to participate in preferred McHugh, J. P., A. Foster, V. Mor, et al. 2017. Reducing hospital provider networks have higher quality and lower costs? Health readmissions through preferred networks of skilled nursing (August 15). Services Research Health Affairs facilities. 36, no. 9 (September 1): 1591–1598. U.S. nursing IQVIA Institute for Human Data Science. 2018. McWilliams, J. M., L. G. Gilstrap, D. G. Stevenson, et al. 2017. home market summary . Parsippany, NJ: IQVIA. Changes in postacute care in the Medicare Shared Savings Irving Levin Associates Inc. 2018. The senior care acquisition JAMA Internal Medicine Program. 177, no. 4 (April 1): 518–526. report: 23rd edition . Norwalk, CT: Irving Levin Associates Inc. Report to the Medicare Payment Advisory Commission. 2013. Jung, H. Y., A. N. Trivedi, D. C. Grabowski, et al. 2016. Does Congress: Medicare payment policy . Washington, DC: MedPAC. more therapy in skilled nursing facilities lead to better outcomes Report to the Medicare Payment Advisory Commission. 2012. 96, no. 1 Physical Therapy in patients with hip fracture? . Washington, DC: MedPAC. Congress: Medicare payment policy (January): 81–89. Medicare Payment Advisory Commission. 2008. Report to the Kaiser Family Foundation. 2017. States and Medicaid provider . Washington, DC: Congress: Reforming the delivery system taxes or fees . Fact sheet. Washington, DC: KFF. http://files.kff. MedPAC. org/attachment/fact-sheet-medicaid-provider-taxesfees-an-update. Report to the Medicare Payment Advisory Commission. 2007. Kaufman, B. 2018. Wrapping up year 2017 in data: Despite . Washington, Congress: Promoting greater efficiency in Medicare numerous challenges last year, investor interest continues in the DC: MedPAC. , April. skilled nursing sector. Provider Medicare Payment Advisory Commission and the Medicaid Konetzka, R. T., E. A. Stuart, and R. M. Werner. 2016. The and CHIP Payment and Access Commission. 2018. Data effect of integration of hospitals and post-acute care providers . book: Beneficiaries dually eligible for Medicare and Medicaid on Medicare payment and patient outcomes. https://ssrn.com/ Washington, DC: MedPAC/MACPAC. abstract=2805172. March 2019 | Report to the Congress: Medicare Payment Policy 223

254 Medicare Payment Advisory Commission and The Urban Spanko, A. 2018. Inside the key metrics skilled nursing buyers The need to reform Medicare’s payments to SNFs Institute. 2015. , August 20. Skilled Nursing News should track before investing. . Washington, DC: MedPAC/Urban Institute. is as strong as ever Temkin-Greener, H., T. Lee, T. Caprio, et al. 2018. Rehabilitation Mendelson, D. A., F. Bentley, and E. Breese. 2018. Medicare therapy for nursing home residents at the end-of-life. Journal of . Washington, patients are using fewer skilled nursing services the American Medical Directors Association (September 24). DC: Avalere Health. Tyler, D. A., J. P. McHugh, R. R. Shield, et al. 2018. Challenges Meyers, D. J., V. Mor, and M. Rahman. 2018. Medicare and consequences of reduced skilled nursing facility lengths of Advantage enrollees more likely to enter lower-quality nursing stay. Health Services Research (June 5). 37, Health Affairs homes compared to fee-for-service enrollees. Urban Institute. 2004. Classifying Medicare SNF patients: no. 1 (January): 78–85. Initial findings from three approaches . Report to the Centers National Investment Center for Seniors Housing & Care. 2018. for Medicare & Medicaid Services. Contract no. 500–00–0025. Skilled nursing data report . Annapolis, MD: National Investment Baltimore, MD: CMS. Center for Seniors Housing & Care. http://info.nic.org/skilled_ White, A., and Q. Zheng. 2018. Differences in resident case-mix data_report_pr. between Medicare and non-Medicare nursing home residents . Office of Inspector General, Department of Health and Human Report prepared by staff from Abt Associates for the Medicare Services. 2015. The Medicare payment system for skilled nursing Payment Advisory Commission. Washington, DC: MedPAC. . Report no. OEI–02–13–00610. facilities needs to be reevaluated https://innovation.cms.gov/Files/reports/bpci-models2-4- Washington, DC: OIG. yr2evalrpt.pdf. Office of the Actuary, Centers for Medicare & Medicaid Services, White, C., S. D. Pizer, and A. J. White. 2002. Assessing the RUG- Department of Health and Human Services. 2018a. Personal III resident classification system for skilled nursing facilities. communication of author with Christopher Truffer. October 11. 24, no. 2 (Winter): 7–15. Health Care Financing Review Office of the Actuary, Centers for Medicare & Medicaid Services, Winblad, U., V. Mor, J. P. McHugh, et al. 2017. ACO-affiliated Department of Health and Human Services. 2018b. Personal hospitals reduced rehospitalizations from skilled nursing facilities communication of author with James Hardesty. September 18. 36, no. 1 (January 1): Health Affairs faster than other hospitals. 67–73. Rahman, M., E. A. Gadbois, D. A. Tyler, et al. 2018. Hospital- skilled nursing facility collaboration: A mixed-methods approach Zhu, J. M., V. Patel, J. A. Shea, et al. 2018. Hospitals using to understanding the effect of linkage strategies. Health Services bundled payment report reducing skilled nursing facility use and (August 5). Research 37, no. 8 (August): improving care integration. Health Affairs 1282–1289. Ritchie, W., and D. Johnson. 2017. Skilled nursing facilities and REITs: A symbiotic relationship facing significant disruption. . Ziegler CFO Hotline. Chicago, Post-acute activity Ziegler. 2018. Cain Brothers’ Comments , April 28. IL: Ziegler. November 24. Skilled nursing facility services: Assessing payment adequacy and updating payments 224

255 CHAPTER 9 Home health care services

256 RECOMMENDATION 9 For 2020, the Congress should reduce the calendar year 2019 Medicare base payment rate for home health agencies by 5 percent. COMMISSIONER VOTES: YES 16 • NO 0 • NOT VOTING 0 • ABSENT 1

257 CHAPTER 9 Home health care services Chapter summary In this chapter Home health agencies (HHAs) provide services to beneficiaries who are • Are Medicare payments homebound and need skilled nursing or therapy. In 2017, about 3.4 million adequate in 2019? Medicare beneficiaries received care, and the program spent $17.7 billion on How should Medicare • home health care services. In that year, almost 12,000 HHAs participated in payments change in 2020? Medicare. Assessment of payment adequacy The indicators of payment adequacy for home health care are generally positive. Access to home health care is adequate: Over Beneficiaries’ access to care— 98 percent of beneficiaries lived in a ZIP code where an HHA operated in 2017, and 84 percent lived in a ZIP code with five or more HHAs. Capacity and supply of providers— The number of HHAs fell slightly • (by 3 percent) in 2017, but this decline follows a long period of growth in prior years. From 2004 to 2016, the number of HHAs increased by 60 percent. The decline in 2017 was concentrated in areas that experienced sharp increases in supply in prior years. Volume of services— From 2002 to 2016, home health utilization • increased substantially, with the number of episodes rising nearly 60 percent and the episodes per home health user climbing from 1.6 to 1.9 March 2019 | Report to the Congress: Medicare Payment Policy 227

258 episodes. In 2017, volume dropped 3.1 percent, the total number of fee-for- service users also fell slightly, and the average number of episodes per home health user declined by 1.4 percent. Episodes not preceded by a hospitalization accounted for most of the growth since 2002, increasing from about half of episodes in 2002 to two-thirds of episodes in 2017. • Marginal profit— In 2017, freestanding HHAs’ marginal profit—that is, the rate at which Medicare payments exceed providers’ marginal cost—was 17.5 percent, suggesting a significant financial incentive for HHAs to serve Medicare patients. Quality of care— In 2017, the rate of home health patients who were hospitalized or received treatment in the emergency room during an episode did not change significantly, similar to the trend in prior years, while measures of functional status, such as improvement in walking and transferring, increased. However, the functional status measures should be interpreted cautiously because these measures are based on provider-reported data and could be affected by agency coding practices. Providers’ access to capital— Access to capital is a less important indicator of Medicare payment adequacy for home health care because this sector is less capital intensive than other health care sectors. The major publicly traded for-profit home health companies had sufficient access to capital markets for their credit needs. Several acquisitions to increase capacity and expansion of capacity by publicly traded home health care firms indicate adequate access to capital. In 2017, the average all-payer margin for HHAs was 4.5 percent. In 2017, Medicare spending for home Medicare payments and providers’ costs— health care declined by 1.6 percent. However, between 2002 and 2016, spending increased by over 88 percent. For more than a decade, payments under the home health prospective payment system (PPS) have consistently and substantially exceeded costs. In 2017, Medicare margins for freestanding agencies averaged 15.2 percent. The projected margin for 2019 is 16 percent. Two factors have contributed to payments exceeding costs: Agencies have reduced episode costs by decreasing the number of visits provided, and cost growth in recent years has been lower than the annual payment updates for home health care. The high margins of freestanding HHAs have led the Commission to recommend a 5 percent reduction in the home health PPS base payment rate for 2020. However, this reduction will likely be inadequate to align Medicare payments with providers’ actual costs, and further reductions through rebasing will likely be necessary. In past years, the Commission has recommended that payments be rebased in the year Home health care services: Assessing payment adequacy and updating payments 228

259 following a payment rate reduction. However, given the congressionally mandated revisions to the home health PPS that are slated for 2020, our recommendation for 2020 addresses only the level of payment. The planned revisions to the home health PPS likely will alter the mix and level of services HHAs provide. Future rebasing should reflect the new patterns of care. Those data will not be available until mid- 2021. ■ March 2019 | Report to the Congress: Medicare Payment Policy 229

260

261 Home health care plays an important role in the care of Background Medicare beneficiaries. Home health can serve as an efficient substitute for or step down from institutional Medicare home health care consists of skilled nursing, post-acute care (PAC), helping to keep beneficiaries physical therapy, occupational therapy, speech therapy, in their homes and potentially reducing Medicare aide services, and medical social work provided to expenditures. Some new models of care—such as value- beneficiaries in their homes. To be eligible for Medicare’s based purchasing, the Hospital Readmission Reduction home health benefit, beneficiaries must need part-time Program (HRRP), and Medicare’s bundled acute care (fewer than eight hours per day) or intermittent skilled care demonstrations—encourage closer cooperation between to treat their illnesses or injuries and must be unable to home health agencies (HHAs) and other providers to leave their homes without considerable effort. In contrast improve care for beneficiaries. In the future, changes in to coverage for skilled nursing facility services, Medicare technology and new models of care may make it possible does not require a preceding hospital stay to qualify for to deliver more care in the home. However, establishing home health care. Also, unlike for most services, Medicare appropriate incentives and levels of payment in FFS does not require copayments or a deductible for home Medicare has proven challenging. health services. In 2017, about 3.4 million Medicare beneficiaries received home care, and the program spent Use and growth of the home health benefit $17.7 billion on home health services. Medicare spending has varied substantially with changes in for home health care more than doubled between 2001 coverage and payment policy and 2017, and this care accounted for about 3 percent of The home health benefit has changed substantially since Medicare fee-for-service (FFS) spending in 2016. the 1980s. Implementation of the inpatient hospital PPS in 1983 led to increased use of home health services as Medicare requires that a physician certify a patient’s hospital lengths of stay decreased. Medicare tightened eligibility for home health care and that a patient receiving coverage of some services, but the courts overturned these services be under the care of a physician. In 2011, curbs in 1988. After this change, the number of HHAs, Medicare implemented a requirement that a beneficiary users, and services expanded rapidly in the early 1990s. have a face-to-face encounter with the physician ordering Between 1990 and 1995, the number of annual users rose home health care. The encounter must take place in the 90 by 75 percent, and the number of visits more than tripled days preceding or 30 days following the initiation of home to about 250 million a year. Spending increased more than health care. Contacts through nonphysician practitioners fourfold between 1990 and 1995, from $3.7 billion to or authorized telehealth services may be used to satisfy the $15.4 billion. As the rates of use and the duration of home requirement. health spells grew, there was concern that the benefit was serving more as a long-term care benefit (Government Medicare pays for home health care in 60-day episodes. Accountability Office 1996). Further, many of the services Payments for an episode are adjusted to account for a provided were believed to be improper. For example, in patient’s clinical and functional characteristics and the one analysis of 1995 to 1996 data, the Office of Inspector number of therapy visits provided in the episode. If General found that about 40 percent of the services in beneficiaries need additional covered home health services a sample of Medicare claims did not meet Medicare at the end of the initial 60-day episode, another episode requirements for payment (mostly because services did not commences and Medicare pays for an additional episode. meet Medicare’s standards for a reasonable and necessary (An overview of the home health prospective payment service, patients did not meet the homebound coverage system (PPS) is available at http://www.medpac.gov/ requirement, or the medical record did not document that docs/default-source/payment-basics/medpac_payment_ a billed service was provided) (Office of Inspector General basics_18_hha_final_sec.pdf?sfvrsn=0.) Coverage for 1996). additional episodes generally has the same requirements as the initial episode (i.e., the beneficiary must be The trends of the early 1990s prompted increased program homebound and need skilled care). The Bipartisan Budget integrity actions, refinements of coverage standards, and Act of 2018 made significant changes to payments temporary spending caps through an interim payment for home health care services in 2020 (see text box on system (IPS). Between 1997 and 2000, the number of revisions to the home health PPS, pp. 232–233). March 2019 | Report to the Congress: Medicare Payment Policy 231

262 Revisions to the home health prospective payment system in 2020 for Medicare to move away from using therapy as a he Bipartisan Budget Act of 2018 requires CMS payment factor (U.S. Senate Committee on Finance to implement two major changes to the home 2011). Eliminating the thresholds in 2020 will mitigate health prospective payment system (PPS) in T the adverse incentives in the home health PPS. 2020: a new 30-day unit of payment in place of the current 60-day unit of payment and the elimination of CMS also plans to implement a new case-mix system, the number of therapy visits as a factor in the payment the Patient-Driven Groupings Model (PDGM), in 2020. system. These changes follow several years of analysis The PDGM categorizes episodes into 432 payment by the Commission and CMS to identify reforms to groups based on the following characteristics: home health payment. The elimination of the therapy thresholds is consistent with a recommendation the Services in the first 30 days of a Episode timing— • Commission first made in 2011 (Medicare Payment spell of home health would be classified as “early,” Advisory Commission 2011). while services in the subsequent 30-day period would be classified as “late.” For example, if a The current payment system has a series of nine beneficiary had two consecutive 60-day payment payment thresholds that increase payment as the episodes under the current system, the first 30-day number of therapy visits in an episode increases; period would be classified as early, while the three in effect, providing more therapy visits increases subsequent 30-day periods would be classified as payments. Such an adjustment encourages agencies to late 30-day periods. Though the unit of payment consider financial incentives when providing therapy will move to a 30-day episode, beneficiaries services. The Commission has noted that home health receiving home health care would continue to be agencies (HHAs) appear to adjust their services to assessed for payment purposes at the beginning of maximize financial results under these thresholds care and at the beginning of each subsequent 60- (Medicare Payment Advisory Commission 2011). An day period of service. investigation by the U.S. Senate Committee on Finance found that many agencies were targeting therapy Referral source— • Cases would be categorized services based on financial incentives and called based on the services received before the beginning (continued next page) beneficiaries using home health services fell by about from 3.9 million to 6.3 million (data not shown). In 2017, 1 million, and the number of visits fell by 65 percent the number of HHAs was 11,844, higher than the level of (Table 9-1, p. 234). The mix of services changed from supply during the 1990s. Almost all the new agencies since predominantly aide services in 1997 to predominantly implementation of the PPS have been for-profit providers skilled nursing visits in 2000, and therapy visits increased (data not shown). between 1997 and 2000 from 10 percent to 19 percent of The steep declines in services under the IPS did not visits. Between 1997 and 2000, total spending for home appear to adversely affect the quality of care beneficiaries health services declined by 52 percent. The reduction in received; one analysis found that patient satisfaction with payments had a swift effect on the supply of HHAs, and home health services was mostly unchanged in that period by 2000, the number of HHAs had fallen by 31 percent. (McCall et al. 2004, McCall et al. 2003). In 2004, the However, after the PPS was implemented in 2000, service Commission also concluded that the quality of care did not use and agency supply rebounded at a rapid pace. Between decline between use of the IPS and the implementation of 2001 and 2017, the number of home health episodes rose Home health care services: Assessing payment adequacy and updating payments 232

263 Revisions to the home health prospective payment system in 2020 (cont.) CMS analyzed the PDGM’s likely impact in the 2019 of the episode: prior hospitalization or institutional home health payment rule, finding that, in general, post-acute care on the one hand, or admission from funds would be redistributed from HHAs that provide the community on the other. more therapy to those that provide relatively more Clinical category— • The new system would create nursing. Specifically: 12 clinical categories based on patients’ reported Payments in 2020 would increase by 2.9 percent • conditions or treatments: need for musculoskeletal for nonprofit agencies and 3.9 percent for facility- rehabilitation; neuro/stroke rehabilitation; wound based HHAs. care; behavioral health care; complex care; and medication management, teaching, and assessment. • Payments would fall by 1.2 percent for freestanding agencies and fall by 2.2 percent for • Similar to the Functional/cognitive level— for-profit HHAs. existing system, the PDGM would classify patients’ cognitive and physical functioning using HHAs in urban areas would see a 0.6 percent • information from the Outcomes Assessment payment decrease, while those in rural areas would Information Set, known as OASIS, home health see a 4.0 percent increase. patient assessment. • Payments would rise for smaller providers and fall The PDGM • Presence of comorbidities— for larger providers. For example, payments would will adjust payment for commonly occurring increase by 1.9 percent for the 2,841 HHAs with comorbidities in home health care. There less than 100 episodes in annual volume and would would be a three-tiered adjustment for selected drop 0.2 percent for larger HHAs with more than a comorbidities. 1,000 episodes a year. ■ the PPS (Medicare Payment Advisory Commission 2004). per episode between 1998 and 2001 would decline about The similarity in quality of care under the IPS and the 15 percent, while the actual decline was about 32 percent PPS suggests that the payment reductions in the Balanced (Table 9-2, p. 235). Between 2001 and 2017, the number Budget Act of 1997 led HHAs to reduce costs and of visits per episode declined. The number of therapy utilization without a measurable difference in the quality services per episode increased, but this increase has been of patient care. more than offset by the decline in all other service types (nursing, home health aide, and medical social services). Medicare has always overpaid for home In addition, HHAs have been able to hold the rate of health services under the PPS episode cost growth below 1 percent in many years, lower than the rate of inflation assumed in the home health Payments for home health care have substantially payment update (data not shown). Consequently, HHAs exceeded costs since Medicare established the PPS. In were able to garner extremely high average payments 2001, the first full year of the PPS, average Medicare relative to the cost of services provided. Between 2001 margins for freestanding HHAs equaled 23 percent (Figure 1 and 2016, freestanding HHA margins averaged 16.3 9-1, p. 235). The high margins in the first year suggest percent (Figure 9-1, p. 235). that the PPS established a base rate well in excess of costs. The base rate assumed that the average number of visits March 2019 | Report to the Congress: Medicare Payment Policy 233

264 TABLE Changes in supply and utilization of home health care, 1997–2017 9–1 Percent change 1997– 2016– 2000– 2016 1997 2017 2000 2000 2017 2016 –3% Home health agencies 10,917 7,528 12,204 11,844 –31% 62% $8.5 $17.7 Total spending (in billions) –52 –2 113 $17.7 $18.1 38 –2 3.6 2.5 3.4 3.4 –31 Users (in millions) 20 –65 104.8 108.3 90.6 258.2 Number of visits (in millions) –3 Visit type (percent of total) 20 48% 49% 49% 41% Skilled nursing –2 –1 –11 48 31 10 9 –37 –68 Home health aide Therapy 101 112 5 43 41 19 10 1 1 1 Medical social services < –0.1 –25 1 1 –2 73 Number of visits per user –15 –49 31 31 37 Percent of FFS beneficiaries who 22 9.4 8.8% 8.9% 7.4% used home health services –1 10.5% Note: FFS (fee-for-service). Medicare did not pay on a per episode basis before October 2000. Yearly figures presented in the table are rounded, but figures in the percent change columns were calculated using unrounded data. Source: Home health standard analytical file 2017; , Medicare and Medicaid Statistical Supplement 2002. Health Care Financing Review 9-3, p. 236). In addition, the annual payment update offset Reductions mandated in 2010 legislation these reductions. The cumulative effect of the PPACA have not significantly lowered payment for home health services reduction and the payment update resulted in a payment reduction of 2.6 percent for the four years of rebasing. In 2010, the Commission recommended that Medicare This modest decrease is smaller than the payment lower home health payments to make them more reductions the industry has weathered in the past; since consistent with costs, a process referred to as payment the implementation of PPS in 2000, Medicare margins for rebasing. The Patient Protection and Affordable Care Act freestanding HHAs have always exceeded 10 percent. of 2010 (PPACA) included several reductions intended to address home health care’s high Medicare payments, Ensuring appropriate use of home health including rebasing the payment system. However, these care is challenging policies have not achieved the goal of making payments Policymakers have long struggled to define the role of more consistent with actual costs. the home health benefit in Medicare (Benjamin 1993). PPACA offset the annual rebasing adjustment by the From the outset, there was a concern that setting a narrow 2 payment update for each year from 2014 through 2017. policy could result in beneficiaries using other, more CMS set the rebasing reduction to the maximum amount expensive services, while a policy that was too broad permitted under the PPACA formula, which was equal to could lead to wasteful or ineffective use of the home 3.5 percent of the 2010 base rate, or an annual reduction health benefit (Feder and Lambrew 1996). Medicare 3 of $81 per 60-day episode. However, the size of the base relies on the skilled care and homebound requirements as rate has increased since 2010, so this reduction averaged primary determinants of home health eligibility, but these 2.8 percent in each year from 2014 through 2017 (Table broad coverage criteria permit beneficiaries to receive Home health care services: Assessing payment adequacy and updating payments 234

