untitled

Transcript

1 REVIEW l-Cause Mortality Association of Al With Overweight and Obesity Using Standard Body Mass Index Categories A Systematic Review and Meta-ana lysis Katherine M. Flegal, PhD Estimates of the relative mortality risks associated with normal weight, Importance Brian K. Kit, MD overweight, and obesity may help to inform decision making in the clinical setting. Heather Orpana, PhD Objective To perform a systematic review of reported hazard ratios (HRs) of all- cause mortality for overweight and obesity relative to normal weight in the general Barry I. Graubard, PhD population. HE TOPIC OF THE MORTALITY PubMed and EMBASE electronic databases were searched through Data Sources differences between weight September 30, 2012, without language restrictions. categories has sometimes ArticlesthatreportedHRsforall-causemortalityusingstandardbody StudySelection T been described as controver- massindex(BMI)categoriesfromprospectivestudiesofgeneralpopulationsofadultswere 1 The appearance of controversy sial. selected by consensus among multiple reviewers. Studies were excluded that used non- may arise in part because studies of standard categories or that were limited to adolescents or to those with specific medical body mass index (BMI; calculated as conditions or to those undergoing specific procedures. PubMed searches yielded 7034 articles, of which 141 (2.0%) were eligible. An EMBASE search yielded 2 additional ar- weight in kilograms divided by ticles. After eliminating overlap, 97 studies were retained for analysis, providing a com- height in meters squared) and mor- bined sample size of more than 2.88 million individuals and more than 270 000 deaths. tality have used a wide variety of BMI Data were extracted by 1 reviewer and then reviewed by 3 inde- Data Extraction categories and varying reference cat- pendent reviewers. We selected the most complex model available for the full sample egories, which can make findings and used a variety of sensitivity analyses to address issues of possible overadjustment appear more variable than when (adjusted for factors in causal pathway) or underadjustment (not adjusted for at least standard categories are used and also age, sex, and smoking). can make it difficult to compare and 2 Results Random-effects summary all-cause mortality HRs for overweight (BMI of in 1997 synthesize studies. A report 30), grade 1 obesity (BMI of 30- 35), and grades 2 and   30), obesity (BMI of  25- from the World Health Organization  35) were calculated relative to normal weight (BMI of 18.5-  3 obesity (BMI of 25). Consultation on Obesity defined The summary HRs were 0.94 (95% CI, 0.91-0.96) for overweight, 1.18 (95% CI, 1.12- BMI-based categories of under- 1.25) for obesity (all grades combined), 0.95 (95% CI, 0.88-1.01) for grade 1 obesity, weight, normal weight, preobesity, and 1.29 (95% CI, 1.18-1.41) for grades 2 and 3 obesity. These findings persisted and obesity. The same cutoff BMI when limited to studies with measured weight and height that were considered to be values were adopted by the National adequately adjusted. The HRs tended to be higher when weight and height were self- reported rather than measured. Heart, Lung, and Blood Institute in 3 1998. Relative to normal weight, both obesity (all grades) Conclusions and Relevance In this study, we used the National and grades 2 and 3 obesity were associated with significantly higher all-cause mor- Heart, Lung, and Blood Institute’s tality. Grade 1 obesity overall was not associated with higher mortality, and over- weight was associated with significantly lower all-cause mortality. The use of pre- defined standard BMI groupings can facilitate between-study comparisons. For editorial comment see p 87. www.jama.com JAMA. 2013;309(1):71-82 CME available online at National Center for Health Author Affiliations: National Cancer Institute, Bethesda, Maryland (Dr www.jamaarchivescm e.com Statistics, Centers for Disease Control and Preven- Graubard). and questions on p 91. tion, Hyattsville, Maryland (Drs Flegal and Kit); Corresponding Author: Katherine M. Flegal, PhD, Na- School of Psychology, University of Ottawa, tional Center for Health Statistics, Centers for Dis- Author Video Interview available at Ottawa, Ontario, Canada (Dr Orpana); and Di- ease Control and Prevention, 3311 Toledo Rd, Room www.jama.com. vision of Cancer Epidemiology and Genetics, 4336, Hyattsville, MD 20782 ([email protected]). ©2013 American Medical Association. All rights reserved. JAMA, January 2, 2013—Vol 309, No. 1 71 Downloaded From: http://jama.jamanetwork.com/ on 01/05/2013

2 ASSOCIATION OF ALL-CAUSE MORTALITY WITH OVERWEIGHT AND OBESITY of self-reported and measured weight reviewed for inclusion by 1 reviewer terminology with categories of  18.5), normal underweight (BMI of and height according to the prepon- (K.M.F.). An independent review of all 25), over-  weight (BMI of 18.5- derant type. articles was conducted by a second set weight (BMI of 25-  30), and obesity Abstracted items included sample of reviewers (B.K.K., H.O., and B.I.G.). (BMI of  30). Grade 1 obesity was size, number of deaths, age at base- The articles were reviewed to identify defined as a BMI of 30 to less than line, length of follow-up, HRs and 95% those that used standard BMI catego- 35; grade 2 obesity, a BMI of 35 to confidence intervals, sex, age, type of ries in prospective, observational co- less than 40; and grade 3 obesity, a weight and height data (measured or hort studies of all-cause mortality BMI of 40 or greater. These standard self-reported), country or region, source among adults with BMI measured or re- categories have been increasingly of study sample, adjustment factors, ex- ported at baseline. Studies that ad- used in published studies of BMI and clusion and inclusion criteria, and sen- dressed these relationships only in ado- mortality, but the literature reporting sitivity analyses. Authors of screened ar- lescents, only in institutional settings, these results has not been systemati- ticles were queried for additional or only among those with specific medi- cally reviewed. information when necessary. In stud- cal conditions or undergoing specific The purpose of this study was to ies that only presented results strati- medical procedures were excluded. We compile and summarize published fied by smoking or health condition, we included multiple articles from a given analyses of BMI and all-cause mortal- selected results for nonsmokers or never data set only when there was little over- ity that provide hazard ratios (HRs) for smokers or for those without the health lap between articles by sex, age group, standard BMI categories. We followed condition. We selected the most com- or some other factor. the guidelines in the Meta-analysis of plex model available for the full sample In some cases, authors used stan- Observational Studies in Epidemiol- and used a variety of sensitivity analy- dard BMI categories for overweight and 4 for report- ogy (MOOSE) statement ses to address issues of possible over- obesity but had used a slightly broader ing of systematic reviews. adjustment or underadjustment. reference BMI category of less than 25 We categorized HRs into 2 age or a slightly narrower reference BMI cat- METHODS groupings either as limited solely to egory of 20 to less than 25 rather than people aged 65 years or older or as a the standard normal BMI category of Articles were identified by searches mixed-age category (eg, aged 25-64 18.5 to less than 25. We included these of PubMed and EMBASE through years or 40-80 years). We classified ar- articles but have noted the cases in September 30, 2012. Details of ticles as adequately adjusted, possibly which the reference BMI category was search strategies appear in eTable 1 at overadjusted, or possibly underad- less than 25 or 20 to less than 25. We http://www.jama.com. No language re- justed. We categorized HRs by adjust- classified studies that included a mix strictions were applied. All articles were Summary Random-Effects Hazard Ratios (HRs) of All-Cause Mortality for Overweight and Obesity Relative to Normal Weight Table 1. Height and Weight Self-reported or Measured Self-reported Measured Height and Weight No. of Summary HR Summary HR No. of Summary HR No. of 2 2 2 I HRs (95% CI) (95% CI) I HRs ,% ,% ,% HRs (95% CI) I 30 BMI of 25-  a a a 0.96 (0.92-1.00) 0.93 (0.89-0.95) 75.8 90.4 51 85.0 89 All ages 140 0.94 (0.91-0.96) a a a 86.8 67 0.93 (0.89-0.96) Mixed ages 79.6 40 0.98 (0.93-1.03) 107 0.95 (0.92-0.98) 90.7 a a  Age 51.2 22 0.90 (0.84-0.95) 31.2 11 0.90 (0.84-0.96) 0.90 (0.86-0.94) 71.0 65 y only 33 30 BMI of  a a a 1.13 (1.06-1.19) 28 73.4 86.7 1.29 (1.18-1.41) 56 89.7 84 1.18 (1.12-1.25) All ages a a a 41 1.16 (1.10-1.24) Mixed ages 74.6 22 63 1.23 (1.16-1.31) 86.1 1.36 (1.25-1.48) 87.2 a a a 65 y only 61.5 15 0.98 (0.86-1.12)  61.1 6 1.09 (0.96-1.23) 1.03 (0.94-1.12) 67.0 Age 21 35  BMI of 30- a a a 30 0.94 (0.86-1.03) 80.5 90.1 23 0.95 (0.85-1.06) 86.8 All ages 0.95 (0.88-1.01) 53 a a a 87.7 Mixed ages 0.95 (0.86-1.06) 42 83.2 18 0.97 (0.87-1.09) 0.96 (0.89-1.04) 90.0 24 a a  Age 76.3 6 0.89 (0.71-1.11) 56.2 5 0.83 (0.58-1.20) 0.87 (0.72-1.05) 85.7 65 y only 11 35  BMI of a a a 30 1.25 (1.13-1.39) 65.4 88.3 23 1.34 (1.16-1.55) 81.7 All ages 1.29 (1.18-1.41) 53 a a a 82.8 Mixed ages 1.28 (1.14-1.44) 42 68.9 18 1.35 (1.16-1.58) 1.32 (1.19-1.45) 89.0 24 a a 25.1  70.6 6 1.10 (0.89-1.34) 11 5 1.29 (0.77-2.17) 1.20 (0.94-1.52) 85.2 65 only Age Abbreviation: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared). a Indicates significant heterogeneity (  .05). P 72 JAMA, January 2, 2013—Vol 309, No. 1 ©2013 American Medical Association. All rights reserved. Downloaded From: http://jama.jamanetwork.com/ on 01/05/2013

3 ASSOCIATION OF ALL-CAUSE MORTALITY WITH OVERWEIGHT AND OBESITY and 75% considered low, moderate, and proximate HRs relative to normal ment level, by whether the data were 6 high, respectively. weight from several recent large stud- measured or self-reported, by whether We also used a se- 9-14 ies that had used finer BMI group- the analysis was performed separately quential approach similar to that de- 7 to assess ings and thus did not meet our inclu- for men and women or for both sexes scribed by Patsopoulos et al consistency of findings when hetero- sion criteria. To do this, we averaged combined, and by region (North geneity was reduced. All analyses were HRs from the finer BMI groupings over America, Europe, and other). 5 to performed with SAS version 9.3 (SAS groups corresponding to the standard We used a random-effects model summarize the results overall and Institute Inc). BMI categories, weighting the HRs by within subgroups and based statisti- For sensitivity analyses, we exam- the number of deaths. cally significant heterogeneity on a P ined the effects on HRs of incorporat- RESULTS value of less than .05. We calculated the ing results from a recent large pooled 2 8 As shown in the eFigure at http://www to describe the degree of For compara- quantity I study for overweight. .jama.com, the primary search strat- tive purposes, we also constructed ap- heterogeneity with values of 25%, 50%, Hazard Ratios for All-Cause Mortality Relative to Normal Weight in Studies That Used Measured Data for Participants With a Body Figure 1. Mass Index of 25 to Less Than 30 Source Source 30 33 2011 (men) 2007 ( ≥ 70 y) Lisko et al, Flegal et al, 34 60 Takata et al, Crespo et al, 2002 