265 FIGURE FIGURE Medicare margins... X-X Medicare margins of freestanding home health agencies 9–1 remained high between 2001 and 2016 25 20 15 10 Medicare margin (in percent) 5 0 2013 2014 2015 2016 2011 2010 2012 2002 2003 2004 2005 2006 2007 2008 2009 2001 Medicare cost reports. Source: to consider alternatives to home health care, such as services in the home even though they are capable of outpatient services. Beneficiaries who meet program leaving home for medical care, which most home health Note: Note and Source are in InDesign. coverage requirements can receive an unlimited number beneficiaries do (Wolff et al. 2008). Medicare does not Source: of home health episodes and face no cost sharing. In provide any incentives for beneficiaries or providers TABLE Medicare visits per episode before and after implementation of PPS 9–2 Notes about this graph: • Data is in the datasheet. Make updates in the datasheet. Visits per episode Percent change in: • WATCH FOR GLITCHY RESETS WHEN YOU UPDATE DATA!!!! Type of visit 1998 2001 2016 2017 1998–2001 2001–2016 2016–2017 • The column totals were added manually. –3% 8.6 –25% –18% Skilled nursing 14.1 10.5 8.4 • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. Therapy (physical, occupational, 5 and speech–language pathology) 3.8 5.2 7.3 7.7 39 40 • I can’t delete the legend, so I’ll just have to crop it out in InDesign. –68 Home health aide 13.4 5.5 1.8 1.6 –59 –11 • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph –0.2 Medical social services 0.3 0.2 0.1 0.1 –36 –33 default when you change the data. 17.8 –0.2 –17 –32 17.8 Total 31.6 21.4 • Use paragraph styles (and object styles) to format. Note: PPS (prospective payment system). The PPS was implemented in October 2000. Data exclude low-utilization episodes. Yearly figures presented in the table are • Data was from: R:\Groups\MGA\data book 2007\data book 2007 chp1 rounded, but figures in the percent change columns were calculated using unrounded data. Source: Home health standard analytic file. March 2019 | Report to the Congress: Medicare Payment Policy 235

266 TABLE Impact of PPACA rebasing on payments for 60-day episodes 9–3 Annual percent change Cumulative change, 2014–2017 2014 2015 2016 2017 Rebasing adjustment –2.8% –2.8% –2.8% –2.8% –10.7% Legislated payment update 2.3 2.1 1.9 2.5 9.1 –0.8 Net payment reduction –1.0 –0.9 –2.6 –0.6 PPACA (Patient Protection and Affordable Care Act of 2010). Effects of payment changes are multiplicative. Note: MedPAC analysis based on home health prospective payment system final rules for 2014 through 2017 from CMS. Source: addition, the program relies on HHAs and physicians to for 10 months. Under the demonstration, Medicare follow program requirements for determining beneficiary conducted a full review of all home health claims in the needs, but evidence from prior years suggests that they state. HHAs were required to submit documentation do not consistently follow Medicare’s standards (Cheh indicating that the beneficiary met program coverage et al. 2007, Department of Health and Human Services standards when filing an initial request for payment. 2018, Office of Inspector General 2001). Concerns about Payment would be reduced by 25 percent for episodes for ensuring the appropriate use of home health episodes which HHAs did not submit supporting documentation not preceded by a hospitalization led the Commission to with the initial claim. Though most claims were approved, recommend a copayment for these episodes (Medicare the Government Accountability Office (GAO) found Payment Advisory Commission 2011). that payments dropped by about $100 million for the 10 months the demonstration was in effect. CMS Program integrity is a continuing challenge suspended the demonstration in March 2017 (Government in home health care Accountability Office 2018). In 2010, the Commission made a recommendation CMS recently announced plans to implement a revised to curb wasteful and fraudulent home health services version of the demonstration. Under the revised (Medicare Payment Advisory Commission 2010). This demonstration, HHAs in Illinois will have the option of recommendation calls on the Health and Human Services submitting additional supporting documentation before Secretary to use the department’s authorities under or after payment. HHAs that have acceptable affirmation current law to examine providers with aberrant patterns of rates during the review process will be released from utilization for possible fraud and abuse. PPACA permits the review requirement. HHAs that do not submit any Medicare to implement temporary moratoriums on the documentation during the demonstration will have enrollment of new HHAs in areas believed to have a their payments reduced by 25 percent and possibly be high incidence of fraud. In 2017, Medicare implemented subject to postpayment review. CMS plans to expand the statewide moratoriums for HHAs in Florida, Illinois, demonstration to Florida, North Carolina, Ohio, and Texas Michigan, and Texas, expanding previously established after gaining experience in Illinois. local moratoriums in these states. There have also been numerous criminal prosecutions for home health fraud, most notably in Miami and Detroit. Are Medicare payments adequate in CMS has experimented with prepayment claims review 2019? as a means to reduce inappropriate billing. In 2016, CMS operated the Pre-claim Review Demonstration in Illinois The Commission reviews several indicators to determine the level at which payments will be adequate to cover Home health care services: Assessing payment adequacy and updating payments 236

267 TABLE Number of participating home health agencies declined 9–4 in 2017 but remained high relative to earlier years Percent change 2017 2016 2015 2012 2008 2004 2004–2016 2016–2017 60% –3.0% Active home health agencies 7,651 11,844 12,311 12,204 12,346 9,787 Number of home health agencies 3.2 3.2 3.3 2.8 2.1 3.1 51 –2.9 per 10,000 FFS beneficiaries Note: FFS (fee-for-service). “Active home health agencies” includes all agencies operating during a year, including agencies that closed or opened at some point during the year. Source: CMS’s Provider of Service file and 2018 annual report of the Boards of Trustees of the Medicare trust funds. the costs of an efficient provider in 2019. We assess criminal investigations and moratoriums on the entry of beneficiary access to care by examining the supply of new HHAs. The number of HHAs exiting the program has home health providers, annual changes in the volume of increased in recent years in these states, and moratoriums services, and marginal profit. The review also examines have likely stopped the entry of new HHAs. Even with quality of care, access to capital, and the relationship declines in these states, however, the supply of HHAs in between Medicare’s payments and providers’ costs. the two states is almost three times the supply of HHAs Overall, the Medicare payment adequacy indicators for that were available in 2001, with supply exceeding 3,400 HHAs are positive. HHAs in 2017. From 2004 to 2017, the number of HHAs per 10,000 FFS Beneficiaries’ access to care: Almost all beneficiaries rose 46.6 percent, from 2.1 to 3.1 (Table beneficiaries live in an area served by HHAs 9-4). Most of the new HHAs were for profit. However, Supply and volume indicators show that almost all supply varies significantly among states. In 2017, Texas beneficiaries have access to home health services. In 2017, averaged 8.7 HHAs per 10,000 beneficiaries, while over 98 percent of beneficiaries lived in a ZIP code served New Jersey averaged less than 1.0 HHA per 10,000 by at least one HHA, 97 percent lived in a ZIP code served beneficiaries. The extreme variation demonstrates that by two or more HHAs, and 84 percent lived in a ZIP the number of providers is a limited measure of capacity code served by five or more agencies. These findings are because HHAs can vary in size. Also, because home 4 consistent with our prior reviews of access. health care is not provided in a medical facility, HHAs can adjust their service areas as local conditions change. Supply of providers: Agency supply remains high Even the number of employees may not be an effective despite recent decline metric because HHAs can use contract staff to meet their Though the supply of HHAs declined slightly in 2017, patients’ needs. supply still remains relatively high. Since 2004, the number of HHAs in Medicare has increased by over 4,000 Episode volume declined slightly in 2017 agencies, reaching 11,844 agencies in 2017 (Table 9-4). Episode volume in 2017 declined by 3.1 percent (Table The slight decline in 2017 was concentrated in Florida 9-5, p. 238). This decline is part of a trend that began in and Texas, states that experienced higher than average 2012, but this period of decline was preceded by a period increases in supply in prior years. These states have been of rapid growth (Figure 9-2, p. 239). Between 2002 and targeted by a myriad of antifraud measures, including 2011, total episodes increased by 67 percent from 4.1 March 2019 | Report to the Congress: Medicare Payment Policy 237

268 TABLE Fee-for-service home health care services have increased significantly since 2002 9–5 Percent change 2016– 2002– 2011 2013 2017 2016 2016 2002 2017 –1.7% Home health users (in millions) 3.4 3.5 3.5 3.4 37.5% 2.5 8.8% 8.9% % 9.2 % 9.4 24.0 –1.4 Share of beneficiaries using home health care 7.2% Episodes (in millions): 4.1 6.8 6.7 –3.1 6.5 6.3 59.4 –1.4 Per home health user 1.6 1.9 1.9 1.9 16.0 2.0 0.19 0.12 Per FFS beneficiary 0.16 44.1 –2.8 0.18 0.17 $9.5 $18.3 $18.0 $17.7 88.8 –1.6 Payments (in billions) $17.8 5,312 0.1 5,196 5,202 37.4 3,783 Per home health user 5,132 2,916 2,645 Per home health episode 2,988 3,039 13.0 1.4 2,896 FFS (fee-for-service). Percent change is calculated on numbers that have not been rounded; payment per episode excludes low-utilization payment adjustment cases. Note: MedPAC analysis of home health standard analytical file. Source: been taken to curb fraud, waste, and abuse in Medicare million episodes to 6.8 million episodes (Table 9-5). The home health care. decline since 2011 has been concentrated in a few states, with five states (Florida, Illinois, Louisiana, Tennessee, The decline in episode volume since 2011 has not been and Texas) accounting for most of the decline in episodes. uniform across the country. Since 2011, Florida, Illinois, However, utilization in these five states had more than Louisiana, Tennessee, and Texas (the five states with the doubled between 2002 and 2011, higher than in most other fastest growing episode volume before 2011) have seen areas (Figure 9-2). a decline of about 25 percent. The remaining 44 states experienced aggregate growth of 4.1 percent, though Changes in average payment per full episode (defined there was a range of increases and declines across these as comprising more than four visits) underscore the states. This geographic variation emphasizes that many limited impact of the PPACA rebasing policy that was 5 areas continue to see growth despite the overall drop implemented in 2014 through 2017. The average payment in episode volume since 2011. The volume decrease per episode in 2017 was 5 percent higher than the average in areas that have been targeted by program integrity payment per episode for 2013, the year before the PPACA efforts suggests that these efforts can address excessive or adjustments were implemented (Table 9-5). The per unwarranted services, and the expansion of these efforts episode payment growth is even more remarkable since to other areas with excessive growth rates would be Medicare implemented additional payment reductions beneficial. during this period, such as reductions for changes in coding practices. Home health care spells of service have increased The decline in home health utilization since 2011 reflects in length and shifted in focus to episodes not changes in both the demand for home health services and preceded by a hospitalization the supply of HHAs. The number of hospital discharges, Between 2002 and 2016, the average number of episodes a common source of referrals, declined by 11 percent per user increased by 16 percent, rising from 1.6 to 1.9 from 2011 to 2014 and has not changed significantly episodes per user (Table 9-5). Though the average number since, indicating that demand for PAC services has not of episodes declined slightly in 2017, the trend since 2002 increased since 2011. In addition, several actions have Home health care services: Assessing payment adequacy and updating payments 238

269 FIGURE FIGURE Title here... 1-XX Volume of fee-for-service home health care 9–2 services have increased significantly since 2002 160 5 states with largest decline in volume since 2011 140 All other states and territories Total 120 100 80 60 40 Cumulative change (in percent) 20 0 2016 2017 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2002 2015 2014 Note: The five states with the largest decline in volume since 2011 include Florida, Illinois, Louisiana, Tennessee, and Texas. MedPAC analysis of home health standard analytic file from CMS. Source: Note: Note and Source are in InDesign. indicates that beneficiaries have been receiving home share of episodes not preceded by inpatient or institutional health care for longer periods. The increase in episodes PAC, which in 2017 accounted for 66 percent of episodes. Source: coincides with Medicare’s PPS incentives that encourage Episodes that qualify for additional payment additional volume: The per episode unit of payment in the Notes about this graph: based on therapy services account for an PPS encourages longer periods of home health use (more increasing share of volume episodes per beneficiary). • Data is in the datasheet. Make updates in the datasheet. Since the October 2000 implementation of the home • I deleted the years from the x-axis and put in my own. The rise in the average number of episodes per home health PPS, Medicare has used the number of therapy health user coincides with a relative shift away from using • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. visits as a factor in payment, and, not surprisingly, home health care after a hospitalization or institutional episodes that qualify for these payments have increased • The dashed line looked ok here, so I didn’t hand draw it. PAC service. Between 2001 and 2011, episodes not faster than episodes that do not. Under the current PPS, • I can’t delete the legend, so I’ll just have to crop it out in InDesign. preceded by a hospitalization or institutional PAC stay additional therapy visits increase payments once six or increased by about 127 percent, while episodes preceded • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph more visits are provided in an episode, and the share of by a prior PAC stay or hospitalization increased by 14.8 these episodes increased between 2008 and 2017 from 37 default when you change the data. percent (Table 9-6, p. 240). Between 2011 to 2017, percent to 49 percent. In past work, the Commission has • Use paragraph styles (and object styles) to format. the volume of episodes not preceded by a hospital or found that agencies that provide more therapy episodes institutional PAC stay dropped by 11.2 percent, while in tend to be more profitable. The higher profitability and the same period, episodes preceded by a hospitalization rapid growth in the number of these episodes suggest that or PAC stay continued to increase slightly in recent years. financial incentives are causing agencies to favor therapy However, this increase has not significantly changed the March 2019 | Report to the Congress: Medicare Payment Policy 239

270 TABLE Home health episodes not preceded by hospitalization or 9–6 PAC stay increased at a higher rate than other episodes Episodes Cumulative percent change 2001 2001–2011 2011 2017 2011–2017 Number of episodes preceded 2.2 by a hospitalization or PAC stay (in millions) 1.9 2.2 14.8% 2.2% Number of episodes not preceded by a hospitalization or PAC stay (in millions) 2.1 4.6 4.2 127.4 –11.2 Share of episodes not preceded 66% 26 67% –1.5 53% by a hospitalization or PAC stay –7.0 Total (in millions) 3.9 6.8 6.3 74.0 Note: PAC (post-acute care). “Episodes preceded by a hospitalization or PAC stay” indicates the episode occurred fewer than 15 days after a stay in a hospital (including a long-term care hospital), skilled nursing facility, or inpatient rehabilitation facility. “Episodes not preceded by a hospitalization or PAC stay” indicates that there was no hospitalization or PAC stay in the 15 days before the episode began. Numbers may not sum to totals due to rounding. Source: Home health standard analytical file and Medicare Provider and Analysis Review file for 2001, 2011, and 2017. services when possible. Consistent with this finding, the and one-half percent in 2020. CMS computed the Bipartisan Budget Act of 2018 requires CMS to remove ratio of home health episodes to FFS beneficiaries in therapy visits as a factor in determining payments under 2015 for all counties (both urban and rural); based the PPS. on this distribution, rural counties with 17.8 or more episodes per 100 beneficiaries were classified as high- New rural payment targets supplemental utilization areas. payments to low-use rural areas Low-population rural counties— Counties that have • In general, the Commission has not found systemic issues a population density of six individuals or fewer per with rural access to care, and Medicare margins of rural square mile and do not have high utilization are HHAs are generally above 10 percent a year, comparable classified as low-population counties and receive a 4 with urban HHAs. Average utilization is not significantly percent add-on in 2019. The add-on will decrease by 1 different between HHAs in urban and rural areas, but some percentage point each year and end after 2022. variation exists around this average, with high-use and low-use areas found in both urban and rural counties. All other rural counties— • Rural counties not in either of the above categories will receive a 3 percent add-on The Bipartisan Budget Act of 2018 implemented a 3 in 2019, also decreasing by 1 percentage point each percent payment increase for home health episodes year to end after 2021. provided in rural areas in 2018. For later years, the Act establishes three categories of rural counties and ties the The rural payment add-on policy for 2019 better targets duration and size of the payment add-on for each category Medicare’s scarce resources. The policy targets payments to the population density and utilization levels of rural to areas with lower population density and limits payments counties. The categories include: to rural areas with higher utilization. This policy is consistent with our June 2012 report to the Congress, Services furnished High-utilization rural counties— • which noted that Medicare should target rural payment in rural counties in the top quartile of utilization will adjustments to areas that may have access challenges. receive a payment add-on of 1.5 percent in 2019 Home health care services: Assessing payment adequacy and updating payments 240

271 TABLE Average home health agency performance on select quality measures 9–7 2014 2015 2016 2017 During an episode, the share of an agency’s beneficiaries who: Used emergency department care 12.0% 12.2% 12.1% 12.7% 15.5 Had to be admitted to the hospital 15.4 16.2 15.4 Share of an agency’s beneficiaries with improvement in: 59% 72% 65% 55% Transferring 63 Walking 74 61 69 All data are for fee-for-service beneficiaries only and are risk adjusted for differences in patient condition among home health patients. Note: Source: MedPAC analysis of data provided by the University of Colorado. rely on patient assessment data, particularly because the Marginal profits measures showing improvement are most sensitive to Another factor we consider when evaluating access to HHA coding practices. The use of patient assessment data care is whether providers have any financial incentive to to determine Medicare payments may also distort these expand the number of Medicare beneficiaries they serve. data, as in some cases more severe debility in function can In considering whether to treat a patient, a provider with yield higher payments. excess capacity compares the marginal revenue it will receive (i.e., the Medicare payment) with its marginal A comparison of trends for 2014 to 2017 illustrates these costs—that is, the costs that vary with volume. If Medicare concerns (Table 9-7). Measures of hospitalization and payments are larger than the marginal costs of treating emergency department use rely on Medicare claims; an additional beneficiary, a provider has a financial these measures indicate mixed or modest changes with incentive to increase its volume of Medicare patients. In no substantial change over this period. In contrast, the contrast, if payments do not cover the marginal costs, the rates of beneficiaries’ functional improvement have risen provider may have a disincentive to care for Medicare substantially, with the share of beneficiaries improving in 6 beneficiaries. In 2017, the marginal profit, on average, transferring and walking increasing 17 percentage points for freestanding HHAs was 17.5 percent. This substantial and 13 percentage points, respectively, over the four-year marginal profit indicates that these HHAs have an period. The higher rates of improvement for the functional incentive to serve Medicare beneficiaries. measures may reflect agency coding practices and should be interpreted cautiously. Quality of care: Divergent trends between claims-based and provider-reported It is also notable that functional improvement data are measures collected only for beneficiaries who do not have their The Commission relies on data from two principal sources home health care stays terminated by a hospitalization, to measure home health care quality: data from patient which means that beneficiaries included in the measure assessment information collected by HHA staff, and may be healthier and more likely to have positive Medicare claims submitted by HHAs and other provider outcomes. 7 types. The Commission has observed that performance Medicare’s home health value-based purchasing for quality measures that rely on claims has not changed program had a limited impact in the first year significantly in 2014 to 2017, while performance for measures that rely on patient assessment data reported by In 2017, Medicare initiated a value-based purchasing HHAs has improved significantly. These divergent trends (VBP) model for home health care. The model is raise concerns about the objectivity of the measures that designed to test whether HHAs in nine states (Arizona, March 2019 | Report to the Congress: Medicare Payment Policy 241

272 Florida, Iowa, Maryland, Massachusetts, Nebraska, Providers’ access to capital: Access to capital North Carolina, Tennessee, and Washington) improve for expansion is adequate or maintain high quality when they are subject to a VBP In 2017, the overall (all-payer) margins for freestanding incentive. Under the demonstration, HHAs with higher HHAs averaged 4.5 percent, indicating that many HHAs performance receive bonuses, while those with lower yield positive financial results that should appeal to scores receive penalties. HHA performance is evaluated capital markets. HHAs are not as capital intensive as other against separate improvement and attainment scores, providers because they do not require extensive physical with payment tied to the higher of these two scores. The infrastructure, and most are too small to attract interest first payment adjustment was implemented January 1, from capital markets. Few HHAs access capital through 2018. publicly traded shares or through public debt such as issuance of bonds. The program determines quality through a composite of 20 measures of process, outcomes, and patient satisfaction. In Information on publicly traded home health care the first year, performance in 2016 was compared with the companies provides some insight into access to capital but prior year. The scores are combined into a composite Total has limitations. Publicly traded companies may have other Performance Score (TPS), following an approach similar lines of business in addition to Medicare home health care, to that used in the hospital VBP program. such as hospice, Medicaid-covered services, and private- duty nursing. Also, publicly traded companies are a small A CMS-contracted report concluded that the impact of portion of the total number of HHAs in the industry. the VBP program on quality was mixed in 2017, the first However, since they are the largest corporate entities in year payments were adjusted under the program. This home health care, they can provide some insight about the analysis compared the TPS values for patients served by industry’s financial status. HHAs in the nine VBP test states with a comparison group of home health patients from other states. The analysis Analysis of for-profit companies indicates that these found that TPS values increased in both VBP states and companies had adequate access to capital in 2017. Publicly the comparison states, but that the increase in the average traded firms continued to invest in home health capacity. TPS value was 7.4 percent greater in the VBP states. The For example, LHC Group merged with Almost Family. report made a caveat with reference to this result, noting Encompass (formerly known as HealthSouth) acquired that the higher annual increase for the VBP states was a multistate hospice and home health company. These principally attributable to better performance on several capacity-driven expansions by publicly traded companies process and outcome measures based on the Outcomes suggest that access to capital remains adequate. Assessment Information Set (OASIS), which are reported by HHAs and not independently verified. The VBP report Medicare payments and providers’ costs: also noted that several HHAs subject to VBP had revised Payments rose while cost per episode patient assessment practices in response to the program remained low in 2017 and that changes in these practices could have contributed In 2017, average Medicare payments per episode increased to the higher rates of improvement on the OASIS-based by 1.4 percent for freestanding HHAs. Meanwhile, low or measures. no cost growth has been typical for home health care, and in some years, cost per episode has declined. In 2017, the Performance on other measures, which relied on data average cost per episode increased by 0.9 percent, slightly from Medicare claims or beneficiary surveys, was mixed. greater than the annual decrease of about 0.1 percent for the Spending on and utilization of skilled nursing facilities previous five years. The ability of freestanding HHAs to keep declined slightly, but emergency department use increased. costs low in most years has contributed to their high margins There was no statistically significant difference between under the Medicare PPS. In 2017, Medicare accounted for VBP states and non-VBP states in rates of hospitalization about 56 percent of revenue for freestanding HHAs. or spending. The analysis also found no statistically meaningful differences between VBP states and non-VBP Medicare margins for freestanding HHAs states for the rates of patient satisfaction collected through remained high in 2017 the Home Health Consumer Assessment of Healthcare In 2017, HHA Medicare margins in aggregate were 15.2 ® Providers . percent for freestanding HHAs (Table 9-8). For these Home health care services: Assessing payment adequacy and updating payments 242