2007 35 35 2009 (women) 2009 (men) Stessman et al, Stessman et al, 36 23 Uretsky et al, Lang et al, 2008 (men) 2010 38 20 2005 (men) 65 y) 2004 (women < Hu et al, Visscher et al, 37 61 Keller and Østbye, Gu et al, 2005 2006 39 22 2002 (men 46-59 y) 2007 Lahmann et al, Arndt et al, 63 40 Tsai et al, Fontaine et al, 70 y SALSA) ≥ 2012 ( 2006 62 41 2011 2007 Faeh et al, Sui et al, 30 25 2007 (60-69 y) 2010 Flegal et al, McAuley et al, 43 48 Seccareccia et al, Walter et al, 1998 (women 20-44 y) 2009 (nondisabled) 42 40 60 y SAHS) < 2012 (18- Fontaine et al, Lubin et al, 2003 64 45 Iribarren et al, 2010 Atlantis et al, 2005 44 58 Lawlor et al, Hanson et al, 1995 2006 (Renfrew/Paisley women) 29 50 Ferrie et al, 2009 (men) 2007 (women) Simpson et al, 56 33 Wändell et al, 2011 (women) Lisko et al, 2009 (women) 65 46 Cabrera et al, 2005 Tice et al, 2006 24 23 Luchsinger et al, 2008 2008 (women) Lang et al, 51 31 2011 Lakoski et al, Hotchkiss and Leyland, 2011 (women) 47 68 Petursson et al, 2006 (whites) 2011 (women) McTigue et al, 69 48 Miller et al, Walter et al, 2009 (disabled) 2002 67 30 2002 Jonsson et al, Flegal et al, 2007 (25-59 y) 50 66 Ferrie et al, Janssen and Bacon, 2009 (women) 2008 20 22 Lahmann et al, Hu et al, 2002 (men 60-75 y) 2005 (women) 49 72 2000 2002 Visscher et al, Farrell et al, 43 71 Gale et al, Seccareccia et al, 1998 (men 20-44 y) 2007 70 29 Batty et al, 2007 (men) 2006 Simpson et al, 58 47 2011 (men) 2006 (collaborative men) Lawlor et al, Petursson et al, 51 26 2010 Nechuta et al, Lakoski et al, 2011 (men) 22 43 2002 (women 46-59 y) 1998 (women 45-69 y) Seccareccia et al, Lahmann et al, 28 22 Lahmann et al, Pednekar et al, 2008 (men) 2002 (women 60-75 y) 40 74 Katzmarzyk et al, 2012 (60-70 y SALSA) Fontaine et al, 2012 73 52 2002 2006 Baldinger et al, Fang et al, 54 59 Greenberg et al, Osler et al, 2007 2001 (women) 53 40 Blain et al, Fontaine et al, 2010 2012 ( 70 y SAHS) ≥ 28 40 2008 (women) 2012 (60-70 y SAHS) Fontaine et al, Pednekar et al, 75 43 1998 (men 45-69 y) 2010 Ioachimescu et al, Seccareccia et al, 56 67 2009 (men) 2006 (blacks) Wändell et al, McTigue et al, 76 55 Janssen, 2011 Heir et al, 2007 77 38 Visscher et al, Katzmarzyk et al, ≥ 2001 2004 (women 65 y) 58 78 2006 (Renfrew/Paisley men) 2009 Lawlor et al, Cesari et al, 38 38 Visscher et al, Visscher et al, ≥ 65 y) 65 y) 2004 (men < 2004 (men 79 57 2011 Bevilacqua and Gimeno, Arnlöv et al, 2010 59 80 2001 (men) 2008 Osler et al, Tsai et al, 81 Locher et al, 2007 0.1 1.0 5 1.0 5 0.1 Hazard Ratio (95% CI) Hazard Ratio (95% CI) Area Latino Data markers indicate hazard ratios and error bars indicate 95% confidence intervals. SAHS indicates San Antonio Heart Study; and SALSA, Sacramento Study on Aging. ©2013 American Medical Association. All rights reserved. JAMA, January 2, 2013—Vol 309, No. 1 73 Downloaded From: http://jama.jamanetwork.com/ on 01/05/2013

4 ASSOCIATION OF ALL-CAUSE MORTALITY WITH OVERWEIGHT AND OBESITY tralia (n=7), China or Taiwan (n=4), lacking sufficient information, 97 ar- egy for PubMed yielded 4142 articles, Japan (n=2), Brazil (n=2), Israel (n=2), ticles remained for analysis; all of which of which 128 met our criteria. A India (n=1), and Mexico (n=1). The had been identified through system- second PubMed search yielded 2892 tabulated studies included more than atic search procedures. The selected additional articles, of which 13 met 2.88 million participants and more than studies are shown in eTable 2 with ad- our criteria. A search of EMBASE 270 000 deaths. ditional information in eTable 3 re- yielded 2 additional eligible articles. In Not all studies reported the specific garding exclusions and adjustment fac- total, 143 eligible studies were identi- categories of interest. There were 93 tors. Regions of origin of participants fied. studies for the BMI category of 25 to less included the United States or Canada After exclusion of 41 articles with than 30 (overweight), 61 studies for the (n=41 studies), Europe (n=37), Aus- overlapping data sets and of 5 articles BMI category of 30 or greater (obe- sity), and 32 studies for the BMI cat- egories of 30 to less than 35 (grade 1 Hazard Ratios for All-Cause Mortality Relative to Normal Weight in Studies That Figure 2. Used Self-reported Data for Participants With a Body Mass Index of 25 to Less Than 30 obesity) and 35 and greater (grades 2 and 3 obesity). Source 82 We considered the results ad- 2001 Taylor and Ostbye, 135 2012 (white men) Cohen et al, equately adjusted if they were ad- 83 Monteverde et al, 2010 84 justed for age, sex, and smoking and not Freedman et al, 2006 (men ≥ 65 y) 85 2011 (disabled) Majer et al, adjusted for factors in the causal path- 15 Al Snih et al, 2007 way between obesity and mortality, or 135 2012 (black women) Cohen et al, 135 Cohen et al, 2012 (black men) if they had reported or demonstrated 86 2012 Jerant and Franks, that adjustments or exclusions to avoid 85 2011 (nondisabled) Majer et al, 87 bias had shown little effect on their find- 2000 Strawbridge et al, 27 Orpana et al, 2010 ings. A number of studies (for ex- 32 Mehta and Chang, 2009 (women) 15-29 89 ) reported qualitatively that ample Niedhammer et al, 2011 88 2008 Ford et al, such adjustments had little or no ef- 19 Hjartåker et al, 2005 (postmenopausal) fect without showing quantitative de- 90 2000 (women) Haapanen-Niemi et al, 90 2000 (men) Haapanen-Niemi et al, tails. 18 Flicker et al, 2010 30-32 ) dem- Other studies (for example 91 < 12 y) Boggs et al, 2011 (education 92 onstrated little effect through a series 2010 Lantz et al, 84 Freedman et al, 2006 (men < 65 y) of sensitivity analyses. We considered 93 Ma et al, 2011 21 the available full sample results from 2004 Krueger et al, 32 Mehta and Chang, 2009 (men) such studies to also be adequately ad- 94 Zunzunegui et al, 2012 justed. Otherwise, we considered stud- 95 2012 (men) Nagai et al, 96 2007 (men) Fujino et al, ies as possibly overadjusted if they ad- 135 Cohen et al, 2012 (white women) justed for factors such as hypertension 98 2008 Yates et al, 97 Corrada et al, 2006 that are considered to be in the causal 95 2012 (women) Nagai et al, pathway between obesity and mortal- 99 2010 Iversen et al, 100 ity or as possibly underadjusted if they Leitzmann et al, 2011 101 Ringbäck Weitoft et al, 2008 (men) did not adjust for age, sex, and smok- 16 2010 (women) Bellocco et al, ing. We classified 53 studies as ad- 102 Bessonova et al, 2011 96 2007 (women) Fujino et al, equately adjusted, 34 studies as possi- 84 65 y) 2006 (women ≥ Freedman et al, bly overadjusted, and 10 studies as 101 Ringbäck Weitoft et al, 2008 (women) 16 2010 (men) Bellocco et al, possibly underadjusted. 103 Gray et al, 2010 ABLE 1 Summary HRs are shown in T 104 Stevens et al, 2000 (women) 105 overall, by age group, and by whether 2008 van Dam et al, 104 Stevens et al, 2000 (men) data were measured or self-reported. 106 Gelber et al, 2007 17 The summary HRs were 0.94 (95% CI, 2011 (men) Carlsson et al, 19 Hjartåker et al, 2005 (premenopausal) 0.91-0.96) for overweight, 1.18 (95% 17 Carlsson et al, 2011 (women) CI, 1.12-1.25) for obesity (all grades), 84 2006 (women 65 y) < Freedman et al, 91 Boggs et al, ≥ 2011 (education 12 y) 0.95 (95% CI, 0.88-1.01) for grade 1 obesity, and 1.29 (95% CI, 1.18-1.41) 5 0.1 1.0 for grades 2 and 3 obesity. Plots of HRs Hazard Ratio (95% CI) for these categories are shown in 33-110 IGURES Additional details are . 8 - 1 Data markers indicate hazard ratios and error bars indicate 95% confidence intervals. F 74 JAMA, January 2, 2013—Vol 309, No. 1 ©2013 American Medical Association. All rights reserved. Downloaded From: http://jama.jamanetwork.com/ on 01/05/2013

5 ASSOCIATION OF ALL-CAUSE MORTALITY WITH OVERWEIGHT AND OBESITY increase of the summary HR from For overweight, excluding these shown in eTables 4-7, which show sum- 1.21 to 1.24. Corresponding values studies led to a uniformly lower HR mary HRs by age, sex, region, and mea- were from 0.97 to 1.05 (neither sig- of 0.89 for both age groups and for surement type. nificantly different from 1) for grade both measured and self-reported Results for studies that we consid- 1 obesity and from 1.34 to 1.39 for data. For obesity, the effects of ered adequately adjusted are shown in grades 2 and 3 obesity. Thus, hetero- excluding these studies were more ABLE 2 . This more select group T geneity appeared to have had little variable and led to an overall showed the same general pattern of overweight associated with reduced mortality, grade 1 obesity not signifi- Figure 3. Hazard Ratios for All-Cause Mortality Relative to Normal Weight in Studies That cantly associated with increased mor- Used Measured Data for Participants With a Body Mass Index of 30 or Greater tality, and the higher grades of obesity significantly associated with in- Source 35 2009 (women) Stessman et al, creased mortality. The summary HRs 36 2010 Uretsky et al, were 0.94 (95% CI, 0.90-0.97) for over- 33 2011 (men) Lisko et al, 45 weight, 1.21 (95% CI, 1.12-1.31) for Iribarren et al, 2005 24 2008 Luchsinger et al, obesity (all grades), 0.97 (95% CI, 0.90- 65 Cabrera et al, 2005 33 1.04) for grade 1 obesity, and 1.34 (95% 2011 (women) Lisko et al, 55 Janssen, 2007 CI, 1.21-1.47) for grades 2 and 3 obe- 38 2004 (women ≥ 65 y) Visscher et al, sity. For overweight, the results from 57 2011 Bevilacqua and Gimeno, 37 2005 Keller and Østbye, possibly overadjusted studies and from 31 Hotchkiss and Leyland, 2011 adequately adjusted studies were simi- 28 Pednekar et al, 2008 (men) 107 2007 Wannamethee et al, lar (eTable 8). However, for obesity, the 29 2007 (women) Simpson et al, possibly overadjusted studies tended to 42 Lubin et al, 2003 78 have lower HRs than the adequately ad- Cesari et al, 2009 38 Visscher et al, 2004 (women <65 y) justed studies. 29 2007 (men) Simpson et al, Between-study heterogeneity was sta- 108 Suadicani et al, 2009 35 2009 (men) Stessman et al, tistically significant in most catego- 64 2010 Atlantis et al, 6 this test ries. According to Higgins et al, 28 Pednekar et al, 2008 (women) 75 Ioachimescu et al, 2010 may have “excessive power when there 60 2002 Crespo et al, are many studies, especially when those 109 Seidell et al, 1996 (women) 58 studies are large.” Heterogeneity (as in- Lawlor et al, 2006 (Renfrew/Paisley men) 69 2 2002 Miller et al, ) was less for dicated by the value of I 61 2006 Gu et al, 59 studies with measured height and 2001 (men) Osler et al, 71 Gale et al, 2007 weight and was lower for studies lim- 20 2005 (women) Hu et al, ited to those older than 65 years. The 20 2005 (men) Hu et al, 2 53 2010 Blain et al, was reduced by limiting find- value of I 66 Janssen and Bacon, 2008 ings to adequately adjusted studies with 49 2002 Farrell et al, 50 measured data. Ferrie et al, 2009 (men) 26 2010 Nechuta et al, Higher levels of heterogeneity, how- 63 Tsai et al, 2006 67 ever, do not necessarily lead to dissimi- Jonsson et al, 2002 58 2006 (Renfrew/Paisley women) Lawlor et al, lar results that would affect the con- 58 Lawlor et al, 2006 (collaborative) clusions. For example, the summary HR 70 2006 Batty et al, 56 2009 (men) Wändell et al, for overweight for older ages (  65 62 Faeh et al, 2011 years) was identical (0.90) for mea- 38 Visscher et al, 2004 (men ≥ 65 y) 2 72 =31.2%) and I sured height and weight ( 2000 Visscher et al, 109 1996 (men) Seidell et al, for self-reported height and weight 59 Osler et al, 2001 (women) 2 50 =71.0%). For adequately adjusted I ( 2009 (women) Ferrie et al, 38 2004 (men < Visscher et al, 65 y) studies, we sequentially excluded HRs 56 Wändell et al, 2009 (women) within age and measurement catego- 79 Arnlöv et al, 2010 2 73 Baldinger et al, 2006 value to ries as needed to reduce the I 80 2008 Tsai et al, below 25%. Within the 4 age- 76 2011 Heir et al, measurement groups, this required ex- 5 1.0 0.1 clusion of 9% to 22% of studies for mea- Hazard Ratio (95% CI) sured data and 14% to 39% of studies Data markers indicate hazard ratios and error bars indicate 95% confidence intervals. for self-reported data. ©2013 American Medical Association. All rights reserved. JAMA, January 2, 2013—Vol 309, No. 1 75 Downloaded From: http://jama.jamanetwork.com/ on 01/05/2013