273 TABLE Medicare margins for freestanding home health agencies, 2016 and 2017 9–8 Medicare margin Share of home health Share of agencies, 2017 2017 episodes, 2017 2016 15.2% 15.5% 100% 100% All Geography Majority urban 16.0 15.8 83 84 Majority rural 16 17 13.4 13.8 Type of ownership 79 For profit 16.8 16.4 88 Nonprofit 10.9 12 21 12.0 Volume quintile First (smallest) 8.5 7.4 20 3 Second 20 9.8 10.8 6 Third 11.5 20 11 11.6 Fourth 14.5 13.6 20 19 Fifth (largest) 17.4 17.0 20 62 Note: Home health agencies were classified as majority urban if they provided more than 50 percent of episodes to beneficiaries in urban counties and were classified as majority rural if they provided more than 50 percent of episodes to beneficiaries in rural counties. Components may not sum to totals due to rounding. MedPAC analysis of home health cost report files from CMS. Source: help their parent institutions financially if they can shorten HHAs, the aggregate Medicare margins varied from 0.7 inpatient stays, lowering expenses in the most costly percent for those at the 25th percentile of the margin setting. distribution to 24.1 percent for those at the 75th percentile (not shown in Table 9-8). For-profit HHAs had higher HHAs’ financial performance in 2016 and 2017 permits margins than nonprofit HHAs, and urban HHAs had an examination of the financial impact of the third and slightly higher margins than rural HHAs. Agencies with fourth years of rebasing under PPACA. In both years, the higher volume had better financial results, likely reflecting margins for freestanding HHAs remained high, reflecting the economies of scale possible for larger operations. the Commission’s concerns that the PPACA policy would For example, HHAs in the bottom quintile of Medicare not make sufficient reductions. The actual performance volume had margins of 7.4 percent while HHAs in the top contrasts starkly with the home health industry’s quintile had margins of 17.0 percent. predictions. In 2013, the industry predicted that Medicare margins for freestanding agencies in 2016 would be –0.3 The Commission includes hospital-based HHAs in its compared with the actual aggregate margins of 15.5 calculation of acute care hospitals’ Medicare margins percent. because these agencies operate in the financial context of hospital operations. In 2017, margins for hospital-based Relatively efficient HHAs serve patients similar to HHAs were –16.0 percent. The lower margins of hospital- patients of all other HHAs based HHAs are attributable chiefly to their higher costs, some of which are a result of overhead costs allocated to Across all health care sectors, the Commission applies the HHA from its parent hospital. Hospital-based HHAs a two-step process when identifying efficient providers. First, the providers must do relatively well across cost March 2019 | Report to the Congress: Medicare Payment Policy 243

274 TABLE Performance of relatively efficient home health agencies in 2016 9–9 Relatively efficient All other providers All Provider characteristics providers 318 4,286 Number of home health agencies 4,604 Share that are for profit 89% 89% 89% Median: Medicare margin % % % 24.4 15.6 15.0 Hospitalization during first 60 days of stay (percent) 16.5% 14.7% 16.7% Cost per episode $2,409 $2,056 $2,445 1.02 0.99 Patient severity case-mix index 0.99 Visits per episode 16.6 Average visits per episode 16.8 15.4 Share of visits by type 48% 47% 47% Skilled nursing visits 7% 8% 8% Aide visits 1% 1% 1% MSS visits 43% 44% 44% Therapy visits Size (number of 60-day payment episodes) 653 504 494 Median 905 1,399 868 Mean Share of episodes 8% 8% 9% Low-use episode 3% 3% 3% Outlier episode 15% 23% 25% Share of episodes provided to rural beneficiaries Note: MSS (medical social services). Sample includes freestanding agencies with complete data for three consecutive years (2014–2016). A home health agency is classified as relatively efficient if it is in the best third of performance for quality or cost and is not in the bottom third of either measure for three consecutive years. Low-use episodes are those with 4 or fewer visits in a 60-day episode. Outlier episodes are those that receive a very high number of visits and qualify for outlier payments. Components may not sum due to rounding. Medicare cost reports and standard analytic file. Source: and quality metrics. Second, the performance has to be basis, adjusted for risk (patient’s health status) and local consistent, meaning that the provider cannot have poor wages; the quality measures were risk-adjusted rates performance on any metric over a three-year period. The of hospitalizations and improvement in walking. Our Commission’s approach is to develop a set of criteria and approach categorized an HHA as relatively efficient if it then examine how many providers meet them. It does not was in the best performing third on at least one measure establish a set share of providers to be considered efficient (low cost per episode, a low hospitalization rate, or a high and then define criteria to meet that pool size. rate of beneficiaries showing improvement in walking) and was not in the worst performing third of any of these We examined the quality and cost efficiency of measures for three consecutive years (2014 to 2016). freestanding HHAs to identify a cohort that demonstrated About 7 percent of freestanding HHAs met these criteria better performance on these metrics relative to its peers in this period. 8 (Table 9-9). The cost measure was on a per episode Home health care services: Assessing payment adequacy and updating payments 244

275 of in institutional settings, and home health care can be In 2016, relatively efficient agencies compared with other provided at lower costs than institutional care. However, HHAs had median margins that were about 9 percentage points higher, a median hospitalization rate that was 2 Medicare’s payments for home health services are too percentage points lower, and a median cost per episode high, and these overpayments diminish the service’s value that was 16 percent lower. Relatively efficient HHAs as a substitute for more costly services. There are also some indications that utilization under fee-for-service is provided more episodes but 1.4 fewer visits per episode. The mix of nursing, therapy, aide, and social services visits not always efficient, as suggested by the broad geographic did not differ significantly between relatively efficient and variation in the use of the benefit. In another example, a recent analysis of home health care utilization in the other HHAs. Efficient providers tended to provide fewer Medicare’s Shared Savings Program found that utilization episodes in rural areas. dropped significantly for patients enrolled in a Medicare The Commission estimates that Medicare accountable care organization (McWilliams et al. 2017). margins will remain high in 2019 The Bipartisan Budget Act of 2018 requires that the policy In modeling 2019 payments and costs, we incorporate changes implemented in 2020 be budget neutral and policy changes that will go into effect between the year of provides CMS with the authority to adjust payments in our most recent data, 2017, and the year for which we are 2020 through 2026 to maintain budget neutrality. CMS has making the margin projection, 2019. The major changes projected that behavioral responses by HHAs to the new are: policies will increase payments by 6.42 percent in 2020 (about $1 billion), and the agency plans to implement an • a 1 percent payment update for 2018 offset by a 0.97 offsetting percentage reduction in 2020. This reduction is percent coding adjustment, necessary to offset the spending increase expected in 2020 • a 2.2 percent payment update for 2019, resulting from the behavioral changes; it does not reflect any assessment of the adequacy of Medicare’s payments. assumed nominal case-mix growth of 0.5 percent in • Further reductions are necessary to better align payments 2018 and 2019, with the costs of services. • rural add-on for 2018 and 2019, and RECOMMENDATION 9 • assumed episode cost growth of 1 percent per year. For 2020, the Congress should reduce the calendar year 2019 Medicare base payment rate for home health On the basis of these policies and assumptions, the agencies by 5 percent. Commission projects a margin of 16.0 percent in 2019. RATIONALE 9 An immediate reduction of 5 percent in 2020 would How should Medicare payments change represent a significant action to address the magnitude of in 2020? the overpayments embedded in Medicare’s rates. However, this reduction will likely be inadequate to align Medicare Our review of payment adequacy for Medicare home payments with providers’ actual costs, and further health service indicates that access is more than adequate reductions will likely be necessary. In past years, the in most areas and that Medicare payments are substantially Commission has recommended that payments be rebased in excess of costs. On the basis of these findings, the in the year following a payment rate reduction. However, Commission has concluded that home health payments given the congressionally mandated revisions to the home should be significantly reduced. Though PPACA included health PPS that are slated for 2020, our recommendation a provision intended to lower payments, the impact of this for 2020 addresses only the level of payment. The planned provision has been modest, and substantial margins for revisions to the home health PPS will likely change many agencies are likely to remain. the mix of services and number of visits provided in an episode, and the payment rate set under a rebasing policy Home health care can be a high-value benefit when should reflect the mix and level of services HHAs provide it is appropriately and efficiently delivered. Medicare under the new payment policies. These data will not be beneficiaries often prefer to receive care at home instead available until mid-2021. March 2019 | Report to the Congress: Medicare Payment Policy 245

276 IMPLICATIONS 9 Beneficiary and provider • Beneficiaries’ access to care should not be affected. Spending Lowering payments should not affect providers’ The payment reductions would lower payments • willingness to deliver appropriate home health care. ■ relative to current law by $750 million to $2 billion in 2020 and by $5 billion to $10 billion over five years. Home health care services: Assessing payment adequacy and updating payments 246

277 Endnotes 6 If we approximate marginal cost as total Medicare costs 1 Freestanding providers accounted for about 90 percent of the minus fixed building and equipment costs, then marginal episodes provided in 2017. profit can be calculated as follows: 2 Payment updates are typically intended to address annual Marginal profit = (Medicare payments – (total Medicare costs increases in provider costs (e.g., salary increases or higher – fixed costs)) divided by Medicare payments prices for other inputs). However, during this period the cost of a home health episode did not increase substantially. In This comparison is a lower bound on the marginal profit recent years, annual cost growth has averaged less than 1 because we do not consider any potential labor costs that are percent, with some years experiencing no growth or decreases fixed. in cost. 7 Medicare collects home health quality data for both fee-for- The average payment in 2017 was $3,030. 3 service and Medicare Advantage beneficiaries. However, the program’s publicly reported measures present aggregate 4 As of November 2018, our measure of access is based on results that do not distinguish between the two programs. data collected and maintained as part of CMS’s Home Health Compare database. The service areas listed are postal ZIP 8 The sample for this analysis is derived from the larger sample codes where an HHA has provided services in the past 12 of freestanding HHA cost reports used to calculate margins months. This definition may overestimate access because in Table 9-8 (p. 243). Of these agencies, 5,147 of them had HHAs need not serve the entire ZIP code to be counted as three years of cost report data necessary for the analysis (2014 serving it. At the same time, the definition may understate through 2016), while 543 agencies did not have quality data access if HHAs are willing to serve a ZIP code but did not necessary to identify an efficient provider. receive a request in the previous 12 months. The analysis excludes beneficiaries with unknown ZIP codes. Medicare makes a case-mix-adjusted 60-day episode payment 5 when more than 4 visits are provided. Low-utilization payment adjustment episodes with four or fewer visits are paid on a per visit basis. March 2019 | Report to the Congress: Medicare Payment Policy 247

278 References McWilliams, J. M., L. G. Gilstrap, D. G. Stevenson, et al. 2017. Benjamin, A. E. 1993. An historical perspective on home care Changes in postacute care in the Medicare Shared Savings 71, no. 1: 129–166. Milbank Quarterly policy. 177, no. 4 (April 1): 518–526. Program. JAMA Internal Medicine Evaluation of Cheh, V., N. Duda, B. L. Carlson, et al. 2007. Report to the Medicare Payment Advisory Commission. 2011. the Home Health Independence Demonstration: Barriers to Washington, DC: MedPAC. Congress: Medicare payment policy. a successful experiment were multifaceted and difficult policy . Prepared for the Centers for Medicare & Medicaid issues remain Medicare Payment Advisory Commission. 2010. Report to the Services. Princeton, NJ: Mathematica Policy Research Inc. Washington, DC: MedPAC. Congress: Medicare payment policy. October. Medicare Payment Advisory Commission. 2004. Report to the 2017 Medicare Department of Health and Human Services. 2018. Congress: Medicare payment policy. Washington, DC: MedPAC. . Washington, fee-for-service supplemental improper payment data DC: HHS. Office of Inspector General, Department of Health and Human The physician’s role in Medicare’s home health Services. 2001. Feder, J., and J. Lambrew. 1996. Why Medicare matters to people OEI–02–00–00620. Washington, DC: OIG. services. 18, no. Health Care Financing Review who need long-term care. 2 (Winter): 99–112. Office of Inspector General, Department of Health and Human Review of Medicare home health services in Services. 1996. Government Accountability Office. 2018. Medicare: CMS should . A–04–96–2121. California, Illinois, New York, and Texas take actions to continue prior authorization efforts to reduce Washington, DC: OIG. spending. HEHS–18–341. Washington, DC: GAO. Staff report on home U.S. Senate Committee on Finance. 2011. Medicare: Home health Government Accountability Office. 1996. Washington, DC: health and the Medicare therapy threshold. utilization expands while program controls deteriorate. GAO/ Government Printing Office. HEHS–96–16. Washington, DC: GAO. Wolff, J., A. Meadow, C. Weiss, et al. 2008. Medicare home McCall, N., J. Korb, A. Petersons, et al. 2004. Decreased home health and the role of physicians. Presentation at Academy health use: Does it decrease satisfaction? Medical Care Research Health, Medicare and Post-Acute Care. June 8. 61, no. 1 (March): 64–88. and Review McCall, N., A. Petersons, S. Moore, et al. 2003. Utilization of home health services before and after the Balanced Budget Act of 38, Health Services Research 1997: What were the initial effects? no. 1 part 1 (February): 85–106. Home health care services: Assessing payment adequacy and updating payments 248

279 CHAPTER 10 Inpatient rehabilitation facility services

280 RECOMMENDATION 10 For 2020, the Congress should reduce the fiscal year 2019 Medicare base payment rate for inpatient rehabilitation facilities by 5 percent. COMMISSIONER VOTES: YES 17 • NO 0 • NOT VOTING 0 • ABSENT 0 ... (Additionally, the Commission reiterates its March 2016 recommendations on the inpatient rehabilitation facility prospective payment system. See text box, p. 261.)

281 CHAPTER 10 Inpatient rehabilitation facility services In this chapter Chapter summary Inpatient rehabilitation facilities (IRFs) provide intensive rehabilitation • Are Medicare payments services to patients after illness, injury, or surgery. Rehabilitation programs adequate in 2019? are supervised by rehabilitation physicians and include services such as How should Medicare • physical and occupational therapy, rehabilitation nursing, speech–language payments change in 2020? pathology, and prosthetic and orthotic services. In 2017, Medicare spent $7.9 billion on IRF care provided to fee-for-service (FFS) beneficiaries in about 1,180 IRFs nationwide. About 340,000 beneficiaries had around 380,000 IRF stays. On average, the Medicare FFS program accounted for 58 percent of IRF discharges. Assessment of payment adequacy Our indicators of Medicare payment adequacy for IRFs are positive. Beneficiaries’ access to care —Our analysis of IRF supply and volume of services provided and of IRFs’ marginal profit under Medicare’s IRF prospective payment system suggest that access remains adequate. —After declining for several years, • Capacity and supply of providers the number of IRFs increased in 2014 and continued to grow through 2016, reaching 1,188 facilities nationwide. In 2017, however, the number of IRFs declined slightly, to 1,178 facilities. Over time, the number of hospital-based and nonprofit IRFs has declined, while the number March 2019 | Report to the Congress: Medicare Payment Policy 251

282 of freestanding and for-profit IRFs has increased. In 2017, the average IRF occupancy rate remained at 65 percent, indicating that capacity is more than adequate to meet demand for IRF services. • Volume of services —From 2016 to 2017, the number of Medicare FFS cases declined 2.7 percent, falling to about 380,000 cases after having experienced small annual growth every year since 2010. • Marginal profit— The marginal profit, an indicator of whether IRFs with excess capacity have an incentive to treat more Medicare beneficiaries, was 19.4 percent for hospital-based IRFs and 38.8 percent for freestanding IRFs—a very positive indicator of patient access. Quality of care— The Commission tracks three broad categories of IRF quality indicators: risk-adjusted facility-level change in patients’ functional and cognitive status during the IRF stay, rates of discharge to the community and to skilled nursing facilities, and rates of readmission to an acute care hospital. Most measures were steady or improved between 2012 and 2017. Providers’ access to capital— The parent institutions of hospital-based IRFs continue to have good access to capital. The major freestanding IRF chain, which accounted for almost half of freestanding IRFs in 2017 and about a quarter of all Medicare IRF discharges, also has good access to capital. This assessment is reflected in the chain’s continued expansion. We were not able to determine the ability of other freestanding facilities to raise capital. IRFs’ access to capital in large part depends on their total (all-payer) profitability, and in 2017, total margins for freestanding IRFs were 10.4 percent. Data on all-payer profitability are not available for hospital-based units, but we can examine the all-payer margins of hospitals with IRF units, which, in 2017, had an aggregate all-payer margin across all lines of business of 7.0 percent. Medicare payments and providers’ costs— The aggregate Medicare margin for IRFs has grown steadily since 2009. In the three-year period between 2015 and 2017, the aggregate IRF Medicare margin remained above 13 percent and in 2017 stood at 13.8 percent. Also in 2017, Medicare margins in freestanding IRFs were 25.5 percent, down slightly from their peak in 2015 of 26.7 percent. In 2017, hospital-based IRF margins were comparatively low at 1.5 percent, but one-quarter of hospital-based IRFs had Medicare margins greater than 11 percent, indicating that many hospitals can manage their IRF units profitably. Lower margins in hospital-based IRFs were driven largely by higher unit costs. In addition, there are notable differences in hospital-based and freestanding IRFs’ mix of cases, which may indicate differences in profitability across case types. Finally, while Inpatient rehabilitation facility services: Assessing payment adequacy and updating payments 252

283 not definitive, evidence indicates that IRFs’ assessments of patients’ motor and cognitive function are not reliably consistent across providers. To the extent that hospital-based IRFs routinely assess their patients as less disabled than do their freestanding counterparts, their payments—and margins—will be systematically lower. Growth in IRFs’ costs historically has been low. From 2009 to 2015, the cumulative growth in cost per discharge was 8.4 percent, well below the 13.5 percent increase in the market basket for IRFs over the period. In 2016, per case cost growth (3.6 percent in aggregate) exceeded payment growth (2.9 percent in aggregate) for the first time since 2008. In 2017, however, per case payments again grew faster than costs (3.4 percent compared with 2.8 percent), resulting in an aggregate IRF margin of 13.8 percent. In 2018 to 2019, we anticipate costs in IRFs will grow faster than payments since updates in those years were constrained to 1.0 percent and 1.35 percent, respectively. For 2019, we project an aggregate Medicare margin of 11.6 percent. This year, the Commission for the first time examined the financial performance of relatively efficient IRFs. Our analysis found that relatively efficient IRFs performed better on quality metrics and had costs 18 percent lower than other IRFs. Relatively efficient IRFs were on average larger and had higher occupancy rates, contributing to greater economies of scale and lower costs. Freestanding and for-profit facilities were more likely to be in the relatively efficient group. On the basis of these factors, the Commission recommends a 5 percent reduction to the IRF payment rate for fiscal year 2020. In addition, the Commission reiterates its March 2016 recommendations that (1) the high-cost outlier pool be expanded to further redistribute payments in the IRF payment system and reduce the impact of misalignments between IRF payments and costs and (2) the Secretary conduct focused medical record review of IRFs that have unusual patterns of case mix and coding and conduct other research necessary to improve the accuracy of payments and protect program integrity. ■ March 2019 | Report to the Congress: Medicare Payment Policy 253

284

285 Medicare facility requirements for IRFs Background To qualify as an IRF for Medicare payment, facilities must meet the Medicare conditions of participation for acute After illness, injury, or surgery, some patients need care hospitals. They must also: intensive, inpatient rehabilitative care, including physical, occupational, and speech therapy. Such services can be have a preadmission screening process to determine • 1 provided in inpatient rehabilitation facilities (IRFs). that each prospective patient is likely to benefit IRFs must be primarily focused on treating conditions significantly from an intensive inpatient rehabilitation that typically require intensive rehabilitation, among program; other requirements. IRFs can be freestanding facilities or specialized units within acute care hospitals. To qualify • ensure that the patient receives close medical for a covered IRF stay, a beneficiary must be able to supervision and provide—through qualified tolerate and benefit from intensive therapy and must personnel—rehabilitation nursing, physical therapy, have a condition that requires frequent and face-to-face occupational therapy, and, as needed, speech– supervision by a rehabilitation physician. Other patient language pathology and psychological (including admission criteria also apply. In 2017, Medicare spent neuropsychological) services, social services, and $7.9 billion on IRF care provided in about 1,180 IRFs orthotic and prosthetic services; nationwide. About 340,000 beneficiaries had almost have a medical director of rehabilitation with training • 380,000 IRF stays. On average, Medicare fee-for-service or experience in rehabilitation who provides services (FFS) beneficiaries accounted for about 58 percent of in the facility on a full-time basis for freestanding IRF discharges. IRFs or at least 20 hours per week for hospital-based Since January 2002, Medicare has paid IRFs under a per IRF units; 2 discharge prospective payment system (PPS). Under • use a coordinated interdisciplinary team led by a the IRF PPS, Medicare patients are assigned to case-mix rehabilitation physician that includes a rehabilitation groups (CMGs) based on the patient’s primary reason nurse, a social worker or case manager, and a licensed for inpatient rehabilitation, age, and level of motor and therapist from each therapy discipline involved in the cognitive function. Within each of these CMGs, patients patient’s treatment; are further categorized into one of four tiers based on the presence of certain comorbidities that have been found to have a plan of treatment for each patient that is • increase the cost of care. Each CMG tier has a designated established, reviewed, and revised as needed by a weight that reflects the group’s average relative costliness physician in consultation with other professional of cases compared with that of the average Medicare personnel who provide services to the patient; and 3 IRF case. The CMG weight is multiplied by a base payment rate and then adjusted to reflect geographic meet the compliance threshold, which requires that • differences in the wages IRFs pay. The payment is no less than 60 percent of patients admitted to an IRF further adjusted based on the IRF’s share of low-income have as a primary diagnosis or comorbidity at least 1 4 patients. Additional adjustments are made for IRFs that of 13 conditions specified by CMS. The intent of the are teaching facilities and for IRFs located in rural areas. compliance threshold is to distinguish IRFs from acute The IRF PPS also has outlier payments for patients who care hospitals. If an IRF does not meet the compliance are extraordinarily costly. Starting in fiscal year 2020, threshold, Medicare pays for all its cases on the basis CMS is changing the patient assessment instrument of the inpatient hospital PPS rather than the IRF PPS. used to help classify patients for payment, shifting from Medicare coverage criteria for beneficiaries IRF-specific measures of motor and cognitive function to measures that are standardized across post-acute Medicare applies additional criteria that govern whether care (PAC) settings. The changes to the assessment IRF services are covered for an individual Medicare 5 instruments will necessitate minor adjustments of the beneficiary. For an IRF claim to be considered reasonable CMG definitions (see text box, pp. 256–257). and necessary, the patient must be reasonably expected to meet the following requirements at admission: March 2019 | Report to the Congress: Medicare Payment Policy 255