6 ASSOCIATION OF ALL-CAUSE MORTALITY WITH OVERWEIGHT AND OBESITY cluded. However, we were able to con- Hazard Ratios for All-Cause Mortality Relative to Normal Weight in Studies That Figure 4. struct approximate HRs from some re- Used Self-reported Data for Participants With a Body Mass Index of 30 or Greater cent large studies that had used Source nonstandard BMI categories (eTable 9). 92 2010 Lantz et al, 85 This approach does not allow for con- 2011 (disabled) Majer et al, 88 2008 Ford et al, struction of appropriate standard er- 18 Flicker et al, 2010 rors or confidence intervals. The ap- 99 2010 Iversen et al, 85 Majer et al, 2011 (nondisabled) proximate HRs were consistent with our 90 2000 (men) Haapanen-Niemi et al, findings from our analyses of indi- 96 Fujino et al, 2007 (men) 89 vidual studies, showing similar minor 2011 Niedhammer et al, 97 2006 Corrada et al, variation. 95 Nagai et al, 2012 (men) 95 2012 (women) Nagai et al, 19 COMMENT 2005 (postmenopausal) Hjartåker et al, 102 2011 Bessonova et al, This study presents comprehensive 90 2000 (women) Haapanen-Niemi et al, 16 2010 (women) Bellocco et al, estimates (derived from a systematic 101 Ringbäck Weitoft et al, 2008 (women) review) of the association of all-cause 98 Yates et al, 2008 96 mortality in adults with current stan- Fujino et al, 2007 (women) 101 Ringbäck Weitoft et al, 2008 (men) dard BMI categories used in the 104 Stevens et al, 2000 (women) 104 United States and internationally. Stevens et al, 2000 (men) 16 2010 (men) Bellocco et al, Estimates of the relative mortality 17 2011 (men) Carlsson et al, risks associated with normal weight, 17 Carlsson et al, 2011 (women) 105 2008 van Dam et al, overweight, and obesity may help to 103 2010 Gray et al, inform decision making in the clini- 19 2005 (premenopausal) Hjartåker et al, cal setting. 1.0 0.1 5 The most recent data from the United Hazard Ratio (95% CI) States show that almost 40% of adult Data markers indicate hazard ratios and error bars indicate 95% confidence intervals. men and almost 30% of adult women fall into the overweight category with 111 to 29.9 relative to those with a BMI of effect on the conclusions of the meta- Compa- a BMI of 25 to less than 30. 20 to less than 25 (Amy Berrington de analysis. rable figures for Canada are 44% of men 112 Gonzalez, DPhil, written communica- The excluded studies varied across and for England and 30% of women 113 tion, June 16, 2011). outcome categories; inspection of the are 42% of men and 32% of women. Our analysis included published excluded studies did not suggest spe- According to the results presented studies using 6 of the same cohorts, cific reasons why they had contrib- herein, overweight (defined as a BMI representing about 60% of the original uted to heterogeneity. Taken to- of 25- 30) is associated with signifi-  8 Berrington de Gonzalez et al sample. gether, the findings suggest that cantly lower mortality overall relative Excluding those studies from our contributors to heterogeneity across all to the normal weight category with an analysis and substituting the above studies include adjustment levels, type overall summary HR of 0.94. For over- results from Berrington de Gonzalez of measurement data, and age group. weight, 75% of HRs with measured et al did not change the summary HR Some degree of heterogeneity may also weight and height and 67% of HRs with for overweight. result from the variation in BMI levels self-reported weight and height were We also repeated the analyses after within the broad BMI categories used, below 1. These results are broadly con- excluding the studies that had used as well as from variations in the type sistent with 2 previous meta-analy- 114,115 slightly different reference categories. of cohorts studied. that used standard categories. ses Excluding studies with a reference BMI In a pooled analysis of 26 observa- 114 Sensitivity Analyses category of less than 25 had no effect found tional studies, McGee et al For the overweight category only, we on the HRs for overweight and de- summary relative risks of all-cause mor- also repeated analyses including the re- creased the HR for obesity by 0.02. Ex- tality for overweight of 0.97 (95% CI, sults from a study that pooled data from cluding studies with a reference BMI 0.92-1.01) for men and 0.97 (95% CI, 19 cohorts. After excluding ever smok- category of 20 to less than 25 in- 0.93-0.99) for women relative to nor- ers and those with a history of cancer creased the HR for overweight by 0.005 mal weight. or heart disease, Berrington de Gonza- and had no effect on the HR for obesity. Recent estimates for the prevalence 8 found a HR of 1.11 (95% CI, lez et al Beyond these slight differences in the  of obesity (defined as a BMI of 30) 1.07-1.16) for men and 1.13 (95% CI, reference category, studies that used among adults are 36% in the United 112 111 1.09-1.16) for women with a BMI of 25 nonstandard BMI categories were ex- 24% in Canada, and 26% in States, 76 JAMA, January 2, 2013—Vol 309, No. 1 ©2013 American Medical Association. All rights reserved. Downloaded From: http://jama.jamanetwork.com/ on 01/05/2013

7 ASSOCIATION OF ALL-CAUSE MORTALITY WITH OVERWEIGHT AND OBESITY 113 England. Obesity was associated with Figure 5. Hazard Ratios for All-Cause Mortality Relative to Normal Weight in Studies That significantly higher all-cause mortal- Used Measured Data for Participants With a Body Mass Index of 30 to Less Than 35 ity relative to the normal weight BMI category with an overall summary HR Source 43 of 1.18. Corresponding estimates for Seccareccia et al, 1998 (men 20-44 y) 40 114 70 y SALSA) ≥ 2012 ( Fontaine et al, were 1.20 obesity from McGee et al 25 2010 McAuley et al, 43 (95% CI, 1.12-1.29) for men and 1.28 1998 (women 45-69 y) Seccareccia et al, 46 2006 Tice et al, (95% CI, 1.18-1.37) for women. In the 39 Arndt et al, 2007 United States and Canada, more than 48 2009 (disabled) Walter et al, 41 Sui et al, 2007 half of those who are obese fall into the 44 1995 Hanson et al, 35). We  grade 1 category (BMI of 30- 40 2012 (60-70 y SALSA) Fontaine et al, 81 did not find significant excess mortal- 2007 Locher et al, 40 ≥ 70 y SAHS) 2012 ( Fontaine et al, ity associated with grade 1 obesity, sug- 40 Fontaine et al, 2012 (60-70 y SAHS) 43 gesting that the main contribution to Seccareccia et al, 1998 (women 20-44 y) 47 2011 (women) Petursson et al, excess mortality in obesity comes from 43 1998 (men 45-69 y) Seccareccia et al, higher levels of BMI. 23 2008 (men) Lang et al, 40 60 y SAHS) Fontaine et al, 2012 (18- < Our findings are consistent with ob- 30 Flegal et al, 2007 ( ≥ 70 y) servations of lower mortality among 48 2009 (nondisabled) Walter et al, 47 Petursson et al, 2011 (men) overweight and moderately obese pa- 68 2006 (whites) McTigue et al, 116-119 Possible explanations have tients. 30 Flegal et al, 2007 (60-69 y) 54 included earlier presentation of heavier Greenberg et al, 2007 110 120 Sonestedt et al, 2011 greater likelihood of receiv- patients, 30 2007 (25-59 y) Flegal et al, 121-123 car- ing optimal medical treatment, 23 Lang et al, 2008 (women) 77 2001 Katzmarzyk et al, dioprotective metabolic effects of in- 74 Katzmarzyk et al, 2012 124,125 and benefits of creased body fat, 68 2006 (blacks) McTigue et al, 118 higher metabolic reserves. 5 1.0 0.1 The results presented herein pro- Hazard Ratio (95% CI) vide little support for the sugges- Data markers indicate hazard ratios and error bars indicate 95% confidence intervals. SAHS indicates San An- 126 that smoking and preexisting tion tonio Heart Study; and SALSA, Sacramento Area Latino Study on Aging. illness are important causes of bias. Most studies that addressed the issue found that adjustments or exclusions Figure 6. Hazard Ratios for All-Cause Mortality Relative to Normal Weight in Studies That for these factors had little or no Used Self-reported Data for Participants With a Body Mass Index of 30 to Less Than 35 effect. However, overadjustment for factors in the causal pathway appears Source 82 2001 Taylor and Ostbye, to decrease HRs for obesity but not 83 Monteverde et al, 2010 for overweight. 135 2012 (black women) Cohen et al, 135 An important source of bias ap- Cohen et al, 2012 (black men) 135 2012 (white men) Cohen et al, pears to be the errors in self-reported 94 Zunzunegui et al, 2012 87 weight and height data. Such errors 2000 Strawbridge et al, 15 Al Snih et al, 2007 have been shown to vary by age, sex, 32 Mehta and Chang, 2009 (women) race, measured values, and data collec- 91 Boggs et al, 2011 (education < 12 y) 127,128 86 2012 Jerant and Franks, The systematic er- tion method. 27 2010 Orpana et al, ror of self-reported data rather than 84 Freedman et al, 2006 (women < 65 y) 32 measured data can result in substan- 2009 (men) Mehta and Chang, 21 Krueger et al, 2004 tial misclassification of individuals into 135 2012 (white women) Cohen et al, 129 84 create er- incorrect BMI categories, Freedman et al, ≥ 2006 (men 65 y) 100 130 Leitzmann et al, 2011 rors that are difficult to correct, and 93 2011 Ma et al, lead to upward bias in the esti- 84 2006 (men < 65 y) Freedman et al, 131 84 ≥ 65 y) 2006 (women Freedman et al, We found a generally lower mates. 106 Gelber et al, 2007 summary HR and less heterogeneity in 91 12 y) ≥ 2011 (education Boggs et al, studies using measured data than in 0.1 1.0 5 studies using self-reported data. The dif- Hazard Ratio (95% CI) ferences were more pronounced in Data markers indicate hazard ratios and error bars indicate 95% confidence intervals. analyses stratified by sex than in analy- ©2013 American Medical Association. All rights reserved. JAMA, January 2, 2013—Vol 309, No. 1 77 Downloaded From: http://jama.jamanetwork.com/ on 01/05/2013

8 ASSOCIATION OF ALL-CAUSE MORTALITY WITH OVERWEIGHT AND OBESITY ses that combined both men and Figure 7. Hazard Ratios for All-Cause Mortality Relative to Normal Weight in Studies That women. Because the errors in self- Used Measured Data for Participants With a Body Mass Index of 35 or Greater reported data tend to differ by sex, there may be an offsetting effect when analy- Source 46 ses combine men and women. 