286 Changes to the IRF assessment instrument and case-mix groups in fiscal year 2020 between and overlap of the FIM and QRP items mean nder the inpatient rehabilitation facility that CMS can replace FIM elements with QRP items (IRF) prospective payment system (PPS), for without materially changing the case-mix classification purposes of payment, patients are assigned to U system. All other aspects of the classification system rehabilitation impairment categories (RICs) based on will be unchanged, including the RIC structure, the the principal diagnosis or primary reason for inpatient assignment of comorbidity tiers, and the methodology rehabilitation. Within each RIC, patients are sorted into for calculating the payment weights. The CMG case-mix groups (CMGs) based on the patient’s level classification system will continue to have 21 RICs of motor and cognitive function at admission and then (plus 2 for patients who have very short stays or who further categorized into one of four tiers based on the die in the IRF). However, the revisions will result in presence of specific comorbidities that have been found some consolidation of CMGs so that, instead of 92 to increase the cost of care. CMGs, there will be 88. At the RIC level, the changes To determine the appropriate CMG, IRFs assess and to the payment weights will be relatively small. score each patient’s motor and cognitive function using CMS plans to implement these revisions in a budget- the IRF–Patient Assessment Instrument (IRF–PAI). neutral manner. CMS’s initial analysis indicates that The IRF–PAI is based on a modified version of the the change will redistribute payments across providers, Uniform Data System for Medical Rehabilitation resulting in increased aggregate payments for hospital- patient assessment instrument, commonly referred to TM TM based and nonprofit IRFs as well as for smaller as the Functional Independence Measure . , or FIM IRFs. This projected shift in payments suggests that The IRF–PAI’s 18 FIM data elements and associated assessments of patients’ motor and cognitive function modifiers, along with the FIM measurement scale, are are not completely consistent across the two sets of data used to measure a patient’s level of disability and the elements; that is, a patient’s FIM function scores are not burden of care for a patient’s caregivers. (All else equal, entirely predictive of the patient’s QRP function scores. a greater level of disability generally results in a higher payment.) One potential reason for these differences is that the FIM score is intended to reflect the patient’s “lowest” The IRF–PAI also includes items that are standardized level of function during the time of assessment, across post-acute care (PAC) settings and are used to whereas the QRP score is intended to measure the collect information on a patient’s motor and cognitive patient’s “usual” functional level during the period function for the IRF Quality Reporting Program (QRP). of assessment. In addition, functional status data are As shown in Table 10-1, the QRP items are very similar generally obtained by observation of the patient and to the FIM elements and associated modifiers. Because are somewhat subjective. Moreover, the FIM scores the QRP elements overlap the FIM data elements, are used to determine payment to IRFs, while the QRP CMS believes that the collection of FIM elements and scores have had no effect on payment to date. Because associated modifiers is no longer necessary and places payment is materially affected by patients’ FIM scores undue burden on providers. Accordingly, in fiscal at admission—with higher payments associated with year 2020, CMS will remove the FIM elements and lower functional status—providers have a financial associated modifiers from the IRF–PAI and will rely on incentive when scoring the FIM elements to minimize QRP items to assign cases to CMGs. patients’ assessed levels of function at admission. No Because the QRP items are defined differently such incentive has existed for QRP scoring. However, from the FIM elements and use a different scale of that situation will change when CMS begins to use measurement, using QRP items for CMG assignment QRP scores to determine payment. will require some revisions to the CMG classification system. However, CMS anticipates the similarity (continued next page) Inpatient rehabilitation facility services: Assessing payment adequacy and updating payments 256

287 Changes to the IRF assessment instrument and case-mix groups in fiscal year 2020 (cont.) PAC-standardized data “as soon as practicable.” At In a comment letter to the Secretary, the Commission the same time, moving toward an IRF classification supported replacing FIM items and modifiers with system that adjusts payments using data elements that QRP items because doing so would relieve providers are standardized across all PAC settings is a necessary of having to report this information on functional status step toward a unified PAC PPS. The Commission noted, twice, using different definitions and measurement however, that once the QRP scores are used to determine scales (Medicare Payment Advisory Commission 2018). payment, providers likely will respond quickly, Further, Section 1899(b)(3) of the Improving Medicare devoting resources to improving the coding of the Post-Acute Care Transformation Act of 2014 requires the Secretary to replace existing setting-specific patient QRP functional measures, altering their QRP scoring assessment data that duplicate or overlap the required practices, or both. ■ TABLE TM Selected FIM elements and QRP counterparts on the IRF–PAI 10–1 QRP FIM FIM item A— The use of suitable utensils Self-care: Eating The ability to use suitable GG130–A— to bring food to the mouth, chewing and utensils to bring food to the mouth and swallowing, once the meal is presented in swallow food once the meal is placed the customary manner on a table or tray. before the patient. Self-care: Bathing FIM item C— Washing, rinsing, and drying The ability to bathe self in GG130–E— the body from the neck down (excluding the shower or tub, including washing, rinsing, back) in either a tub, shower, or sponge/ and drying self. bed bath. Self-care: Dressing upper body FIM item D— Dressing and undressing above GG130–F— The ability to put on and remove shirt or pajama top; includes buttoning, if the waist, as well as applying and removing applicable. a prosthesis or orthosis when applicable. GG130–C— Self-care: Toileting FIM item F— Maintaining perineal hygiene The ability to maintain perineal and adjusting clothing before and after using hygiene, adjust clothes before and after using a toilet, commode, bedpan, or urinal. the toilet, commode, bedpan, or urinal. FIM item I— All aspects of transferring from Transfers: Bed, chair, wheelchair The ability to come to a GG170–D— bed to a chair, or wheelchair, or coming to standing position from sitting in a chair, or on a standing position, if walking is the typical the side of the bed. mode of locomotion. GG170–E— The ability to safely transfer to and from a bed to a chair (or wheelchair). Transfers: Toilet Includes safely getting on and off GG170–F— FIM item J— The ability to safely get on and off a toilet or commode. a standard toilet. Locomotion: Walk FIM item L— Ability to/level of assistance GG170–K— Once standing, the ability to needed to walk 150 feet. walk at least 150 feet in a corridor or similar space. TM TM ), QRP (Quality Reporting Program), IRF–PAI (Inpatient Rehabilitation Facility–Patient Assessment Instrument). Note: FIM (Functional Independence Measure CMS, Inpatient Rehabilitation Facility–Patient Assessment Instrument, Version 1.5. Source: March 2019 | Report to the Congress: Medicare Payment Policy 257

288 TABLE 10–2 The number and share of FFS IRF cases with neurological conditions and brain injury continued to grow, 2004–2017 Share of IRF Medicare Percentage point change FFS cases Meets 2008– 2004– 2016– compliance a 2017 2008 2004 Condition 2008 2016 2017 2016 threshold –0.2 16.6 % 20.4 % 20.2 % 20.5 yes 3.8 Stroke 0.2 % Other neurological conditions 5.2 8.0 13.6 15.0 yes 2.9 5.6 1.3 –0.4 –5.2 3.0 yes 10.9 16.0 13.1 Fracture of the lower extremity 10.4 no 0.0 1.5 2.9 10.6 10.6 9.1 6.2 Debility 2.9 3.0 yes 10.7 9.9 7.0 3.9 Brain injury 0.8 no 2.1 0.9 7.9 8.2 6.1 5.2 Other orthopedic conditions –0.2 Cardiac conditions 5.3 4.6 6.0 –0.3 no –0.6 1.4 5.5 b –1.1 –11.0 –7.7 Major joint replacement of lower extremity 24.1 13.1 5.4 4.4 0.0 4.2 4.3 4.9 4.9 yes 0.1 0.6 Spinal cord injury c –5.0 –1.1 –0.3 9.8 All other 11.3 16.3 10.1 Note: FFS (fee-for-service), IRF (inpatient rehabilitation facility). “Other neurological conditions” includes multiple sclerosis, Parkinson’s disease, polyneuropathy, and neuromuscular disorders. “Fracture of the lower extremity” includes hip, pelvis, and femur fractures. Patients with debility have generalized deconditioning not attributable to other conditions. “Other orthopedic conditions” excludes fractures of the hip, pelvis, and femur, and hip and knee replacements. “All other” includes conditions such as amputations, arthritis, and pain syndrome. All Medicare FFS IRF cases with valid patient assessment information were included in this analysis. Yearly figures presented in the table are rounded, but figures in the percentage point change columns were calculated using unrounded data. a The compliance threshold requires that at least 60 percent of an IRF’s patients have 1 of 13 specified diagnoses or have a comorbidity that could cause significant decline in functional ability such that the patient requires intensive rehabilitation. Some FFS cases with conditions that do not meet the compliance threshold could thus be counted toward the threshold if they had certain comorbidities. b Cases admitted for rehabilitation after major joint replacement of the lower extremity count toward the compliance threshold if joint replacement was bilateral, if the patient had a body mass index of 50 or greater, or if the patient was age 85 or older. c Conditions in the “all other” category that meet the compliance threshold include congenital deformity, lower-limb amputations, major multiple trauma, burns, and certain arthritis cases. Source: MedPAC analysis of Inpatient Rehabilitation Facility–Patient Assessment Instrument data from CMS. • The patient requires active and ongoing therapy in at Patterns of use in IRFs least two modalities, one of which must be physical or In 2004, CMS began to consistently enforce the IRF occupational therapy. compliance threshold and enacted revisions to some of 6 the qualifying conditions. The combination of renewed The patient can actively participate in and benefit from • enforcement of the threshold and additional restrictions intensive therapy that most typically consists of three resulted—as intended—in a substantial decline in the hours of therapy a day at least five days a week. volume of Medicare patients treated in IRFs. By 2008, The patient is sufficiently stable at the time of • the number of IRF discharges had fallen 26 percent, with admission to actively participate in the intensive the biggest declines seen in the number of medically rehabilitation program. complex (–73 percent), arthritis (–68 percent), and hip and knee replacement (–60 percent) cases. Average The patient requires supervision by a rehabilitation • case-mix severity and cost per case increased as IRFs physician. This requirement is satisfied by face-to- shifted their mix of cases to conditions that count face physician visits with a patient at least three days toward the threshold, such as stroke, brain injury, and a week. other neurological conditions (Table 10-2). IRF volume stabilized after 2008, but increases in certain neurological Inpatient rehabilitation facility services: Assessing payment adequacy and updating payments 258

289 TABLE 10–3 Mix of Medicare FFS IRF cases differed by provider type, selected conditions, 2017 Freestanding Hospital based Condition For profit Nonprofit For profit Nonprofit 16 Stroke % 26 % 20 % 26% 21 8 13 10 Other neurological conditions Fracture of the lower extremity 9 8 13 11 Debility 11 8 12 10 12 10 11 Brain injury 11 6 6 Other orthopedic conditions 10 7 Note: FFS (fee-for-service), IRF (inpatient rehabilitation facility). “Other neurological conditions” includes multiple sclerosis, Parkinson’s disease, polyneuropathy, and neuromuscular disorders. “Fracture of the lower extremity” includes hip, pelvis, and femur fractures. Patients with debility have generalized deconditioning not attributable to other conditions. “Other orthopedic conditions” excludes fractures of the hip, pelvis, and femur, and hip and knee replacements. All Medicare FFS IRF cases with valid patient assessment information were included in this analysis. Source: MedPAC analysis of Inpatient Rehabilitation Facility–Patient Assessment Instrument data from CMS. share of cases with brain injury or debility was similar conditions—Parkinson’s disease and neuromuscular across IRF types. disorders—continued. Between 2008 and 2017, the number of IRF discharges with other neurological In 2017, 8.5 percent of IRF cases received high-cost conditions almost doubled, climbing 99 percent, and the outlier payments, although the share varied by case number of discharges with brain injuries (traumatic and type. For example, high-cost outlier cases accounted for nontraumatic combined) rose 63 percent, while the total 12.6 percent of spinal cord injury cases, 10.7 percent of number of Medicare IRF discharges increased 6 percent stroke cases, 6.3 percent of cases with other neurological (data not shown). Notably, the number of cases with other conditions, and 5.2 percent of other orthopedic conditions. orthopedic conditions, cardiac conditions, and debility also Outlier cases were also distributed unevenly among IRFs. rose, though a sizable share of these cases do not count High-cost outliers accounted for almost 15 percent of 7 toward the compliance threshold. The number of hip and hospital-based IRF cases compared with 2.6 percent of knee replacement cases going to IRFs continued their freestanding IRF cases. On average, high-cost outliers had downward trajectory, declining an additional 55 percent an average length of stay that was 7.3 days longer than from 2008 to 2016. IRFs also saw a large decline in cases non-outlier cases (19.4 days vs. 12.1 days). Outlier cases for fractures of the lower extremity, falling 26 percent were also more likely to have comorbidities that increased over the same period, even though they count toward the case mix (65.6 percent of outlier cases vs. 55.1 percent for compliance threshold. non-outlier cases). The distribution of case types differs by type of IRF High-margin IRFs have a different mix of (Table 10-3). For example, in 2017, only 16 percent cases of cases in freestanding for-profit IRFs were admitted for rehabilitation following a stroke, compared with A previous Commission analysis of differences in the 26 percent of cases in hospital-based nonprofit IRFs. mix of cases across IRFs suggested that patient selection Likewise, 21 percent of cases in freestanding for-profit contributes to provider profitability (Medicare Payment IRFs were admitted with other neurological conditions, Advisory Commission 2016). We found that IRFs with more than twice the share admitted to hospital-based the highest margins in 2013 had a higher share of other 8 nonprofit IRFs. Cases with other orthopedic conditions neurological cases and a lower share of stroke cases. also made up a higher share of cases in freestanding for- Further, we observed differences in the types of stroke profit facilities than in all other IRFs. By contrast, the March 2019 | Report to the Congress: Medicare Payment Policy 259

290 • Patients in high-margin IRFs were less likely to have and other neurological conditions admitted to high-margin been high-cost outliers in the acute care hospital or to and low-margin IRFs. Stroke cases in the highest margin IRFs were two-and-a-half times more likely than those have spent four or more days in the hospital intensive care or coronary care unit. in the lowest margin IRFs to have no paralysis. Likewise, other neurological cases in the highest margin IRFs were But once patients were admitted to and assessed by the almost three times more likely than those in the lowest IRF, the average patient profile changed, with patients margin IRFs to have a neuromuscular disorder (such as treated in high-margin IRFs appearing to be more amyotrophic lateral sclerosis or muscular dystrophy) as disabled than those in low-margin IRFs (as measured by opposed to neurological conditions like multiple sclerosis motor impairment scores assigned by IRFs). This pattern or Parkinson’s disease. persisted across case types. As noted in our March 2016 report to the Congress, these As noted in our March 2016 report to the Congress, the findings suggest that, under the IRF PPS, some case types consistent finding that high-margin IRFs have patients are more profitable than others. The Commission plans who are, on average, less severely ill in the acute care to assess variation in costs among the IRF CMGs and hospital but appear more functionally disabled upon differences in relative profitability across CMGs in future assessment in the IRF suggests that assessment and analyses. It is necessary to identify and reduce variation in scoring practices contribute to greater profitability in some costs among CMGs and properly calibrate payments with IRFs, especially given the comparatively low level of costs for each group to avoid overpayments and reduce costs and cost growth observed in high-margin facilities. financial incentives for providers to admit certain types of If providers differ in their assessment and scoring of cases and avoid others. In the short term, the Commission patients’ motor and cognitive function, payments will not has recommended that the Secretary effect changes to be properly aligned with the resource needs of patients. reduce potential misalignments between IRF payments Some IRFs will receive payments that are too high relative and costs by redistributing payments within the IRF PPS to the costs incurred in treating their patients, while other through the high-cost outlier pool (see text box on March IRFs will receive payments that are too low. 2016 recommendations). Expanding the outlier pool would increase outlier payments for the most costly cases, These findings led the Commission to recommend that easing the financial burden for IRFs that have a relatively CMS ensure payment accuracy and help improve program high share of these cases. integrity by reviewing medical records and conducting other research as necessary (see text box on March 2016 Data suggest patients not assessed recommendations). More recently, the Commission has uniformly across IRFs begun to explore data integrity issues related to post-acute A previous Commission analysis of acute care hospital care (PAC) patient assessment data more broadly, and we claims data and data from the Inpatient Rehabilitation expect to evaluate whether such data can continue to be Facility–Patient Assessment Instrument (IRF–PAI), used in Medicare’s payment systems or quality incentive while not definitive, strongly suggests that IRFs differ programs. in their assessment of patients’ motor and cognitive function, raising more generalized concerns about patient assessment data (Medicare Payment Advisory Are Medicare payments adequate in Commission 2016). 2019? Overall, when we compared patients in high-margin and low-margin IRFs, we found that patients in high-margin To assess whether payments for fiscal year 2019 are IRFs were less severely ill and resource intensive during adequate to cover the costs providers incur and how much the acute care hospitalization that preceded the IRF stay: providers’ costs are expected to change in the coming year (2020), we examine several indicators of payment adequacy. • Patients in high-margin IRFs had, on average, a lower Specifically, we assess beneficiaries’ access to care by case-mix index in the acute care hospital as well as a examining the capacity and supply of IRFs and changes over lower level of severity of illness and a shorter length time in the volume of services provided, quality of care, of stay. providers’ access to capital, and the relationship between Medicare payments and providers’ costs. Inpatient rehabilitation facility services: Assessing payment adequacy and updating payments 260

291 The Commission reiterates its March 2016 recommendations on the IRF prospective payment system Beneficiary and provider Recommendation 9-2 • We do not expect this recommendation to have The Secretary should conduct focused medical adverse effects on Medicare beneficiaries with record review of inpatient rehabilitation facilities respect to access to care or out-of-pocket spending that have unusual patterns of case mix and coding. or on providers’ willingness and ability to care for Medicare beneficiaries. Rationale 9-2 The Commission’s finding that high-margin inpatient Recommendation 9-3 rehabilitation facilities (IRFs) have patients who The Secretary should expand the inpatient are, on average, less severely ill in the acute care rehabilitation facility outlier pool to redistribute hospital but appear more functionally disabled in the payments more equitably across cases and IRF suggests the possibility that coding practices providers. contribute to greater profitability in some IRFs. Providers may differ in their assessment of patients’ Rationale 9-3 motor and cognitive function, resulting in payments for some IRFs that are too high relative to the costs The Commission’s finding that high-margin IRFs may incurred in treating their patients. To improve the be selecting certain types of cases suggests that some accuracy of payments and protect program integrity, case-mix groups (CMGs) may be more profitable CMS should review medical records merged with IRF than others. At the same time, our finding that IRFs patient assessment data, reassess inter-rater reliability may differ in their assessments of patients’ motor and across IRFs, and conduct other research as necessary. cognitive function suggests that the IRF CMGs may not Because medical record review is resource intensive, be adequately capturing differences in patient acuity CMS should begin by focusing on providers that have and costs across cases and providers. The potential an atypical mix of cases, such as a high concentration for financial loss may therefore be greater for some of neuromuscular disorders and stroke cases without providers than for others. Expanding the outlier pool paralysis, and on providers that have anomalous would increase outlier payments for the most costly patterns of coding, such as wide discrepancies in cases, easing the financial burden for IRFs that have a their patients’ levels of severity as coded in the relatively high share of these cases. acute care hospital compared with that coded in the IRF. However, system-wide assessment of payment Implications 9-3 accuracy is also needed. Spending This recommendation would be implemented in • Implications 9-2 a budget-neutral manner and should not have an Spending overall impact on spending. Implementing this recommendation could result • in changes to the payment system that would be Beneficiary and provider budget neutral but could also reduce Medicare’s • We do not expect this recommendation to have spending on IRF services if CMS were to make adverse effects on Medicare beneficiaries with payment adjustments to account for assessment and respect to access to care or out-of-pocket spending. coding differences across providers or for coding This recommendation may relieve the financial changes that do not reflect real case-mix change. pressure on some providers and may improve CMS would incur some administrative expenses to equity among providers by diminishing the effects conduct these activities. of inaccurate coding. ■ March 2019 | Report to the Congress: Medicare Payment Policy 261

292 TABLE The number of for-profit and freestanding IRFs continued to grow in 2017 10–4 Average Share of annual change Number of IRFs Medicare FFS 2016– 2013– 2009– discharges 2017 2013 Type of IRF 2016 2017 2013 2014 2015 2016 2017 2009 1,196 100% All IRFs 1,178 –0.8% 0.8% –0.7% 1,188 1,182 1,177 1,161 1,020 Urban 992 977 1,013 93 1,026 1,019 –0.4 1.6 –0.7 162 –1.9 –4.2 –2.5 159 162 Rural 164 184 204 7 262 2.2 52 225 243 251 Freestanding 273 279 0.8 4.0 899 920 926 918 –1.4 –0.1 –1.7 Hospital based 915 48 971 –3.1 681 732 Nonprofit 677 39 0.0 –1.9 655 676 681 5.9 54 4.7 For profit 2.2 392 370 352 338 322 295 169 7 Government –6.0 –5.0 –2.1 125 133 138 149 155 Note: IRF (inpatient rehabilitation facility), FFS (fee-for-service). The number of facilities are for the calendar year. The large decline in the number of rural IRFs between 2013 and 2014 was due primarily to changes in the core-based statistical areas, as defined by the Office of Management and Budget, which determine whether geographic areas are considered urban or rural. Because of these changes, 19 IRFs that were previously considered rural are now designated urban. Components may not sum to totals due to missing data. Source: MedPAC analysis of Provider of Services data and Medicare Provider Analysis and Review data from CMS. 2016 to 1,188 facilities nationwide (Table 10-4). But Beneficiaries’ access to care: IRF supply and in 2017, the number of IRFs fell 0.8 percent to 1,178 service volume suggest sufficient access facilities. IRFs are not the sole provider of rehabilitation We have no direct indicator of beneficiaries’ access to IRF services in communities; SNFs also provide rehabilitation care. Although there are criteria for admission to an IRF, it services in an institutional setting, and home health is not clear when IRF care is necessary or beneficial for a agencies, comprehensive outpatient rehabilitation given patient or when another, potentially lower cost PAC facilities, and independent therapy providers furnish care provider (such as a skilled nursing facility (SNF)) could at home or on an outpatient basis. Given the number and provide appropriate care. The absence of IRFs in some distribution of these other rehabilitation therapy providers, areas of the country makes it particularly difficult to assess it is unlikely that areas exist where IRFs are the only the need for IRF care since beneficiaries in areas without provider of rehabilitation therapy services available to IRFs presumably receive similar services in other settings. Medicare beneficiaries. Nevertheless, our analysis of IRF supply and volume of services provided suggests that capacity remains In 2017, about 76 percent of IRFs were distinct units in adequate to meet demand. Moreover, the marginal profit, acute care hospitals; the rest were freestanding facilities. an indicator of whether IRFs with excess capacity have However, because hospital-based units have, on average, an incentive to treat more Medicare beneficiaries, was fewer beds and a lower share of Medicare discharges, they robust for both freestanding and hospital-based IRFs, thus accounted for only 48 percent of Medicare discharges. providing a very positive indicator of patient access. Overall, 33 percent of IRFs were for-profit entities. Freestanding IRFs were far more likely to be for profit Number of IRFs and occupancy rates suggest than were hospital-based IRFs (78 percent vs. 19 percent; adequate capacity and supply data not shown). In 2017, 54 percent of Medicare After declining from a peak of 1,235 facilities in 2005 discharges were from for-profit facilities. Over time, the (data not shown) to 1,161 facilities in 2013, the number number of hospital-based and nonprofit IRFs has declined, of IRFs increased in 2014 and continued to grow through Inpatient rehabilitation facility services: Assessing payment adequacy and updating payments 262