2006 Tice et al, 43 1998 (men 20-44 y) Seccareccia et al, Publication bias can potentially affect 40 ≥ 2012 ( 70 y SALSA) Fontaine et al, 81 systematic reviews. Studies that find Locher et al, 2007 43 1998 (women 45-69 y) Seccareccia et al, little or no association of overweight or 40 Fontaine et al, 2012 ( ≥ 70 y SAHS) obesity with mortality risk sometimes 39 2007 Arndt et al, 25 McAuley et al, 2010 only mention these results in passing 40 2012 (18- < 60 y SAHS) Fontaine et al, without providing details. For ex- 44 1995 Hanson et al, 132 41 did not include terms ample, He et al 2007 Sui et al, 48 Walter et al, 2009 (disabled) for overweight or obesity in their mod- 43 Seccareccia et al, 1998 (men 45-69 y) 30 els, reporting only that overweight and ≥ 2007 ( Flegal et al, 70 y) 40 2012 (60-70 y SAHS) Fontaine et al, obesity were not associated with in- 47 Petursson et al, 2011 (women) creased mortality. Studies of BMI and 54 Greenberg et al, 2007 48 2009 (nondisabled) Walter et al, mortality sometimes selectively re- 47 Petursson et al, 2011 (men) port analyses of certain subgroups, an 68 2006 (whites) McTigue et al, 133,134 40 approach that can lead to bias. Fontaine et al, 2012 (60-70 y SALSA) 110 2011 Sonestedt et al, The study by Berrington de Gonza- 23 Lang et al, 2008 (women) 8 30 and the overlapping study by lez et al Flegal et al, 2007 (60-69 y) 74 1 Katzmarzyk et al, 2012 found results similar to Adams et al 30 2007 (25-59 y) Flegal et al, ours in their full sample but based their 68 McTigue et al, 2006 (blacks) 23 2008 (men) Lang et al, final results on a subgroup with less 77 2001 Katzmarzyk et al, than half of their original sample, ar- 43 1998 (women 20-44 y) Seccareccia et al, guing that this subgroup provided more 0.1 1.0 5 valid results than the full sample. The Hazard Ratio (95% CI) validity of this assertion has not been demonstrated, and such large-scale ex- Data markers indicate hazard ratios and error bars indicate 95% confidence intervals. SAHS indicates San An- clusions may introduce additional bias, tonio Heart Study; and SALSA, Sacramento Area Latino Study on Aging. particularly when using self-reported have shown little * data. Other studies Figure 8. Hazard Ratios for All-Cause Mortality Relative to Normal Weight in Studies That or no effect of similar exclusions. Used Self-reported Data for Participants With a Body Mass Index of 35 or Greater Strengths and Limitations Source 82 2001 Taylor and Ostbye, One of the strengths of our study is the 135 Cohen et al, 2012 (black women) large sample size and number of stud- 135 2012 (black men) Cohen et al, 15 ies included, which make the findings Al Snih et al, 2007 135 2012 (white women) Cohen et al, robust to the effects of any single study. 94 Zunzunegui et al, 2012 135 Additionally, we used a comprehen- 2012 (white men) Cohen et al, 21 Krueger et al, 2004 sive search strategy and prespecified 83 2010 Monteverde et al, standard categories. Although stan- 86 Jerant and Franks, 2012 87 2000 Strawbridge et al, dard BMI categories were developed by 27 Orpana et al, 2010 the World Health Organization and by 32 Mehta and Chang, 2009 (women) 91 the National Institutes of Health in the < Boggs et al, 12 y) 2011 (education 32 2009 (men) Mehta and Chang, 1990s, not all studies of BMI and mor- 106 Gelber et al, 2007 100 tality use standard categories as part of 2011 Leitzmann et al, 84 65 y) 2006 (women < Freedman et al, their analyses. The combination of flex- 91 Boggs et al, ≥ 12 y) 2011 (education ible categorization and selective report- 84 2006 (men < 65 y) Freedman et al, 93 Ma et al, 2011 ing can lead to wide variations in HRs 84 ≥ 2006 (women 65 y) Freedman et al, 136 Cat- even within a single data set. 84 65 y) ≥ Freedman et al, 2006 (men egorization of BMI has both advan- 5 1.0 0.1 137,138 How- tages and disadvantages. Hazard Ratio (95% CI) * References 15, 16, 18, 19, 22-25, 27, 28, 30-32, 135. Data markers indicate hazard ratios and error bars indicate 95% confidence intervals. 78 JAMA, January 2, 2013—Vol 309, No. 1 ©2013 American Medical Association. All rights reserved. Downloaded From: http://jama.jamanetwork.com/ on 01/05/2013

9 ASSOCIATION OF ALL-CAUSE MORTALITY WITH OVERWEIGHT AND OBESITY (Queensland Institute of Medical Research), Iain nificantly lower all-cause mortality. The ever, the use of predefined standard Lang, MD (Peninsula College of Medicine and Den- use of predefined standard BMI group- groupings avoids issues of post hoc and tisty), Malena Monteverde, PhD (National Council of Scientific and Technical Research, Argentina), ings can facilitate between-study com- ad hoc selection of categories and ref- Mangesh Pednekar, PhD (Sekhsaria Institute for parisons. erence categories. Public Health), Julie Simpson, PhD (University of Our study also has limitations. It Melbourne), and Joachanan Stessman, MD Dr Flegal had full access to all Author Contributions: (Hadassah-Hebrew University Medical Center), for of the data in the study and takes responsibility for addresses only all-cause mortality and providing additional information about their stud- the integrity of the data and the accuracy of the data not morbidity or cause-specific mor- ies; Yinong Chong, PhD (Centers for Disease Con- analysis. trol and Prevention), for assistance with an article in Study concept and design: Flegal, Kit, Graubard. tality. It addresses only findings Chinese; Eduardo Simoes, MD (Centers for Disease Acquisition of data: Flegal, Kit, Orpana, Graubard. related to BMI and not to other Control and Prevention), for assistance with an Flegal, Kit, Orpana, Analysis and interpretation of data: article in Portuguese; and David Check, BS (Na- aspects of body composition such as Graubard. tional Cancer Institute), for assistance with the fig- Drafting of the manuscript: Flegal. visceral fat or fat distribution. Our ures. No financial compensation was provided to Critical revision of the manuscript for important in- any of these individuals. census of these articles may be incom- Flegal, Kit, Orpana, Graubard. tellectual content: Statistical analysis: Flegal, Graubard. plete. Our coding and data abstraction The authors have com- Conflict of Interest Disclosures: REFERENCE procedures may have introduced pleted and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were re- 1. Adams KF, Schatzkin A, Harris TB, et al. Over- errors. Our information on age was ported. weight, obesity, and mortality in a large prospective limited. Because of publication bias Funding/Support: There was no external funding . cohort of persons 50 to 71 years old. N Engl J Med and selective reporting, null or nega- for this work. The Centers for Disease Control and 2006;355(8):763-778. Prevention and the National Cancer Institute Obesity: preventing and managing the global epi- 2. tive HRs may have been less likely to reviewed and approved the manuscript before sub- World Health demic: report of a WHO consultation. be published. Geographical coverage mission. Organ Tech Rep Ser . 