293 TABLE The number of IRF cases per FFS beneficiary decreased in 2017 10–5 Average annual change 2016– 2004– 2008– 2017 2016 2008 2017 2004 2016 2014 2012 2010 2008 2006 Number of 1.2 % % –2.7 % 379,885 390,514 375,590 373,284 359,307 356,312 404,633 495,349 cases –7.9 Cases per 10,000 FFS 99.7 –2.4 0.1 –7.2 98.5 100.1 99.2 100.4 111.9 135.6 beneficiaries 100.9 Payment 5.8 2.1 3.1 per case $15,380 $16,646 $17,085 $17,795 $18,632 $19,714 $20,322 $13,290 ALOS 12.8 –0.6 12.9 0.0 12.7 1.3 (in days) 12.7 13.0 13.3 13.1 12.7 325,506 350,353 1.0 –2.9 Users 338,887 339,087 –7.9 323,897 369,269 449,362 340,175 Note: IRF (inpatient rehabilitation facility), FFS (fee-for-service), ALOS (average length of stay). Source: MedPAC analysis of Medicare Provider Analysis and Review data from CMS. IRF Medicare volume decreased in 2017 while the number of freestanding and for-profit IRFs has increased. Between 2009 and 2017, the number of The number of Medicare FFS IRF cases grew rapidly hospital-based IRFs fell by 7 percent and the number of throughout the 1990s and the early years of the IRF PPS, nonprofit IRFs fell by 10 percent, while the number of reaching a peak of about 495,000 in 2004. After CMS freestanding IRFs and for-profit IRFs rose by 19 percent renewed its enforcement of the compliance threshold in and 33 percent, respectively. 2004, IRF volume declined substantially, as expected, falling almost 8 percent per year from 2004 to 2008 (Table In 2017, 28 IRFs closed; most were hospital-based units. 10-5). At that point, volume began to increase slowly, At the same time, 19 new IRFs opened. Slightly more than rising an average of 1.2 percent per year from 2008 to half of the new IRFs were hospital-based units. Of the 2016. Between 2016 and 2017, however, the number of new hospital-based units, about a third were for profit; of FFS IRF cases fell 2.7 percent, to a little less than 380,000 the new freestanding facilities, half were for profit. Acute cases. care hospitals find that IRF units can help reduce inpatient lengths of stay. Previous Commission analyses have found In 2017, the number of IRF cases per 10,000 FFS that hospitals with IRF units have higher inpatient margins beneficiaries fell to 98.5, down 2.4 percent from the than hospitals without such units (Medicare Payment previous year. Relatively few Medicare beneficiaries use Advisory Commission 2015). IRF services because, to qualify for Medicare coverage, IRF patients must be able to tolerate and benefit from In 2017, the average IRF occupancy rate remained at 65 rehabilitation therapy that is intensive, which is usually percent, the same level as in 2016. Occupancy rates were interpreted to mean at least three hours of therapy a day higher in freestanding IRFs (69 percent) than in hospital- for at least five days a week. Yet, compared with all based IRFs (61 percent). These rates suggest that capacity Medicare beneficiaries, those admitted to IRFs in 2017 is more than adequate to meet demand for IRF services. were disproportionately over age 85. March 2019 | Report to the Congress: Medicare Payment Policy 263

294 With the decline in the number of IRF cases per FFS increase Medicare spending. There has been relatively beneficiary, FFS Medicare’s share of IRF discharges fell to little research on rehospitalization of IRF patients in 58 percent of total discharges as the volume of IRF cases aggregate, though some studies have focused on one or across all payers rose slightly in 2017 (data not shown). more rehabilitation impairment categories (Dejong et al. 2009, Galloway et al. 2013, Ottenbacher et al. 2014, Marginal profit provides incentive to treat more Schneider et al. 2013, Schneider et al. 2012). However, Medicare beneficiaries research regarding rehospitalization of SNF and nursing home patients has identified several contributing factors Another measure of access is whether providers have a that may be within a PAC provider’s control. These factors financial incentive to expand the number of Medicare include staffing level, skill mix, and frequency of staff beneficiaries they serve. In considering whether to treat turnover; drug management; and adherence to transitional a patient, a provider with excess capacity compares care protocols such as discharge counseling, medication the marginal revenue it will receive (i.e., the Medicare reconciliation, patient education regarding self-care, and payment) with its marginal costs—that is, the costs that communication among providers, staff, and the patient’s vary with volume. If Medicare payments are larger than family (Grabowski et al. 2008, Kane et al. 2003, Konetzka the marginal costs of treating an additional beneficiary, a et al. 2008a, Konetzka et al. 2008b, Lau et al. 2005, provider has a financial incentive to increase its volume of Mustard and Mayer 1997). Medicare patients. In contrast, if payments do not cover the marginal costs, the provider may have a disincentive The Commission’s rates of rehospitalization during the to care for Medicare beneficiaries. Given the difference IRF stay and during the 30 days after discharge are risk in financial performance across IRFs, we examined adjusted and reflect those readmissions that are potentially freestanding and hospital-based IRFs’ marginal profit to avoidable with adequate care in the IRF setting (Kramer assess whether both types of providers have a financial 10 et al. 2015). The measure of rehospitalization in the 30 incentive to increase the number of Medicare beneficiaries days after discharge reflects in part how well facilities 9 they serve. We found that Medicare payments exceed prepare beneficiaries and their caregivers for safe and marginal costs by a substantial amount—19.4 percent appropriate transitions to the home or the next health for hospital-based IRFs and 38.8 percent for freestanding care setting. Since 2013, the national average rate of risk- IRFs—suggesting that IRFs with available beds have a adjusted potentially avoidable rehospitalizations during strong incentive to admit Medicare patients. This finding the IRF stay has been about 2.6 percent (Table 10-6, p. is a very positive indicator of patient access, even in IRFs 266). (Lower rates are better.) Meanwhile, between 2012 with lower overall Medicare margins. and 2017, the rate of risk-adjusted potentially avoidable rehospitalization within 30 days after discharge from an Quality of care: Steady or improved for IRF declined from 4.8 percent to 4.3 percent in 2015, then most measures rose to 4.7 percent in 2016 and 2017. Between 2012 and 2017, the Commission has tracked three broad categories of IRF quality indicators: risk- We also examined rates of discharge to the community adjusted facility-level change in functional and cognitive and to SNFs. We found that between 2012 and 2017, status during the IRF stay, rates of discharge to the the national average for the risk-adjusted community community and to SNFs, and rates of readmission to an discharge rate increased from 74.2 percent to 76.0 11 acute care hospital (see text box on measures of quality). percent. (Higher rates are better.) Between 2012 and During this period, most measures were steady or 2014, the national average for the risk-adjusted rate of improved. discharge to SNFs increased from 6.9 percent to 7.1 percent, but subsequently declined to 6.8 percent in 2017 Risk-adjusted rates of potentially avoidable (lower rates are better). rehospitalization, discharge to the community, and discharge to SNF The Commission also considers functional status at admission and discharge, measured using the motor Avoidable rehospitalizations expose beneficiaries to and cognitive scores on the IRF–PAI. This instrument hospital-acquired infections, increase the number of incorporates the 18-item Functional Independence transitions between settings (which are disruptive to TM TM Measure ) scale to assess the level of disability (FIM patients), and can result in medical errors (such as in motor and cognitive functioning and the burden of medication errors). In addition, they unnecessarily Inpatient rehabilitation facility services: Assessing payment adequacy and updating payments 264

295 Measures of inpatient rehabilitation facility quality Patients who were discharged from the IRF to a nursing n its assessment of the quality of care in inpatient home for a non-SNF episode are not considered rehabilitation facilities (IRFs), the Commission discharged to a SNF. has historically examined risk-adjusted rates of I readmission to the hospital, discharge to the community The community discharge measure reflects the share of and to skilled nursing facilities (SNFs), and change in stays in which the patient was not discharged directly functional status during the IRF stay. from the IRF to a hospital or a SNF. Individuals who were discharged from the IRF to a nursing home as a Two readmission measures are calculated: one that non-SNF resident (that is, for long-term care financed occurs during the IRF stay and one that occurs within by payers other than Medicare) are included in the 30 days after discharge from the IRF (Kramer et al. measure of community discharge. Patients who were 2015). Individuals who died in the IRF or during discharged from the IRF to the community but were the 30 days after discharge from the IRF were admitted to a hospital within one day of discharge are excluded from the facilities’ readmission rates. The not considered discharged to the community. readmission measures count patients whose primary diagnosis for rehospitalization was considered The change in the Functional Independence Measure™ potentially avoidable; that is, the condition typically from admission to discharge is calculated for both could have been managed in the IRF. The potentially motor function and cognition. The measures represent avoidable readmissions are respiratory-related illness the average change among patients for 13 motor items (pneumonia, influenza, bronchitis, chronic obstructive and 5 cognitive items on the IRF–Patient Assessment pulmonary disease, and asthma); sepsis; congestive Instrument. Patients with missing information for any heart failure; fractures or fall with a major injury; of the items are not included when calculating average urinary tract or kidney infection; blood pressure change. management; electrolyte imbalance; anticoagulant therapy complications; diabetes-related complications; The observed rates of readmission to the hospital, cellulitis or wound infection; pressure ulcer; medication discharge to the community and to SNFs, and change error or adverse drug reaction; and delirium. For in functional status during the IRF stay are risk the measure of potentially avoidable readmission adjusted for medical comorbidities, functional status at during the IRF stay, delirium could be a primary or a IRF admission, rehabilitation impairment category, and secondary rehospitalization diagnosis. demographic characteristics. The data sources used for risk adjustment were Part A hospital and IRF claims. To account for beneficiaries who are discharged from Risk-adjusted rates compare a facility’s observed rates the IRF to a SNF, a measure of discharge to SNF is with its expected rates based on the mix of patients. calculated. This measure reflects the share of stays in The rates reported are the average risk-adjusted rates which the patient was discharged directly from the IRF for Medicare fee-for-service beneficiaries in all IRFs for additional rehabilitation in a SNF that was financed with 25 or more stays during the year. ■ under Medicare Part A’s skilled nursing benefit. indicating greater functional independence. To measure care for a patient’s caregivers (Deutsch et al. 2005). observed improvement in motor function and cognition, Scores for each of the 18 FIM items can be summed to we subtracted the respective FIM scores at admission calculate a motor score (based on 13 FIM items) and from the FIM scores at discharge to calculate FIM motor a cognitive score (based on 5 FIM items). The motor and cognitive gains (Kramer et al. 2015). A larger number score at discharge can range from 13 to 91, while the indicates more improvement in functional independence cognitive score can range from 5 to 35, with higher scores March 2019 | Report to the Congress: Medicare Payment Policy 265

296 TABLE Risk-adjusted quality indicators for IRFs held 10–6 steady or improved slightly from 2012 to 2017 Percent change 2012–2017 2012 Measure 2016 2015 2014 2017 2013 2.8% Potentially avoidable rehospitalizations during IRF stay –7.1% 2.6% 2.7% 2.6% 2.7% 2.6% 6.9% Discharged to a SNF 6.9% 7.1% 7.0% 6.8% 6.8% –1.4 75.0% 2.4 76.0% 75.9% Discharged to the community 75.2% 74.9% 74.2% Potentially avoidable rehospitalizations during 30 days –2.1 after discharge from IRF 4.8% 4.8% 4.7% 4.3% 4.7% 4.7% 22.1 22.4 22.9 23.1 23.7 24.0 8.6 Motor FIM™ gain 10.3 3.5 Cognitive FIM™ gain 3.7 3.7 3.8 3.9 3.7 Note: IRF (inpatient rehabilitation facility), SNF (skilled nursing facility), FIM™ (Functional Independence Measure™). High rates of discharge to the community indicate better quality. High rates of rehospitalization and discharge to SNF indicate worse quality. Rates are the average of facility rates and calculated for all facilities with 25 or more Medicare fee-for-service stays. The motor FIM measures the level of disability in motor functioning on a 91-point scale. The cognitive FIM measures the level of cognitive impairment on a 35-point scale. FIM gain is calculated as the FIM score at discharge minus the FIM score at admission. Higher FIM gain indicates more improvement. Mean FIM gain averages the change of all facilities with 25 or more Medicare fee-for-service stays. Source: MedPAC analysis of Inpatient Rehabilitation Facility–Patient Assessment Instrument data from CMS. years. However, improvements in the functional status and cognition between admission and discharge. Each measures should be viewed with some caution given that risk-adjusted rate was calculated by comparing a facility’s they are self-reported rather than claims-based measures. observed rate with its expected rate and multiplying this The Commission is evaluating the reliability of patient ratio by the national rate. assessment data and the appropriateness of using these In 2017, the mean gain (positive change) in the motor data for payment on quality assessment of PAC providers. FIM score during an IRF stay was 24.0, while the mean gain for the cognitive FIM score was 3.9 (Table 10-6). Variation in quality measures across IRFs (Bigger gains are better.) From 2012 to 2017, the average IRFs varied widely in their performance on Medicare’s risk-adjusted gain in IRF patients’ motor and cognitive quality measures (Table 10-7). In 2017, the lowest FIM scores (as assigned by IRFs) increased about 9 performing quartile of IRFs had a risk-adjusted rate percent and 10 percent, respectively. However, changes of discharge to a SNF that was 8.7 percent or higher, in motor function and cognition must be interpreted with compared with 4.2 percent or lower for the best caution. Functional status data are generally obtained by performing quartile of providers. (A lower rate of observation of the patient and are somewhat subjective. discharge to a SNF is better.) Risk-adjusted rates of Because payment is based in part on patients’ functional discharge to the community varied as well: The worst status at admission—with higher payments associated performing quartile of IRFs had a community discharge with lower functional status—providers have a financial rate of 73.1 percent or lower, compared with 79.2 percent incentive to minimize their assessments of patients’ levels or higher for the best performing quartile of providers. of function at admission. If IRFs minimize patients’ (A higher rate of discharge to the community is better.) functional status at admission, gains in function during the Rehospitalization rates also varied: The worst performing patients’ stays will be overstated. quartile had risk-adjusted rates of potentially avoidable rehospitalization during the IRF stay that were at or above Overall, the Commission finds that most quality measures 3.5 percent, compared with 1.7 percent or below for the have been stable or improved slightly over the past five Inpatient rehabilitation facility services: Assessing payment adequacy and updating payments 266

297 TABLE Performance on risk-adjusted quality measures varied across IRFs in 2017 10–7 Risk-adjusted rate Ratio of best to Worst Best worst performing performing performing Measure Mean quartile quartile quartile 1.7 Potentially avoidable rehospitalizations during IRF stay 2.6 % 3.5 % % 0.49 % % 8.7 % 4.2 0.48 Discharged to a SNF 6.8 % 73.1 % 79.2 % 1.08 Discharged to the community 76.0 % 5.8 0.59 % % 3.4 4.7 Potentially avoidable rehospitalizations during 30 days after discharge from IRF Motor FIM™ gain 24.0 21.2 26.4 1.25 Cognitive FIM gain 3.9 3.0 4.7 1.34 Note: IRF (inpatient rehabilitation facility), FIM™ (Functional Independence Measure™), SNF (skilled nursing facility). High rates of discharge to the community indicate better quality. High rates of rehospitalization and discharge to SNF indicate worse quality. Mean rates are calculated for all facilities with 25 or more Medicare fee-for- service stays. The motor FIM measures the level of disability in motor functioning on a 91-point scale. The cognitive FIM measures the level of cognitive impairment on a 35-point scale. FIM gain is calculated as the FIM score at discharge minus the FIM score at admission. Higher FIM gain indicates more improvement. Source: MedPAC analysis of Inpatient Rehabilitation Facility–Patient Assessment Instrument data from CMS. which owned almost half of freestanding IRFs in 2017 best performing quartile. (A lower rate of readmissions is and accounted for about a quarter of all Medicare IRF better.) Variation was also observed in the two FIM gain discharges—has good access to capital. This assessment measures, but because these measures are self-reported, is reflected in the chain’s continued expansion. Analysts they could reflect reporting differences more than note that Encompass Health traditionally has prioritized performance differences. building new facilities over acquiring existing facilities, Providers’ access to capital: IRFs appear to which allows the company to maintain control over have adequate access to capital facility size, layout, and amenities. In 2017, the company opened four new facilities and two more in 2018, with More than three-quarters of IRF providers are hospital- two additional facilities scheduled to open in 2019. The based units that would access any necessary capital new facilities are frequently joint ventures with acute care through their parent institutions. Overall, as detailed in hospitals (HealthSouth Corporation 2018). As part of a the hospital chapter, hospitals’ access to capital remained vertical integration strategy, the company has acquired strong in 2017 with a continued high level of bond home health agencies and hospice providers to expand issuances. New construction spending has declined and its PAC business and drive more effective collaboration has shifted more to outpatient than inpatient capacity between its rehabilitation facilities and home health (Conn 2017). Large hospital systems in recent years have agencies. invested significantly in the ambulatory setting, as opposed to the acute inpatient setting, in an effort to access faster Most other freestanding IRFs are independent or local growing markets and offer access to lower cost settings in chains with a limited number of facilities. The extent to a business environment shifting toward value-based care which these providers have access to capital is less clear. (Barclays 2018). IRFs’ access to capital depends in large part on their Market analysts indicate that the IRF industry’s largest total (all-payer) profitability. In 2017, total margins for chain, Encompass Health (formerly HealthSouth)— freestanding IRFs remained healthy, with an aggregate March 2019 | Report to the Congress: Medicare Payment Policy 267

298 FIGURE Medicare margins... X-X FIGURE Program spending for IRF services has grown steadily since 2009 10–1 9 7.9 8 7.7 7.5 7.2 6.9 6.7 7 6.4 6.2 6.0 6 5 4 Dollars (in billions) 3 2 1 0 2013 2012 2015 2017 2016 2014 2011 2010 2009 Note: IRF (inpatient rehabilitation facility). Office of the Actuary 2018. Source: Note: Note and Source are in InDesign. Source: margin of 10.4 percent, up 0.8 percentage point from 11 percent, indicating that many hospitals can manage 2016. Profitability varied by ownership. In 2017, for- their IRF units profitably. Lower margins in hospital-based profit IRFs had an aggregate total margin of 12.5 percent IRFs were driven largely by higher unit costs. compared with 5.6 percent for nonprofit IRFs. Data are Trends in spending and cost growth not available to calculate total margins for hospital-based Notes about this graph: IRFs. However, in 2017, hospitals’ aggregate total margins The Office of the Actuary estimates that Medicare FFS across all lines of service for hospitals with and without • Data is in the datasheet. Make updates in the datasheet. spending for IRF services in fiscal year 2017 was $7.9 IRF units were similar, at 7.0 percent and 7.2 percent, billion (Figure 10-1). Program spending has been growing, • WATCH FOR GLITCHY RESETS WHEN YOU UPDATE DATA!!!! respectively. on average, more than 3 percent per year since 2009. • The column totals were added manually. A combination of increases in the number of Medicare Medicare payments and providers’ costs: • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. beneficiaries receiving care in IRFs (average growth of Medicare margins remained high in 2017 0.5 percent per year) and payment increases averaging 2.6 • I can’t delete the legend, so I’ll just have to crop it out in InDesign. Aggregate Medicare margins grew steadily between 2009 percent contributed to this growth in spending. • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph and 2015 and increased again in 2017 to 13.8 percent Since 2009, payments have been growing faster than costs default when you change the data. (Table 10-8, p. 270). Medicare margins in freestanding (Figure 10-2). From 2009 to 2015, the cumulative growth IRFs were 25.5 percent in 2017, down slightly from a peak • Use paragraph styles (and object styles) to format. in cost per discharge was 8.4 percent, an average of just of 26.7 percent in 2015. Hospital-based IRF margins were • Data was from: R:\Groups\MGA\data book 2007\data book 2007 chp1 1.4 percent per year. The cumulative growth in cost per comparatively low at 1.5 percent in 2017, but one-quarter discharge for freestanding for-profit IRFs was especially of hospital-based IRFs had Medicare margins greater than Inpatient rehabilitation facility services: Assessing payment adequacy and updating payments 268

299 FIGURE Title here... 1-XX FIGURE IRFs’ payments per discharge increased cumulatively more than costs, 2009–2017 10–2 25 Payment per discharge 20.8 Cost per discharge 20 17.3 14.4 15 14.5 11.3 9.3 10 11.9 7.7 8.4 5.0 Cumulative percent change 7.1 5 2.6 5.8 4.9 3.5 2.3 0 2014–2015 2015–2016 2016–2017 2009–2010 2010–2011 2011–2012 2012–2013 2013–2014 Note: IRF (inpatient rehabilitation facility). Percent changes are calculated based on consistent two-year cohorts. Source: MedPAC analysis of Medicare cost report data from CMS. slow over this period, at just 2.2 percent (data not shown). Margins vary widely Note: Note and Source are in InDesign. In contrast, payments per discharge grew more rapidly Financial performance varied across IRFs. In 2017, the Source: than costs, climbing a cumulative 14.4 percent over this aggregate margin for freestanding IRFs (which accounted period (an average of 2.2 percent per year) and 15.1 for 53 percent of Medicare discharges from IRFs) was percent for freestanding for-profit IRFs (latter figure not Notes about this graph: 25.5 percent; hospital-based IRFs had an aggregate shown). These differences in per case cost and payment margin of 1.5 percent (Table 10-8, p. 270). Margins • Data is in the datasheet. Make updates in the datasheet. growth led to a steady rise between 2009 and 2015 in varied by ownership as well, with for-profit IRFs having • I deleted the years from the x-axis and put in my own. aggregate Medicare margins, which climbed from 8.4 a substantially higher aggregate Medicare margin in percent to 13.9 percent (Table 10-8, p. 270; 2009 data not • I had to manually draw tick marks and axis lines because they kept resetting when I changed any data. 2017 than nonprofit IRFs (23.8 percent vs. 2.2 percent). shown). (Hospital-based IRFs are far more likely than freestanding • The dashed line looked ok here, so I didn’t hand draw it. IRFs to be nonprofit.) Among freestanding IRFs, nonprofit Between 2015 and 2016, cost growth outpaced payment • I can’t delete the legend, so I’ll just have to crop it out in InDesign. facilities (which accounted for 7 percent of Medicare growth for the first time since 2009, climbing 3.6 percent, discharges from IRFs) had an aggregate margin of 12.0 • Use direct selection tool to select items for modification. Otherwise if you use the black selection tool, they will reset to graph the fastest rate of cost growth since 2008. However, from percent (data not shown). Freestanding for-profit IRFs default when you change the data. 2016 to 2017, payments per discharge again increased (which accounted for 45 percent of Medicare discharges faster than costs, growing by 3.4 percent compared with • Use paragraph styles (and object styles) to format. from IRFs) had an aggregate margin of 27.8 percent (data 2.6 percent for costs, contributing to an increase in the not shown). Among hospital-based IRFs, the aggregate 2017 Medicare margin to 13.8 percent. From 2015 through margin for nonprofit units (which accounted for 32 percent 2017, aggregate Medicare margins for IRFs remained of Medicare discharges from IRFs) was 0.1 percent, above 13 percent (Table 10-8, p. 270). compared with 6.6 percent for for-profit units (which | March 2019 Report to the Congress: Medicare Payment Policy 269

300 TABLE 10–8 Aggregate FFS Medicare IRF margins remained high in 2017 Margins Share of Medicare discharges, 2017 2004 Type of IRF 2006 2008 2010 2012 2014 2015 2016 2017 All IRFs 100% 16.7 % 12.5 % 9.4 % 8.6 % 11.2 % 12.2% 13.9% 13.3% 13.8% –0.6 1.5 12.2 9.9 3.8 47 0.7 0.7 2.2 0.9 Hospital based 25.2 25.5 21.4 25.8 26.7 Freestanding 53 24.7 17.5 18.2 23.9 2.1 38 Nonprofit 2.1 5.3 10.9 12.8 2.2 1.6 3.5 1.7 19.6 23.8 24.4 16.3 16.8 For profit 22.9 23.6 24.9 24.2 55 N/A N/A N/A N/A N/A N/A 7 Government N/A N/A N/A 9.6 9.0 93 17.0 12.8 Urban 11.6 12.6 14.3 13.6 14.2 7.2 6.4 8.4 9.4 8.6 6.3 Rural 7 13.2 9.0 4.7 Number of beds –9.9 –7.5 –10.9 –6.9 –10.3 –4.9 –3.6 2 3.7 –10.5 1 to 10 –1.2 0.6 –0.2 –0.4 –0.3 11 to 24 –3.3 21 10.5 7.3 1.2 15.0 15.8 10.6 10.0 16.0 13.7 14.0 18.3 48 25 to 64 12.3 17.5 22.4 23.1 20.6 21.0 21.9 17.4 17.8 21.5 29 65 or more Medicare share 2.3 2.4 0.4 5.1 11.1 12.9 19 3.0 <50% 2.9 3.7 56 50% to 75% 17.1 12.6 9.5 9.6 12.5 14.1 16.1 15.4 15.8 20.2 13.9 21.1 >75% 13.5 13.6 20.5 19.6 20.8 20.2 25 Note: FFS (fee-for-service), IRF (inpatient rehabilitation facility), N/A (not applicable). Government-owned facilities operate in a different financial context from other facilities, so their margins are not necessarily comparable. Their margins are not presented separately here, although they are included in the margins for other groups (e.g., “all IRFs”), where applicable. Percentages may not sum to 100 due to rounding. MedPAC analysis of cost report data from CMS. Source: accounted for 10 percent of Medicare discharges from components found that hospital-based IRFs had higher IRFs; data not shown). costs than freestanding IRFs across all cost categories, with the biggest difference manifesting in routine costs Higher unit costs were the primary driver of differences in (Medicare Payment Advisory Commission 2015). financial performance between freestanding and hospital- based IRFs. Freestanding IRFs had a median standardized Nevertheless, one-quarter of hospital-based IRFs had cost per discharge that was 27 percent lower than that of Medicare margins greater than 11 percent, indicating that hospital-based IRFs ($12,069 vs. $16,645, respectively). many hospitals can manage their IRF units profitably. Hospital-based IRFs are far more likely than freestanding Further, despite comparatively low average margins in IRFs to be nonprofit, which could contribute to the disparity hospital-based IRFs, evidence suggests that these units in unit costs. But even nonprofit freestanding IRFs had make a positive financial contribution to their parent a median standardized cost per discharge that was 15 hospitals. For example, aggregate inpatient Medicare percent lower than that of hospital-based IRFs (data not margins for hospitals are consistently higher for hospitals shown). Previous Commission analysis of underlying cost with IRF units versus hospitals without (0.8 percentage Inpatient rehabilitation facility services: Assessing payment adequacy and updating payments 270

301 the primary reason for rehabilitation (24 percent vs. 17 point higher in 2017). Aggregate overall Medicare margins percent). Similarly, freestanding IRFs compared with for hospitals with IRF units were 2.0 percentage points hospital-based IRFs admitted larger shares of cases with higher in 2017. other neurological conditions (19 percent vs. 10 percent) Margins also varied by facility size. In 2017, the aggregate and other orthopedic conditions (10 percent vs. 6 percent). Medicare margin for IRFs with 10 or fewer beds was Notably, the impairment groups of other neurological –10.5 percent, compared with 21.9 percent for IRFs with and other orthopedic conditions encompass a broader 65 or more beds (Table 10-8). These differences are in range of conditions than do other impairment groups. large measure due to differences in economies of scale This clinical heterogeneity can allow favorable selection leading to higher costs in smaller facilities. The median of patients within these groups based on their likely costs standardized cost for IRFs with fewer than 10 beds was of care. Cases with other neurological conditions also 53 percent higher than for IRFs with 65 or more beds count toward the compliance threshold, so IRFs with ($18,636 compared with $12,200; data not shown). higher shares of these cases can more easily meet the Smaller facilities also tend to have lower occupancy rates requirements of the 60 percent rule while keeping down than large facilities (54 percent compared with 68 percent costs. Further, some case types are more profitable than in 2017), also contributing to differences in costs. others, resulting in higher margins for facilities that admit larger shares of those cases. The Commission plans to Medicare margins tended to rise as the share of Medicare examine the relative profitability of the IRF case-mix patients increased. The aggregate Medicare margin was groups in a future analysis. 3.0 percent for IRFs in which fewer than half of discharges were covered by FFS Medicare, compared with 21.1 In general, hospital-based IRFs also have a much larger percent for IRFs in which more than three-quarters of share of cases with extraordinarily high costs. In 2017, 15 discharges were covered by FFS Medicare (Table 10-8). percent of hospital-based IRF cases qualified for high-cost outlier payments, compared with 3 percent of freestanding Numerous factors contribute to lower margins in IRF cases. Indeed, 85 percent of Medicare’s IRF outlier hospital-based IRFs payments were made to hospital-based facilities. Though Several factors account for the disparity in margins these payments diminish losses per case for such outliers, between hospital-based and freestanding IRFs, including they do not completely cover the costs. It is not clear differences in economies of scale, stringency of cost whether the large number of outlier cases in hospital-based control, service mix, and patient mix. Differences in IRFs’ IRFs stems from differences in efficiency, unmeasured assessment of patients’ motor function and cognition case complexity, or both. likely play a role as well. Hospital-based IRFs appear to assess their patients Hospital-based IRFs may be less stringent in cost control differently Historically, evidence suggests that assessments Hospital-based IRFs appear to be less stringent in their of patients’ motor and cognitive function are not reliably cost control. Between 2009 and 2017, costs per case for consistent across IRFs. Some in the industry have hospital-based IRFs grew 21.1 percent, compared with postulated that hospital-based IRFs devote less time to 10.3 percent for freestanding IRFs. Notably, hospital- training assessment staff and verifying the accuracy of based IRFs are far less likely than freestanding IRFs to assessments, resulting in less reliable measures of patients’ be for profit and therefore are likely to be less focused on motor and cognitive function in hospital-based IRFs. controlling costs to maximize returns to investors. We see Others assert that some freestanding IRFs aggressively this effect among freestanding IRFs, where the cumulative assess their patients in a way that maximizes payment. increase in costs per case from 2009 to 2017 for nonprofits To the extent that hospital-based IRFs consistently assess (26.5 percent) far outstripped that of for-profit facilities their patients as less disabled than do their freestanding (8.2 percent). counterparts, for whatever reason, their payments—and margins—will be systematically lower. Hospital-based IRFs have a different mix of patients There are marked differences in hospital-based and Efficient provider analysis freestanding IRFs’ mix of cases. Between 2009 and The Commission is required by the Medicare Prescription 2015, freestanding IRFs compared with hospital-based Drug, Improvement, and Modernization Act of 2003 to IRFs admitted a larger share of patients with stroke as March 2019 | Report to the Congress: Medicare Payment Policy 271

302 Identifying relatively efficient inpatient rehabilitation facilities years. Only IRFs with at least 25 Medicare fee-for- he Commission is required by the Medicare service discharges were included in the analysis. Prescription Drug, Improvement, and Modernization Act of 2003 to consider the T The method we used to assess performance attempts to costs associated with an efficient provider. This year, limit drawing incorrect conclusions about performance we attempted to identify and examine the financial based on poor data. Using three years to categorize performance of inpatient rehabilitation facilities (IRFs) IRFs as efficient (rather than just one year) avoids that had consistently low costs per discharge and high categorizing providers based on random variation or quality. We calculated the cost per discharge using cost on one “unusual” year. After determining whether an report and claims data and adjusted for differences IRF was relatively efficient based on having relatively in area wages; mix of cases; and prevalence of high- low costs and good quality care for three years in a row, cost outliers, short-stay outliers, and transfer cases. we calculated performance on several quality and cost For quality measures, we used risk-adjusted rates of measures in 2017. By first assigning an IRF to a group potentially avoidable rehospitalizations during the IRF (relatively efficient or other) and then examining the stay and risk-adjusted rates of discharge to a skilled group’s performance in the next year, we avoid having nursing facility. To be included in the group of IRFs a facility’s poor data affect both its own categorization that furnished relatively low-cost, high-quality care, and the assessment of the group’s performance. Thus, an IRF had to be (1) in the best performing third of the an IRF’s erroneous data in 2014, 2015, or 2016 could distribution of adjusted cost per discharge or of one of result in its inaccurate assignment to a group, but the quality measures for three consecutive years (2014 because the group’s performance is assessed with data through 2016) and (2) not in the worst performing from 2017, these “bad” data would not directly affect third of the distribution of adjusted cost per discharge the assessment of the group’s performance. ■ or either of the quality measures for three consecutive consider the costs associated with efficient providers. in the median Medicare margin, which was 16.5 percent The Commission follows two principles when selecting for the relatively efficient group compared with 1.0 percent a set of efficient providers. First, the providers must do for other IRFs (Table 10-9). relatively well on both cost and quality metrics. Second, Relatively efficient IRFs were on average larger and had the performance has to be consistent, meaning that the higher occupancy rates compared with other IRFs, leading provider cannot have poor performance on any metric in to greater economies of scale. The mix of cases also any of three consecutive years preceding the year under differed somewhat between the relatively efficient and evaluation. The Commission’s approach is to develop a other IRFs. Relatively efficient IRFs had a higher average set of criteria and then examine how many providers meet case-mix index, more cases with other neurological them. It does not establish a set share (for example, 10 conditions, but smaller shares of stroke cases compared percent) of providers to be considered efficient and then with other IRFs. define criteria to meet that pool size. Although all types of facilities were represented in the This year is the first one in which the Commission has relatively efficient group of IRFs, they were much more examined the financial performance of relatively efficient likely to be freestanding and/or for profit. In fact, over half IRFs. The text box explains how we identified relatively of Encompass Health facilities (formerly HealthSouth) efficient IRFs. Our analysis finds that relatively efficient were in the relatively efficient IRF group. Hospital-based IRFs had lower rehospitalization rates and discharge to nonprofit IRFs were less likely to be in the relatively SNFs than other IRFs. While payment rates to all IRFs efficient group, although they accounted for over a third were similar, standardized costs per discharge for this (37.2 percent) of this group. group were 18 percent lower, leading to a large difference Inpatient rehabilitation facility services: Assessing payment adequacy and updating payments 272

303 TABLE Characteristics of relatively efficient providers, 2017 10–9 Type of IRF Ratio of Relatively relatively efficient to other IRFs efficient IRFs Other IRFs Performance in 2017 Median: 0.91 Rehospitalization rate % % 2.6 2.4 Discharge to SNF rate 0.65 7.0% 4.6% Payment per discharge $20,624 $20,569 1.00 0.82 $13,385 Standardized cost per discharge $16,390 16.5% 1.0% N/A Medicare margin 1.34 1.05 1.28 Facility case-mix index 12.7 12.7 1.00 Length of stay (in days) 61% 1.21 69% Occupancy rate 23 30 1.30 Number of beds Share of discharges that were for: 23.2% 19.5% 0.84 Stroke 10.3% 1.49 6.9% Other neurological conditions Share of facilities that were: 40.5% 20.7% N/A Freestanding 51.2% 34.3% N/A For profit 37.2% 52.5% N/A Hospital-based nonprofit IRF (inpatient rehabilitation facility), SNF (skilled nursing facility). IRFs were identified as “relatively efficient” based on a cost measure (costs per discharge) and two Note: quality measures (rates of readmission and discharge to SNFs) between 2014 and 2016. Relatively efficient IRFs were those in the best third of the distribution for one measure and not in the worst third for any measure in each of the three years. Costs per discharge were standardized for differences in area wages; mix of cases; and prevalence of high-cost outliers, short-stay outliers, and transfer cases. Quality measures were calculated for all facilities with 25 or more fee-for-service stays. “Rehospitalization rate” refers to potentially avoidable rehospitalizations during the IRF stay. High rates of rehospitalization and discharge to SNF indicate worse quality. “Other neurological conditions” includes multiple sclerosis, Parkinson’s disease, polyneuropathy, and neuromuscular disorders. MedPAC analysis of Medicare cost report data, Medicare Provider Analysis and Review data, and Inpatient Rehabilitation Facility–Patient Assessment Instrument Source: data from CMS for 2013 to 2016. an update of 1.0 percent for fiscal year 2018, as • 12 How should Medicare payments change required by MACRA ; and in 2020? • an update of 1.35 percent in 2019 based on an IRF market basket increase of 2.9 percent with offsetting To estimate 2019 payments, costs, and margins with 2017 productivity adjustment and PPACA adjustments of data, the Commission considers policy changes effective 0.8 percent and 0.75 percent, and changes to the high- in 2018 and 2019, including those in the Patient Protection cost outlier fixed loss amount in 2019, which will and Affordable Care Act of 2010 (PPACA) and the lower payments. Medicare Access and CHIP Reauthorization Act of 2015 (MACRA). Those changes that affect our estimate of the Historically, cost growth in this sector has been at or below 2019 margin include: market basket levels, though between 2015 and 2016, cost March 2019 | Report to the Congress: Medicare Payment Policy 273

304 growth exceeded the market basket. We use a three-year the case-mix system. Nevertheless, because of concerns historical average to estimate cost growth in 2018 and about the accuracy of Medicare’s payments for resource- 2019. intensive cases, the Commission continues to believe that an expanded outlier pool is warranted in the near Considering these assumptions, we project an aggregate term. Over the longer term, however, CMS must ensure Medicare margin of 11.6 percent for IRFs in 2019. the accuracy of Medicare’s payments by determining that IRFs’ assessment and scoring consistently reflects For fiscal years 2009 through 2017, the Commission patients’ level of disability. Research is also needed recommended a 0 percent update to the IRF payment rate. to assess variation in costs within the IRF CMGs and In its calculations for fiscal year 2019, however, as the differences in relative profitability across CMGs. In the aggregate margin neared historic highs, the Commission future, CMS could enact payment system reforms that recommended in its March 2017 and March 2018 reports necessitate reassessment of IRF outlier payments and that the Congress reduce IRF payment rates by 5 percent. adjustments to the outlier pool, including a return to a Because such action was not taken and because, in the smaller pool. absence of legislative action, CMS is required by statute to apply an adjusted market basket increase, payments The Commission also reiterates its March 2016 have continued to rise: From 2009 to 2015, the cumulative recommendation that the Secretary conduct focused growth in payments per discharge was 14.4 percent, while medical record review of IRFs that have unusual patterns cost growth was 8.4 percent—well below market basket of case mix and coding and conduct other research levels. In 2016, the gap between payments and costs necessary to improve the accuracy of payments and narrowed somewhat as per case cost growth (3.6 percent protect program integrity. With the shift to using the QRP in aggregate) exceeded payment growth (2.9 percent in functional measures in 2020 to classify cases into CMGs, aggregate) for the first time since 2008. As a result, the it is important that CMS conduct focused medical reviews aggregate margin in 2016 declined but remained high at to ensure consistency in reporting across providers using 13.3 percent. In 2017, payments again increased faster the new measures. than costs, raising margins to 13.8 percent. This high aggregate margin indicates that aggregate Medicare The Commission estimates that reducing the payment payments continue to substantially exceed the costs of rate for IRFs by 5 percent and expanding the outlier pool caring for beneficiaries in IRFs. Absent congressional from 3 percent to 5 percent would decrease total payments action, payments to IRFs will continue to increase in fiscal to IRFs by 5 percent. We estimate the combined effect year 2020 by an estimated 2.7 percent, the largest payment of reducing the payment rate for IRFs by 5 percent and rate update in the past decade. expanding the outlier pool would decrease aggregate payments to freestanding IRFs by 6.2 percent; to hospital- Reducing the payment rate for IRFs would better align based IRFs by 3.8 percent; to for-profit IRFs by 6.0 Medicare payments with the costs of IRF care. The percent; and to nonprofit IRFs by 4.2 percent. Changes Commission continues to believe that the high-cost outlier being made by the Secretary to the CMGs by using the pool should be expanded, as previously recommended in QRP functional measures in place of the FIM, though 2016, to further redistribute payments within the IRF PPS budget neutral, may result in some small shift in payments and reduce the impact of potential misalignments between toward hospital-based and nonprofit facilities in the short IRF payments and costs. Currently, the outlier pool is set term. at 3 percent of total IRF payments. Expanding the outlier pool would increase outlier payments for the most costly RECOMMENDATION 10 cases, ameliorating the financial burden for IRFs that have For 2020, the Congress should reduce the fiscal year 2019 a relatively high share of these cases. The expanded outlier Medicare base payment rate for inpatient rehabilitation pool would be funded by an offset to the national base facilities by 5 percent. payment amount, which would further reduce all CMG payment rates by the same percentage across the board. RATIONALE 10 As noted in our March 2016 and March 2017 reports to The combination of low historical cost growth and the Congress, expanding the outlier pool could increase increasing average payments has resulted in overpayments payments for providers who are less efficient as well as for to IRFs. The high aggregate margin in 2017 and our providers whose patients’ acuity is not well captured by Inpatient rehabilitation facility services: Assessing payment adequacy and updating payments 274

305 update and a forecasted –0.5 percent productivity projected margin for 2019 indicate that Medicare 13 adjustment of the market basket update. payments substantially exceed the costs of caring for Relative beneficiaries. This excess contributes to Medicare’s long- to current law, this recommendation would decrease run sustainability challenges. For every fiscal year since Medicare spending by between $250 million and $750 2009, the Commission has recommended that the update million in 2019 and by between $5 billion and $10 to the IRF payment rate be eliminated or that the payment billion over five years. rate be reduced. However, CMS has been required by Beneficiary and provider statute to apply an adjusted market basket increase each We do not expect this combination of • year. Between 2009 and 2017, the cumulative increase in recommendations to have an adverse effect on either payments per case for all IRFs was 20.8 percent, while Medicare beneficiaries’ access to care or out-of- costs per case rose 14.5 percent, a difference of more than pocket spending. This recommendation could increase 6 percentage points. Reducing the payment rate for IRFs the financial pressure on some providers. We expect by 5 percent would better align Medicare payments with relatively efficient providers will continue to be the costs of IRF care. willing and able to care for Medicare beneficiaries. ■ IMPLICATIONS 10 Spending The payment update for IRFs in fiscal year 2020 • consists of a forecasted 3.2 percent market basket March 2019 | Report to the Congress: Medicare Payment Policy 275

306 Endnotes If we approximate marginal cost as total Medicare cost minus 9 More frequently, Medicare beneficiaries receive inpatient 1 fixed building and equipment cost, then: rehabilitation services in skilled nursing facilities (SNFs), in part because there are many more SNFs than IRFs Marginal profit = (payments for Medicare services – (total nationwide. Medicare costs – fixed building and equipment costs)) / Medicare payments More information about the prospective payment system for 2 IRFs is available at http://medpac.gov/docs/default-source/ The result is a lower bound on the marginal profit because we payment-basics/medpac_payment_basics_18_irf_final_sec. ignore any potential labor costs that are fixed. pdf?sfvrsn=0. The potentially avoidable readmissions we measure are 10 3 Patients with a length of stay of fewer than four days are respiratory-related illness (pneumonia, influenza, bronchitis, assigned to a single CMG, regardless of diagnosis, age, level chronic obstructive pulmonary disease, and asthma); sepsis; of motor or cognitive function, or presence of comorbidities. congestive heart failure; fractures or fall with a major injury; urinary tract or kidney infection; blood pressure management; 4 The 13 conditions are stroke; spinal cord injury; congenital electrolyte imbalance; anticoagulant therapy complications; deformity; amputation of a lower limb; major multiple diabetes-related complications; cellulitis or wound infection; trauma; hip fracture; brain injury; certain other neurological pressure ulcer; medication error or adverse drug reaction; and conditions (multiple sclerosis, Parkinson’s disease, cerebral delirium. palsy, and neuromuscular disorders); burns; 3 arthritis conditions for which appropriate, aggressive, and sustained 11 Our measure of community discharge does not give outpatient therapy has failed; and hip or knee replacement IRFs credit for discharging a Medicare beneficiary to the when it is bilateral, the patient’s body mass index is greater community if the beneficiary is subsequently readmitted to an than or equal to 50, or the patient is age 85 or older. acute care hospital within 30 days of the IRF discharge. In September 2018, the Office of Inspector General (OIG) 5 12 The market basket increase for fiscal year 2018 was 2.6 released a report indicating that many inpatient rehabilitation percent. That update would have been offset by PPACA- stays did not comply with all Medicare coverage and required reductions totaling 1.35 percentage points, for a net documentation requirements for reasonable and necessary update of 1.25 percent. However, Section 411(b) of MACRA care. OIG’s analysis found that only 45 of 220 sampled stays requires that the increase factor for fiscal year 2018 be 1.0 met the requirements (Office of Inspector General 2018). percent. 6 CMS’s major revisions to the compliance threshold policy This market basket forecast was made in the third quarter of 13 in 2004 were to (1) increase the number of conditions that 2018. When setting the update for fiscal year 2020, CMS will count toward the threshold from 10 to 13 and (2) revise the use the most recent forecast available at that time, which may qualifying criteria of major joint replacement—a condition differ from the number we report here. that was commonly treated in IRFs at that time—such that only a certain subset of patients with that condition would count toward the compliance threshold. 7 Other orthopedic conditions, cardiac conditions, and debility are not among the 13 conditions that count toward the compliance threshold, but such cases may count if they have specified comorbidities. Prior Commission analysis of 2013 data showed that less than a third of these cases met the compliance threshold. This analysis of FFS IRF claims and assessment data from 8 2013 excluded cases that were not preceded by an acute care hospital stay within 30 days of the IRF admission. Inpatient rehabilitation facility services: Assessing payment adequacy and updating payments 276

307 References Lau, D. T., J. D. Kasper, D. E. Potter, et al. 2005. Hospitalization Health care services: Initiating coverage of Barclays. 2018. and death associated with potentially inappropriate medication . August 14. hospital sector Archives of prescriptions among elderly nursing home residents. Conn, J. 2017. Consumers fueling outpatient construction. Internal Medicine 165, no. 1 (January 10): 68–74. Modern Healthcare , March. Medicare Payment Advisory Commission. 2018. MedPAC Dejong, G., S. D. Horn, R. J. Smout, et al. 2009. Joint comment on CMS’s proposed rule on the inpatient rehabilitation replacement rehabilitation outcomes on discharge from skilled facility PPS for FY 2019. June 25. nursing facilities and inpatient rehabilitation facilities. Archives Report to the Medicare Payment Advisory Commission. 2016. 90, no. 8 (August): of Physical Medicine and Rehabilitation Washington, DC: MedPAC. Congress: Medicare payment policy. 1284–1296. Report to the Medicare Payment Advisory Commission. 2015. Deutsch, A., C. V. Granger, R. C. Fiedler, et al. 2005. Outcomes Congress: Medicare payment policy. Washington, DC: MedPAC. and reimbursement of inpatient rehabilitation facilities and subacute rehabilitation programs for Medicare beneficiaries with Mustard, C. A., and T. Mayer. 1997. Case-control study of hip fracture. Medical Care 43, no. 9 (September): 892–901. exposure to medication and the risk of injurious falls requiring hospitalization among nursing home residents. American Journal Galloway, R. V., C. V. Granger, A. M. Karmarkar, et al. 2013. of Epidemiology 145, no. 8 (April 15): 738–745. The Uniform Data System for Medical Rehabilitation: Report of patients with debility discharged from inpatient rehabilitation Office of Inspector General, Department of Health and Human American Journal of Physical Medicine programs in 2000–2010. Many inpatient rehabilitation facility stays did Services. 2018. 92, no. 1 (January): 14–27. & Rehabilitation not meet Medicare coverage and documentation requirements . A–01–15–00500. Washington, DC: OIG. Grabowski, D. C., K. A. Stewart, S. M. Broderick, et al. 2008. Predictors of nursing home hospitalization: A review Office of the Actuary, Centers for Medicare & Medicaid Services, of the literature. 65, no. 1 Medical Care Research and Review Department of Health and Human Servivces. 2017. Personal (February): 3–39. communication of author with Theresa Bean, November 1. HealthSouth Corporation. 2018. Annual report (Form 10–K) for Ottenbacher, K. J., A. Karmarkar, J. E. Graham, et al. 2014. fiscal year ending December 31, 2017. Filing submitted to the Thirty-day hospital readmission following discharge from Securities and Exchange Commission. February 28. postacute rehabilitation in fee-for-service Medicare patients. 311, no. 6 Journal of the American Medical Association Kane, R. L., G. Keckhafer, S. Flood, et al. 2003. The effect of (February 12): 604–614. Journal of the American Geriatrics Evercare on hospital use. Society 51, no. 10 (October): 1427–1434. Schneider, J. C., P. Gerrard, R. Goldstein, et al. 2013. The impact of comorbidities and complications on burn injury inpatient Konetzka, R. T., W. Spector, and M. R. Limcangco. 2008a. rehabilitation outcomes. 5, Physical Medicine and Rehabilitation Reducing hospitalizations from long-term care settings. Medical no. 2 (February): 114–121. Care Research and Review 65, no. 1 (February): 40–66. Schneider, J. C., P. Gerrard, R. Goldstein, et al. 2012. Predictors Konetzka, R. T., S. C. Stearns, and J. Park. 2008b. The staffing- of transfer from rehabilitation to acute care in burn injuries. Health Services outcomes relationship in nursing homes. Journal of Trauma and Acute Care Surgery 73, no. 6 (December): Research 43, no. 3 (June): 1025–1042. 1596–1601. Development of inpatient Kramer, A., M. Lin, R. Fish, et al. 2015. rehabilitation facility quality measures: Potentially avoidable . readmissions, community discharge, and functional improvement Report prepared for the Medicare Payment Advisory Commission. Washington, DC: MedPAC. March 2019 | Report to the Congress: Medicare Payment Policy 277

308

309 CHAPTER 11 Long-term care hospital services

310 RECOMMENDATION 11 For 2020, the Secretary should increase the fiscal year 2019 Medicare base payment rates for long-term care hospitals by 2 percent. COMMISSIONER VOTES: YES 17 • NO 0 • NOT VOTING 0 • ABSENT 0

311 CHAPTER 11 Long-term care hospital services In this chapter Chapter summary Long-term care hospitals (LTCHs) provide care to beneficiaries who need • Are Medicare payments hospital-level care for relatively extended periods. To qualify as an LTCH for adequate in 2019? Medicare payment, a facility must meet Medicare’s conditions of participation How should Medicare • for acute care hospitals and, for certain Medicare patients, have an average payments change in 2020? length of stay greater than 25 days. In 2017, Medicare spent $4.5 billion on care provided in LTCHs nationwide. About 103,000 fee-for-service (FFS) beneficiaries had roughly 116,000 LTCH stays. On average, Medicare FFS beneficiaries accounted for about two-thirds of LTCHs’ discharges. In fiscal year 2016, CMS began implementing a dual payment-rate structure for LTCHs that decreased payment rates for certain cases that do not meet criteria specified in the Pathway for SGR Reform Act of 2013. The extent to which LTCHs alter admission patterns for cases that meet the criteria and are thus paid the standard LTCH prospective payment system rate will ultimately determine the industry’s financial performance under Medicare. We focus some analyses on a cohort of LTCHs with a high share (85 percent or more) of cases meeting the criteria in 2017, consistent with the goals of the dual payment-rate policy. Assessment of payment adequacy We have no direct measures of beneficiaries’ Beneficiaries’ access to care— access to needed LTCH services. While we consider the capacity and supply March 2019 | Report to the Congress: Medicare Payment Policy 281

312 of LTCH providers and changes over time in the volume of services they furnish, we expect reductions in these metrics since the implementation of the new dual payment-rate structure that began in fiscal year 2016, as mandated by the Pathway for SGR Reform Act of 2013. • Capacity and supply of providers— The number of LTCHs began to decrease in 2013, but the decline has been more rapid since the implementation of the dual payment-rate structure. We estimate that the number of LTCHs decreased by 4.1 percent from 2016 to 2017 and by an additional 2.3 percent from 2017 to 2018. However, the average LTCH occupancy rate was 64 percent in 2017, suggesting that LTCHs have adequate capacity in the markets they serve. • Volume of services— From 2016 to 2017, the number of LTCH cases decreased by 7.3 percent, continuing a four-year trend that began in 2013. The number of LTCH cases per FFS beneficiary also declined during this period (2016 to 2017) by 7 percent. However, from 2016 to 2017, the number of LTCH cases that met the criteria per 10,000 FFS beneficiaries increased by 3.6 percent. Marginal profit— In 2017, marginal profit, an indicator of whether LTCHs • with excess capacity have an incentive to admit Medicare patients, averaged about 14 percent across all LTCHs. The marginal profit in 2017 was about 6 percentage points lower than in 2016, reflecting payment reductions associated with the implementation of the dual payment-rate structure. For LTCHs with a high share (85 percent or more) of cases meeting the criteria specified in the Pathway for SGR Reform Act of 2013, marginal profit totaled 16 percent, about 1 percentage point lower than in 2016. Quality of care— Consistent with prior years, non-risk-adjusted rates of direct LTCH to acute care hospital readmission, death in the LTCH, and death within 30 days of discharge were stable across all LTCH cases. Providers’ access to capital— LTCHs have begun altering their cost structures and referral patterns in response to the dual payment-rate structure, which reduces payment for cases that do not meet the criteria specified in law. This transition, coupled with payment reductions to annual updates required by statute, have limited opportunities for growth in the near term and reduced the industry’s need for capital. Medicare payments and providers’ costs— From 2012 through 2015, Medicare payments increased, but more slowly than provider costs. Payments per case remained stable from 2015 through 2016, resulting in an aggregate 2016 Medicare margin of 3.9 percent across all cases. The first year that all LTCHs began transitioning to the dual payment-rate structure was 2017. The extent to which each Long-term care hospital services: Assessing payment adequacy and updating payments 282

313 facility admits cases that meet the criteria directly impacts the Medicare payments it receives and may affect the costs incurred in providing care. In 2017, the aggregate Medicare margin was –2.2 percent. However, when we consider a cohort of LTCHs with a high share of cases that met the criteria, and thus admission patterns consistent with the goals of the dual payment-rate structure, the Medicare margin remained positive. Indeed, in 2017, LTCHs with 85 percent or more of Medicare cases that met the criteria had a Medicare margin of 4.6 percent. We expect continued changes in admission patterns and cost structures of LTCHs in response to the implementation of the dual payment-rate structure. We project that LTCHs’ aggregate Medicare margin for facilities with more than 85 percent of Medicare discharges that meet the criteria will be 1.2 percent in 2019. On the basis of these indicators, and in the context of recent changes in payment policy, our recommendation for fiscal year 2020 would increase the 2019 LTCH payment rate by 2 percent. This update supports LTCHs in their provision of safe and effective care for Medicare beneficiaries meeting the criteria for payment at the standard LTCH prospective payment system rate. ■ March 2019 | Report to the Congress: Medicare Payment Policy 283

314

315 LTCHs historically have constituted about 1 percent of Background post-acute care (PAC) use; however, this share varies substantially across ACH diagnoses. For example, about Patients with chronic critical illness—those who exhibit 60 percent of beneficiaries requiring a tracheostomy metabolic, endocrine, physiologic, and immunologic with more than 96 hours of ventilator support in an ACH abnormalities that result in profound debilitation and often were discharged to an LTCH, as were about 15 percent ongoing respiratory failure—frequently need hospital- of beneficiaries discharged with either septicemia or level care for extended periods. Some are treated in respiratory failure requiring mechanical ventilation for long-term care hospitals (LTCHs). These facilities can be more than 96 hours. The variation in LTCH use suggests freestanding or colocated with other hospitals as hospitals that many Medicare beneficiaries receive care during an within hospitals or satellites. To qualify as an LTCH ACH stay or during an ACH stay that is subsequently for Medicare payment, a facility must meet Medicare’s followed by a PAC stay in a non-LTCH setting. However, conditions of participation for acute care hospitals (ACHs) in 2013, close to 80 percent of ventilator-dependent and, for certain Medicare patients, have an average length beneficiaries using PAC were treated in LTCHs compared 1 of stay greater than 25 days. In aggregate, LTCHs had an with 14 percent in skilled nursing facilities (SNFs) average length of stay of 26.3 days; by comparison, the (Medicare Payment Advisory Commission 2017a). average Medicare length of stay in ACHs is about 5 days. In fiscal year 2016, CMS began phasing in a payment In 2017, Medicare spent $4.5 billion on care provided change for LTCH cases that do not meet certain criteria in LTCHs nationwide. About 103,000 beneficiaries had specified in the Pathway for SGR Reform Act of 2013 roughly 116,000 LTCH stays. On average, Medicare fee- (see text box on the development of the long-term care for-service (FFS) beneficiaries accounted for about two- 4 hospital dual payment-rate structure, pp. 288–289). thirds of LTCHs’ discharges. Under this new dual payment-rate structure, Medicare Since October 2002, Medicare has paid LTCHs cases are paid the standard LTCH PPS rate if the patient prospective per discharge rates based primarily on the had an immediately preceding ACH stay that included 2 patient’s diagnosis and the facility’s wage index. Under 3 or more days in an intensive care unit (ICU) or if the this prospective payment system (PPS), LTCH payment patient received mechanical ventilation services for at rates are based on the Medicare severity long-term least 96 hours in the LTCH. These cases are referred to care diagnosis related group (MS–LTC–DRG) patient as “cases meeting the criteria.” LTCH cases not meeting classification system, which groups patients primarily that specified criteria receive a “site-neutral” rate based according to diagnoses and procedures. MS–LTC–DRGs on the lesser of an IPPS-comparable amount or 100 include the same groupings used in ACHs paid under the percent of the cost for the case. For the first four years of inpatient PPS (IPPS) but have relative weights specific to implementation, cases that do not meet the criteria receive LTCH patients that reflect the average relative costliness payment of 50 percent of the standard LTCH PPS rate of cases in the group compared with that of the average and 50 percent of the site-neutral rate. Given this phase-in LTCH case. The LTCH PPS has outlier payments for period, the policy will not be fully in effect for all LTCH 3 patients who are extraordinarily costly. The LTCH PPS facilities until fiscal year 2021. However, data from fiscal pays differently for short-stay outlier cases (patients with year 2017 include the partial phase-in of the dual payment- shorter-than-average lengths of stay), reflecting CMS’s rate structure across all LTCHs. contention that Medicare should adjust payment rates for Because the impact of the dual payment-rate structure patients with relatively short stays to reflect the reduced is expected to be substantial, we focus some analyses costs of caring for them (see text box discussing short-stay on LTCHs that have a high share of cases that meet the outliers, p. 286). criteria, consistent with the goals of the dual payment-rate LTCHs are not distributed uniformly across the country. structure, which creates a financial disincentive for LTCHs Due in part to state certificate-of-need programs that to admit Medicare cases that do not meet the criteria. We prevent or limit the opening of certain types of health care define this subgroup of LTCHs as those with more than facilities in some states, many areas have no LTCHs, while 85 percent of their Medicare cases meeting the criteria in 5 others have a high concentration of them, underscoring the 2017, accounting for about 30 percent of LTCHs. fact that some medically complex patients can be treated appropriately in other settings. March 2019 | Report to the Congress: Medicare Payment Policy 285

316 Payment for short-stay outliers in long-term care hospitals the incremental payments for short-stay cases to the n the long-term care hospital (LTCH) payment provider’s incremental costs. system, Medicare adjusts payments for cases with short stays. CMS defines a short-stay outlier I Beginning in fiscal year 2018, CMS changed how (SSO) case as having a length of stay less than or LTCHs are paid for SSOs. Instead of paying LTCHs equal to five-sixths of the geometric mean length of for SSO cases based on the lesser of four payment stay for the case type. The SSO policy reflects CMS’s rates, CMS now pays a rate equal to an amount contention that patients with lengths of stay similar to that is a blend of the IPPS amount for the Medicare those in acute care hospitals (ACHs) should be paid at severity–diagnosis related group and 120 percent of rates comparable with the cases paid under the ACH the LTCH per diem payment amount up to the full inpatient prospective payment system (IPPS). LTCH prospective payment system (PPS) standard federal payment rate. As the length of stay for the SSO Previously, the Commission expressed concern increases, the blended payment includes an increasing regarding the financial incentives associated with the share of payment attributable to the LTCH per diem. payment structure of the SSO policy and the inherent The longer the length of stay, the more closely payment payment cliffs it created. Historically, Medicare paid resembles the full LTCH PPS amount, greatly reducing LTCHs for SSO discharges based on the lesser of four the payment cliff that existed under the prior policy. payment calculations, including up to the full LTCH 6 CMS also updated this policy to no longer differentiate standard payment amount. This payment structure between the SSO cases and cases with “very short” created large differences between the SSO payment and lengths of stay. the full LTCH payment, resulting in a strong financial incentive for LTCHs to keep patients until their lengths In fiscal year 2017, the prior SSO structure remained of stay exceed the SSO threshold for the relevant in place. Under this structure, 32.8 percent of LTCH case type. In its March 2017 report to the Congress, discharges received SSO payment adjustments, an the Commission stated that CMS could reduce the increase from 2016. This increase in part reflects financial incentives to increase a beneficiary’s length reductions in the length of stay for cases that do of stay beyond the SSO threshold by better aligning not meet the criteria under the dual payment-rate structure. ■ Beneficiaries’ access to care: Expected Are Medicare payments adequate in reductions in supply and volume continue, without affecting access to care 2019? We have no direct measures of beneficiaries’ access to To address whether payments for 2019 are adequate needed LTCH services. The absence of LTCHs in many to cover the costs that providers incur in furnishing areas of the country does not necessarily indicate an services to Medicare beneficiaries, we examine several inadequacy of supply since beneficiaries in areas without indicators of payment adequacy. Specifically, we assess LTCHs have access to similar services in other settings, beneficiaries’ access to care (by examining the capacity including ACHs and some skilled nursing facilities and supply of LTCH providers, changes over time in the (SNFs). However, in 2013, among PAC users requiring volume of services furnished, and providers’ willingness mechanical ventilation, close to 80 percent of these to admit Medicare beneficiaries), quality of care, beneficiaries were treated in LTCHs (Medicare Payment providers’ access to capital, and the relationship between Advisory Commission 2017a). In 2018, LTCHs were Medicare payments and providers’ costs. located in just 8.5 percent of counties, but these LTCHs Long-term care hospital services: Assessing payment adequacy and updating payments 286

317 TABLE The number of LTCHs decreased in 2017 11–1 Congressionally Congressionally imposed imposed moratorium moratorium Average annual change a a a 2015 2016–2017 2016 2017 2012–2016 Type of LTCH 2014 2012 2013 LTCHs paid under b 413 416 421 394 % –4.1 % –0.6 411 412 the LTCH PPS LTCHs with valid b 426 cost reports 411 407 398 392 –2.2 –1.1 399 c 373 –2.8 401 Urban 385 378 389 373 c Rural 25 26 26 19 18 20 11.1 78 78 71 –2.3 0.0 71 66 73 Nonprofit 320 328 315 312 –0.6 –2.5 309 For profit 307 20 18 19 17 Government –5.4 15 16 –6.3 Note: LTCH (long-term care hospital), PPS (prospective payment system). The Medicare, Medicaid, and SCHIP Extension Act of 2008 and subsequent legislation imposed a moratorium on new LTCHs and new LTCH beds in existing facilities from December 29, 2007, through December 28, 2012. The Pathway for SGR Reform Act of 2013 and subsequent legislation implemented a new moratorium from April 1, 2014, through September 30, 2017. a Data for 2013 through 2015 should not be compared with prior or subsequent years because of an anomalous number of facilities that underwent an acquisition and changes in the cost reporting period. b Data for hospitals paid under the LTCH PPS are from the Provider of Services file based on the applicable fiscal year. The count of hospitals with valid cost reports is based on each hospital’s cost reporting period that most aligns with the fiscal year; however, this timing contributes to differences between the two facility counts. c In addition to the anomalous numbers of facilities that underwent an acquisition and changes in the cost reporting period, there were new core-based statistical area codes for LTCHs that CMS adopted beginning fiscal year 2015. This change reclassified as urban several facilities previously classified as rural, and therefore the number of facilities between 2014 and 2015 should not be compared. Source: MedPAC analysis of cost report data and the Medicare Provider of Services file from CMS. a limited moratorium on new LTCHs and new beds served beneficiaries from over 90 percent of counties in existing LTCHs from December 29, 2007, through nationwide. A recent study found that 80 percent of December 28, 2012. During that time, new LTCHs were Medicare beneficiaries reside in a hospital referral region able to enter the Medicare program only if they met with at least one LTCH (National Association of Long 7 specific exceptions to the moratorium. Term Care Hospitals 2017). At the median, beneficiaries The Pathway traveled about 17 miles to receive LTCH care. About 10 for SGR Reform Act of 2013 and subsequent legislation percent of beneficiaries traveled in excess of 90 miles. implemented a new moratorium from April 1, 2014, 8 While changes in the overall capacity and supply of LTCH through September 30, 2017. providers and in the volume of services they furnish might We examined Medicare cost report data to assess the typically suggest declining access to care, we fully expect number of LTCH beds and facilities. Growth in the reductions in these metrics following the implementation number of LTCHs filing Medicare cost reports slowed of the dual payment-rate structure that began in fiscal year considerably in the later years of the moratorium (Table 2016. 11-1). Between 2012 and 2015, a larger-than-usual number of facilities made changes to their cost reporting Capacity and supply of providers: Number of period, thereby affecting the number of facilities with LTCHs began to decrease in 2013 sufficient cost report data to be used for this payment The Medicare, Medicaid, and SCHIP Extension Act of 9 adequacy analysis. Between 2012 and 2017, the number 2007 (MMSEA) and subsequent legislation imposed March 2019 | Report to the Congress: Medicare Payment Policy 287

318 Development of the long-term care hospital dual payment-rate structure have multiple organ failures, and patients with he Pathway for SGR Reform Act of 2013 septicemia and other complex infections (Dalton et al. mandated changes to the long-term care 2012). hospital (LTCH) prospective payment system, T including limiting the standard LTCH payment rate Analysis of findings from the Post-Acute Care to cases that spent at least three days in an intensive Payment Reform Demonstration, which tested the care unit (ICU) during an immediately preceding use of a standardized patient assessment tool in acute care hospital (ACH) stay or to discharges that various post-acute care settings, revealed meaningful received an LTCH principal diagnosis indicating differences in the intensity of nursing care and prolonged mechanical ventilation. In March 2014, the nutritional, rehabilitation, and physician services Commission recommended that the LTCH payment between LTCH users and other post-acute care (PAC) system be reformed to better align payments for both users. Length of time in an ICU during an immediately chronically critically ill (CCI) and non-CCI cases preceding ACH stay was a distinguishing characteristic across LTCH and ACH settings. of patients who used LTCHs as opposed to patients who used only skilled nursing facilities, inpatient Defining an LTCH patient rehabilitation facilities, or care provided by home For almost two decades, given the variation in LTCH health agencies. Post-acute care episodes that had a use across the country and the relatively high cost of preceding ACH ICU stay of seven days or more were providing care to Medicare beneficiaries in LTCHs, found only among LTCH users (Gage et al. 2011). policymakers and researchers alike have attempted LTCH care is commonly used for other, less acutely to define the type of patient most appropriate for the ill, patients as well. These patients may require lengthy LTCH setting. Recent research using data from 2012 hospitalizations and subsequent post-acute care, showed that, after adjusting for case mix, about half but they do not have (or no longer have) intensive of the variation in LTCH use is explained by patient nursing care needs (Centers for Medicare & Medicaid factors, including the presence of a tracheostomy. This Services 2013). Research has consistently shown research found that the remaining variation in LTCH that caring for these lower acuity patients in LTCHs use is explained by regional and hospital factors, increases Medicare expenditures without demonstrable including the proximity of a beneficiary’s discharging improvements in quality of care or outcomes (Koenig ACH to an LTCH (Makam et al. 2018). et al. 2015). Yet such patients have historically made Defining the most medically complex patients who up a substantial share of cases in most LTCHs. might be the most appropriate for LTCH-level care has been elusive. Some clinicians have described Commission recommendation for long-term CCI patients as exhibiting metabolic, endocrine, care hospitals physiologic, and immunologic abnormalities that result The Commission has maintained that LTCHs should in profound debilitation and often ongoing respiratory serve only the most medically complex patients failure (Nierman and Nelson 2002). Many of these and has determined, with general agreement from abnormalities and debilities in hospital patients are industry representatives, that the best available proxy not readily identifiable using available administrative for intensive resource needs in LTCH patients is data. However, the research literature is consistent in ICU length of stay during an immediately preceding describing such patients as having long ACH stays ACH stay. The Commission has also long held that with heavy use of intensive care services. Another payments to providers should be properly aligned study defined LTCH-appropriate patients as ventilator- with patients’ service needs. Further, subject to risk dependent with major comorbidities, patients who differentials, payment for the same services should (continued next page) Long-term care hospital services: Assessing payment adequacy and updating payments 288

319 Development of the long-term care hospital dual payment-rate structure (cont.) ACH stay immediately preceding LTCH admission be comparable regardless of where the services are and for which: provided. • the ACH stay included at least 3 days in an The Commission recommended that the Congress intensive care unit or limit standard LTCH payments to cases that spent eight or more days in an ICU during an immediately • the discharge was assigned to the Medicare preceding ACH stay (Medicare Payment Advisory severity long-term care diagnosis related group Commission 2014). The Commission’s analysis of (MS–LTC–DRG) based on the receipt of inpatient prospective payment system (IPPS) claims mechanical ventilation services for at least 96 data found that cases with eight or more days in an hours. ICU accounted for about 6 percent of all Medicare IPPS discharges and had a geometric mean cost per All other LTCH discharges (cases that do not meet discharge that was four times that of IPPS cases the criteria)—including any discharges assigned with seven or fewer ICU days. Further, these cases to psychiatric or rehabilitation MS–LTC–DRGs, were concentrated in a small number of Medicare regardless of intensive care unit use—are paid a site- severity–diagnosis related groups that correspond with neutral amount (an amount based on either Medicare’s descriptions of LTCH patients provided by critical IPPS or 100 percent of the costs of the case, whichever care clinicians (Dalton et al. 2012). is lower). These site-neutral payments are being phased in over a four-year period. In cost reporting Setting the ICU length of stay threshold for CCI periods starting fiscal year 2016, cases that do not cases at eight days captures a large share of LTCH meet the criteria receive a blended rate of one-half cases requiring prolonged mechanical ventilation—a the standard LTCH payment and one-half the site- service specialty of many LTCHs. However, the neutral payment. In cost reporting periods starting on Commission was concerned that LTCH care could be or after October 1, 2019, these cases will receive 100 appropriate for some patients requiring mechanical percent of the site-neutral payment rate. Given LTCHs’ ventilation even if they did not spend eight or more varying cost reporting periods, the Commission days in an ICU during an immediately preceding ACH expects fiscal year 2021 to be the first full year in stay. The Commission therefore recommended that which this policy is completely phased in. patients requiring prolonged ventilation care qualify for CCI status. For LTCH cases that did not spend Congressionally mandated facility-level criteria eight or more days in an ICU during an immediately To qualify as an LTCH for Medicare payment, a preceding ACH stay, the Commission recommended facility must meet Medicare’s hospital conditions that the Secretary of Health and Human Services of participation and certain Medicare patients must set the payment rates equal to those of ACHs. The have an average length of stay greater than 25 days. Commission recommended that savings from this The Pathway for SGR Reform Act of 2013 loosens policy be used to create additional inpatient outlier these criteria such that, beginning in fiscal year 2016, payments for CCI cases in IPPS hospitals. CMS calculates the LTCH average length of stay only for Medicare fee-for-service cases that are not paid Congressionally mandated patient-level criteria the site-neutral rate. However, the Pathway for SGR The Pathway for SGR Reform Act of 2013 established Reform Act of 2013 requires that, for cost reporting “site-neutral” payments for certain cases in LTCHs, periods starting on or after October 1, 2019, at least beginning in fiscal year 2016. Under the law, the half of an LTCH’s cases meet the criteria to continue LTCH payment rate applies only to qualifying LTCH to be paid the standard LTCH prospective payment discharges (cases that meet the criteria) that had an system rate. ■ March 2019 | Report to the Congress: Medicare Payment Policy 289

320 TABLE The number of Medicare LTCH cases and users 11–2 continued to decrease between 2016 and 2017 Average annual change 2016– 2012– 2015– 2016 2016 2012 2013 2014 2017 2015 2017 2015 137,827 Cases 140,463 % 116,424 133,984 125,586 –2.3 % –4.2 % –7.3 131,129 37.7 Cases per 10,000 FFS beneficiaries –5.7 35.4 36.6 34.4 –7.0 32.5 –3.0 30.2 Spending per FFS beneficiary $146.64 $148.78 $140.17 $141.61 –12.5 –2.0 $115.44 $131.94 –5.9 $39,493 –0.2 $40,015 $40,719 $40,656 $38,253 1.0 Payment per case –5.9 $40,070 26.2 26.3 26.3 26.6 26.8 26.5 Average length of stay (in days) 0.4 1.1 –2.2 103,322 Users 121,532 118,288 116,088 111,171 –7.1 –2.1 –4.2 123,652 LTCH (long-term care hospital), FFS (fee-for-service). Note: MedPAC analysis of Medicare Provider Analysis and Review data from CMS and the annual reports of the Boards of Trustees of the Medicare trust funds. Source: of LTCHs with valid cost reports decreased by about 7 been 1 percentage point to 2 percentage points higher percent from 426 to 398, or about a 1.4 percent average than for nonprofit LTCHs. However, in 2017, occupancy annual decrease, roughly consistent with the 1.3 percent rates dropped to 64 percent, and the difference between average annual decrease in hospitals paid under the LTCH occupancy rates at for-profit and nonprofit LTCHs 10 PPS in the Provider of Services file. widened. For-profit LTCHs had an occupancy rate of 65 From 2017 to 2018, percent compared with 59 percent for nonprofit LTCHs the number of LTCHs decreased by another 2.3 percent (data not shown). In aggregate, LTCHs with a high share (data not shown), totaling a nearly 10 percent decline since of Medicare cases meeting the criteria had an occupancy 2012. Cost report data indicate that the number of LTCH rate of 69 percent in 2017. beds nationwide decreased about 2.1 percent annually from 2012 through 2017 (data not shown). Volume of services: Number of LTCH users Consistent with historical trends, the Commission decreased estimates that, in 2017, more than 75 percent of LTCHs Beneficiaries’ use of LTCH services suggests that access were for profit, and 95 percent were located in urban is adequate. The volume of services provided by LTCHs areas. In our analysis of urban and rural facilities, the data has fluctuated in response to payment policy changes. presented in Table 11-1 (p. 287) beginning in 2015 are not Following a moratorium on new facilities and new beds comparable with prior years because CMS adopted new in existing facilities, from 2012 through 2015, the number core-based statistical area codes based on the 2010 census of LTCH cases per capita decreased by 3.0 percent (Table for LTCHs that year, in addition to the aforementioned 11-2). From 2015 to 2016, as the new dual payment-rate anomalous cost reporting trends. This change reclassified structure was implemented, LTCH cases per 10,000 FFS as urban several facilities previously classified as rural. beneficiaries further dropped by 5.7 percent and by 7.0 percent from 2016 to 2017. These decreases occurred, in Aggregate occupancy rates for LTCHs from 2012 through part, because LTCHs changed their admitting practices to 2016 remained largely unchanged at 66 percent, and, admit fewer cases that do not meet the criteria to be paid historically, occupancy rates for for-profit LTCHs have the standard LTCH PPS rate. Long-term care hospital services: Assessing payment adequacy and updating payments 290

321 TABLE The volume and share of cases meeting the criteria for 11–3 the standard LTCH PPS rate increased from 2016 to 2017 Percent change 2016 2017 2015 2015–2016 2016–2017 % 72,429 72,318 74,666 –0.2 % 3.2 Cases meeting the criteria 64 % 55 % 58 % Share of all LTCH cases Cases per 10,000 FFS beneficiaries –1.7 3.6 19.0 19.4 18.7 Spending (in billions) $3.3 $3.4 –0.1 3.0 $3.3 3.4 $87.90 $86.40 $89.30 –1.7 Spending per FFS beneficiary 0.0 $46,127 Payment per case $46,217 $46,223 –0.2 Length of stay (in days) 28.5 27.9 27.9 –2.0 –0.1 Note: LTCH (long-term care hospital), PPS (prospective payment system), FFS (fee for service). “Cases meeting the criteria” refers to Medicare discharges that meet the criteria specified in the Pathway for SGR Reform Act of 2013 to be paid the standard LTCH PPS rate. MedPAC analysis of Medicare Provider Analysis and Review data from CMS and the annual reports of the Boards of Trustees of the Medicare trust funds. Source: Since 2015, the share of Medicare cases in LTCHs The higher rate of LTCH use by African American meeting the criteria increased by 9 percentage points to 64 beneficiaries may be due to the concentration of LTCHs percent in 2017, driven primarily by a reduction in volume in areas of the country with larger African American of cases not meeting the criteria (Table 11-3). From populations (Dalton et al. 2012, Kahn et al. 2010). Another 2012 through 2017, the total number of cases meeting contributing factor may be a greater incidence of critical the criteria in LTCHs remained stable, with a decrease illness in this population (Mayr et al. 2010). At the same occurring between 2014 and 2015 but an increase between time, African American Medicare beneficiaries may be 2016 and 2017. Controlling for changes in the number of more likely to opt for LTCH care since they are less likely FFS beneficiaries, we found the number of LTCH cases than White beneficiaries to elect hospice care (Medicare meeting the criteria increased by 3.6 percent from 2016 to Payment Advisory Commission 2017b). 2017. LTCH patient discharges are concentrated in a relatively In 2017, Medicare FFS beneficiaries accounted for 63 small number of diagnosis groups. In fiscal year 2017, the percent of LTCH discharges and just over half of patient top 20 LTCH diagnoses made up 63 percent of all LTCH days in aggregate, representing a slight decline in the discharges. The most frequently occurring diagnosis share of Medicare FFS discharges and patient days was pulmonary edema and respiratory failure (Medicare following a period of relative stability since 2010. In 2016, severity–long-term care diagnosis related group (MS– dual-eligible beneficiaries (enrolled in both Medicare LTC–DRG) 189). Over 35 percent of LTCH cases were and Medicaid) accounted for about 45 percent of FFS diagnoses that included respiratory conditions, an increase 11 Medicare days (data not shown). from 2016. Compared with all Medicare beneficiaries, those admitted Not unexpectedly, the patient diagnoses become even to LTCHs are disproportionately disabled (under age 65), more concentrated when we consider cases from the over age 85, or diagnosed with end-stage renal disease. cohort of LTCHs with the highest share of cases (85 They are also more likely to be African American. percent or more) meeting the criteria for the standard March 2019 | Report to the Congress: Medicare Payment Policy 291

322 TABLE 11–4 Among LTCHs with a high share of cases meeting the criteria for the standard LTCH PPS rate, the top 20 MS–LTC–DRGs made up 77 percent of discharges in 2017 MS–LTC– Share of cases Description Discharges DRG 189 Pulmonary edema and respiratory failure 5,888 22.1% 5,530 207 Respiratory system diagnosis with ventilator support 96+ hours 20.8 208 Respiratory system diagnosis with ventilator support ≤96 hours 1,157 4.4 3.8 Septicemia without ventilator support 96+ hours with MCC 1,021 871 949 Aftercare with CC/MCC 803 3.0 166 Other respiratory system OR procedures with MCC 681 2.6 629 682 Renal failure with MCC 2.4 Tracheostomy with ventilator support 96+ hours or primary diagnosis except face, mouth and neck 4 2.3 620 without major OR procedure 2.0 529 Extensive OR procedure unrelated to principal diagnosis with MCC 981 539 Osteomyelitis with MCC 425 1.6 1.5 405 Respiratory infections and inflammations with MCC 177 1.5 Skin ulcers with MCC 396 592 190 360 Chronic obstructive pulmonary disease with MCC 1.4 862 Postoperative and post-traumatic infections with MCC 1.3 348 1.2 314 Other circulatory system diagnoses with MCC 330 1.2 313 Complications of treatment with MCC 919 559 Aftercare, musculoskeletal system, and connective tissue with MCC 297 1.1 291 Heart failure and shock with MCC 291 1.1 Degenerative nervous system disorders with MCC 1.1 281 56 0.8 371 Major gastrointestinal disorders and peritoneal infections with MCC 224 Top 20 MS–LTC–DRGs 20,528 77.2 Note: MS–LTC–DRG (Medicare severity–long-term care diagnosis related group), LTCH (long-term care hospital), PPS (prospective payment system), MCC (major complication or comorbidity), CC (complication or comorbidity), OR (operating room). MS–LTC–DRGs are the case-mix system for LTCH facilities. “Cases meeting the criteria” refers to Medicare discharges that meet the criteria specified in the Pathway for SGR Reform Act of 2013 to be paid the standard LTCH PPS rate. MedPAC analysis of Medicare Provider Analysis and Review data from CMS. Source: LTCH PPS rate in 2017. For these LTCHs, the top 20 Financial incentives to serve Medicare beneficiaries across LTCHs diagnoses made up 77 percent of discharges (Table 12 11-4). The top two diagnoses, pulmonary edema and Another measure of access is whether providers have a respiratory failure and respiratory system diagnosis with financial incentive to expand the number of Medicare ventilator support, accounted for almost 43 percent of all beneficiaries they serve. In considering whether to treat discharges in the subset of LTCHs with a high share of a patient, a provider with excess capacity compares Medicare cases that met the criteria in 2017, compared the marginal revenue it will receive (i.e., the Medicare with less than 30 percent of discharges across all LTCHs. payment) with its marginal costs—that is, the costs that Further, more than 55 percent of these cases involved vary with volume. If Medicare payments are larger than diagnoses that were respiratory conditions or involved the marginal costs of treating an additional beneficiary, prolonged mechanical ventilation in the cohort of LTCHs a provider with sufficient capacity has a financial with a high share of cases meeting the criteria. Long-term care hospital services: Assessing payment adequacy and updating payments 292

323 FIGURE Cumulative change... FIGURE X-X Rates of unadjusted quality measures remain stable 11–1 50 Mortality within 30 days of discharge 40 In-LTCH mortality Readmissions 13% 13% 13% 30 12% 12% 12% 20 9% 9% 17% Percent of cases 9% 16% 16% 13% 12% 12% 8% 7% 10 6% 10% 10% 10% 9% 9% 9% 7% 7% 7% 0 2017 2015 2015 2015 2016 2017 2016 2016 2017 Cases meeting the criteria Cases not meeting the criteria All cases Note: LTCH (long-term care hospital). “Cases meeting the criteria” refers to Medicare discharges that meet the criteria specified in the Pathway for SGR Reform Act of 2013 to qualify for the standard LTCH prospective payment system rate. “Cases not meeting the criteria” refers to Medicare discharges that do not meet the criteria specified in the Pathway for SGR Reform Act of 2013. Results are preliminary and subject to change. MedPAC analysis of Medicare Provider Analysis and Review and enrollment data from CMS. Source: incentive to increase its volume of Medicare patients. In Quality of care: Meaningful measures contrast, if payments do not cover the marginal costs, the becoming available; trends for unadjusted indicators remain stable provider may have a disincentive to care for Medicare 13 beneficiaries. The Commission historically has assessed aggregate quality of care trends by examining three claims-calculated In 2017, the average LTCH marginal profit was about measures: unadjusted in-facility mortality rates, mortality 14 percent, down from almost 20 percent in 2016. within 30 days postdischarge, and direct ACH readmissions This decrease is not unexpected given the industry- from LTCHs. LTCHs began reporting a limited set of wide changes that are occurring as a result of the quality measures to CMS in fiscal year 2013 and recently congressionally mandated implementation of the dual started publicly reporting some risk-adjusted quality payment-rate structure. However, the change in marginal measures for LTCHs that are included in our discussion. profit was much smaller for LTCHs with a high share of Medicare cases meeting the criteria. For these LTCHs, Aggregate unadjusted quality measures marginal profit in 2017 was about 16 percent, 1 percentage For this report, we continued to analyze unadjusted point lower than in 2016. Both statistics suggest that readmission and mortality rates for LTCH cases from LTCHs with available beds continue to have a financial 2015 through 2017. We generally found stable rates of Note: Note and Source are in InDesign. incentive to increase their occupancy rates with Medicare readmissions to ACHs and stable mortality rates both beneficiaries who meet the criteria, representing a positive Source: in the facility and 30 days postdischarge (Figure 11-1). indicator of access. However, we caution that these measures are not risk March 2019 | Report to the Congress: Medicare Payment Policy 293

324 TABLE 11–5 Among cases meeting the criteria, rates of unadjusted measures varied across diagnoses related to respiratory illness or prolonged use of mechanical ventilation, 2017 Total mortality 30-day (in-LTCH post plus In-LTCH discharge 30-day mortality Readmission MS–LTC– mortality post rate Description DRG rate rate discharge) 29% 5% 14% Tracheostomy with ventilator support 96+ hrs or primary diagnosis 43% 4 except face, mouth and neck without major OR procedure 37 Other respiratory system OR procedures with MCC 11 21 16 166 177 7 27 Respiratory infections and inflammations with MCC 13 14 189 7 15 14 29 Pulmonary edema and respiratory failure Chronic obstructive pulmonary disease with MCC 10 190 15 25 6 207 12 22 14 36 Respiratory system diagnosis with ventilator support 96+ hours 208 Respiratory system diagnosis with ventilator support ≤96 hours 22 30 15 45 870 38 12 50 9 Septicemia with ventilator support 96+ hours with MCC 34 10 20 14 Total diagnoses related to respiratory illness or using prolonged mechanical ventilation Note: LTCH (long-term care hospital), OR (operating room), MCC (major complication or comorbidity). “Cases meeting the criteria” refers to Medicare discharges that meet the criteria specified in the Pathway for SGR Reform Act of 2013 to be paid the standard LTCH prospective payment system rate. MedPAC analysis of Medicare Provider Analysis and Review and enrollment data from CMS. Source: meeting the criteria. The rates of readmission and 30- adjusted, so patient characteristics were not taken into day postdischarge mortality were consistent from 2015 account when calculating rates, and trends may therefore to 2017, but the share of cases that died in the LTCH be muted or exaggerated by changes in patient mix over appears to have dropped. Six percent of cases not meeting time. In aggregate, in 2017, 9 percent of LTCH cases were the criteria died during the LTCH stay in 2017, down readmitted to an ACH directly from the LTCH, 12 percent from 8 percent in 2015. Given that these measures are not died in the LTCH, and another 12 percent died within 30 adjusted for patient risk factors, this decrease could be days of discharge from the LTCH (Figure 11-1, p. 293). attributable to improvements in quality or changes in case The rates have been stable since 2015. mix or admission patterns. We will monitor these cases as Not unexpectedly, given differences in patient severity, the dual payment-rate structure is fully phased in. the unadjusted rates for the three quality measures varied depending on whether the case met the criteria, but the For cases meeting the criteria, the unadjusted readmission rates were stable over time. In 2017, for cases meeting and mortality rates varied markedly by respiratory the criteria, 10 percent were readmitted to the ACH diagnosis group (Table 11-5). For example, among patients directly from the LTCH, 16 percent died in the LTCH, with a principal diagnosis of septicemia with prolonged and 13 percent died within 30 days of discharge from the ventilator support with major complication or comorbidity LTCH. Thus, combined, almost 40 percent of LTCH cases (MCC) (MS–LTC–DRG 870), 38 percent died in the meeting the criteria in 2017 were readmitted or died in the LTCH and another 12 percent died within 30 days of LTCH or within 30 days of discharge. discharge. By comparison, among patients with a primary diagnosis of chronic obstructive pulmonary disease with By comparison, cases not meeting the criteria had MCC (MS–LTC–DRG 190), 10 percent died in the LTCH lower rates of readmission and mortality than cases and another 15 percent died within 30 days of discharge. Long-term care hospital services: Assessing payment adequacy and updating payments 294

325 TABLE 11–6 Measures collected for the LTCH Quality Reporting Program for 2019 Publicly Collection Collection available start date Measure name instrument Catheter-associated urinary tract infection outcome measure 10/01/12 NHSN 12/2016 Central line–associated bloodstream infection outcome measure 12/2016 NHSN 10/01/12 12/2017 LTCH CARE 10/01/14 Percent of residents or patients who were assessed and appropriately given the seasonal influenza vaccine Influenza vaccination coverage among healthcare personnel 10/01/14 NHSN 12/2017 Facility-wide inpatient hospital-onset Clostridium difficile infection outcome measure NHSN 12/2017 01/01/15 LTCH CARE Application of percent of residents experiencing one or more falls with major injury (long stay) 09/2018 04/01/16 LTCH CARE Percent of LTCH patients with an admission and discharge functional assessment and a care 04/01/16 09/2018 plan that addresses function Discharge to community Claims 09/2018 Medicare spending per beneficiary Claims 09/2018 Potentially preventable 30-day post-discharge readmission Claims 04/01/16 LTCH CARE Change in mobility among LTCH patients requiring ventilator support Application of percent of LTCH patients with an admission and discharge functional 04/01/16 LTCH CARE assessment and a care plan that addresses function LTCH CARE 07/01/18 Drug regimen review conducted with follow-up for identified issues Changes in skin integrity PAC: Pressure ulcer/injury LTCH CARE 07/01/18 Compliance with spontaneous breathing trial by Day 2 of the LTCH stay LTCH CARE 07/01/18 Ventilator liberation rate LTCH CARE 07/01/18 LTCH (long-term care hospital), NHSN (National Healthcare Safety Network), LTCH CARE (LTCH Continuity Assessment Record and Evaluation), PAC (post-acute Note: care). Source: CMS LTCH quality reporting measure information and CMS LTCH Compare website. Overall, 34 percent of patients meeting the criteria with a Network (NHSN)), and Medicare claims data. CMS has diagnosis related to respiratory illness or using prolonged published two years of outcomes data for four outcome mechanical ventilation died within the LTCH or within 30 measures, including rates of pressure ulcers, catheter- days of discharge. associated urinary tract infection (CAUTI), central line– associated blood stream infection (CLABSI), and 30-day Adjusted measures for quality reporting all-cause unplanned readmissions. For several measures, CMS compares each facility’s risk-adjusted rate with the Medicare’s LTCH Quality Reporting Program (QRP) national rate. for fiscal year 2019 includes 16 measures (Table 11-6). CMS currently reports some of these measures on its The rate of pressure ulcers reported by LTCHs for the data LTCH Compare website, which is updated quarterly. collection period of October 1, 2016, through September The data elements needed to calculate the LTCH quality 30, 2017, was relatively low at 1.3 percent (Table 11-7, measures are collec