2000;894:i-xii, 1-253. Disclaimer: The findings and conclusions in this re- 3. Expert Panel on the Identification, Evaluation, was limited. port are those of the authors and not necessarily the and Treatment of Overweight in Adults. Clinical official views of the Centers for Disease Control and guidelines on the identification, evaluation, and CONCLUSIONS Prevention or the National Cancer Institute. treatment of overweight and obesity in adults: ex- Online-Only Material: The Author Video Interview, ecutive summary. Am J Clin Nutr . 1998;68(4): Relative to normal weight, obesity (all eTables 1 through 9, eFigure, and eReferences are 899-917. grades) and grades 2 and 3 obesity were available at http://www.jama.com. 4. Stroup DF, Berlin JA, Morton SC, et al. Meta- We thank Amy Berring- Additional Contributions: analysis of observational studies in epidemiology: a pro- both associated with significantly higher ton de Gonzalez, DPhil (National Cancer Institute), posal for reporting: Meta-analysis Of Observational all-cause mortality. Grade 1 obesity was o ̃ Marselle Bevilacqua (Universidade Federal de Sa Studies in Epidemiology (MOOSE) group. . 2000; JAMA Paulo, Brazil), Michael Bursztyn, MD (Hadassah- 283(15):2008-2012. not associated with higher mortality, Hebrew University Medical Center), Sarah Cohen, DerSimonian R, Kacker R. Random-effects model 5. suggesting that the excess mortality in PhD (International Epidemiology Institute), Jane Con- for meta-analysis of clinical trials: an update. Ferrie, PhD (University College, London), Trond temp Clin Trials . 2007;28(2):105-114. obesity may predominantly be due to Heir, MD (Oslo University Hospital), Heather Keller, 6. Higgins JP, Thompson SG, Deeks JJ, Altman DG. elevated mortality at higher BMI lev- PhD (University of Waterloo), Patrick Krueger, PhD BMJ Measuring inconsistency in meta-analyses. . 2003; els. Overweight was associated with sig- (University of Colorado), Petra Lahmann, PhD 327(7414):557-560. Summary Hazard Ratios (HRs) of All-Cause Mortality for Overweight and Obesity Relative to Normal Weight From Studies Table 2. Considered Adequately Adjusted Height and Weight Self-reported or Measured Measured Self-reported Height and Weight Summary HR No. of Summary HR No. of Summary HR No. of 2 2 2 I HRs ,% ,% HRs ,% (95% CI) I (95% CI) HRs I (95% CI) 30 BMI of 25-  a a a 87.6 45 74.8 41 0.95 (0.90-1.01) 91.0 0.92 (0.88-0.96) 0.94 (0.90-0.97) All ages 86 a a a 68 0.93 (0.88-0.98) Mixed ages 79.2 34 0.96 (0.91-1.02) 89.3 91.8 0.95 (0.91-0.99) 34 65 y only  27.9 11 0.90 (0.84-0.96) 23.4 7 0.91 (0.84-0.98) 42.9 18 Age 0.90 (0.86-0.95) BMI of  30 a a a 1.11 (1.03-1.20) 20 1.33 (1.21-1.47) 22 88.0 89.3 67.1 All ages 42 1.21 (1.12-1.31) a a a 16 89.7 1.26 (1.16-1.37) 66.7 17 1.39 (1.27-1.53) 33 84.3 Mixed ages 1.13 (1.04-1.23) a a 65 y only 39.7 63.9 6 1.02 (0.81-1.29) 1.05 (0.92-1.21) 73.1 3 1.08 (0.93-1.25)  Age 9 35  BMI of 30- a a a 21 1.00 (0.92-1.09) 64.2 89.6 21 0.94 (0.84-1.05) 83.8 All ages 0.97 (0.90-1.04) 42 a a a 84.8 16 1.03 (0.94-1.12) 33 64.8 17 0.95 (0.85-1.07) 0.98 (0.91-1.06) 90.3 Mixed ages a a a  78.0 5 0.90 (0.70-1.16) 0.88 (0.69-1.12) 64.1 4 0.82 (0.46-1.47) Age 88.1 65 y only 9 35 BMI of  a a a 21 1.32 (1.20-1.46) 88.7 46.6 21 1.35 (1.16-1.57) 81.2 1.34 (1.21-1.47) 42 All ages a a 82.2 Mixed ages 16 1.37 (1.24-1.52) 40.4 17 1.34 (1.14-1.57) 33 89.6 1.35 (1.22-1.50) a a 37.8  75.2 5 1.12 (0.89-1.43) 9 4 1.40 (0.64-3.07) 1.28 (0.93-1.76) 86.8 65 y only Age Abbreviation: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared). a Indicates significant heterogeneity (  .05). P ©2013 American Medical Association. All rights reserved. JAMA, January 2, 2013—Vol 309, No. 1 79 Downloaded From: http://jama.jamanetwork.com/ on 01/05/2013

10 ASSOCIATION OF ALL-CAUSE MORTALITY WITH OVERWEIGHT AND OBESITY 7. . 1995;48 Patsopoulos NA, Evangelou E, Ioannidis JP. Sen- J Clin Epidemiol American population. specific mortality among Chinese women: prospec- tive cohort study. PLoS Med . 2010;7(9). sitivity of between-study heterogeneity in meta- (7):903-916. 27. Orpana HM, Berthelot JM, Kaplan MS, Feeny DH, analysis: proposed metrics and empirical evaluation. 45. Iribarren C, Jacobs DR, Kiefe CI, et al. Causes and McFarland B, Ross NA. BMI and mortality: results from . 2008;37(5):1148-1157. demographic, medical, lifestyle and psychosocial pre- Int J Epidemiol Obe- a national longitudinal study of Canadian adults. Berrington de Gonzalez A, Hartge P, Cerhan JR, Soc dictors of premature mortality: the CARDIA Study. 8. . 2010;18(1):214-218. sity (Silver Spring) et al. Body-mass index and mortality among 1.46 mil- Sci Med . 2005;60(3):471-482. 28. Pednekar MS, Hakama M, Hebert JR, Gupta PC. 46. N Engl J Med lion white adults. Tice JA, Kanaya A, Hue T, et al. Risk factors for . 2010;363(23): Association of body mass index with all-cause and 2211-2219. . mortality in middle-aged women. Arch Intern Med cause-specific mortality: findings from a prospective 9. Jacobs EJ, Newton CC, Wang Y, et al. Waist cir- 2006;166(22):2469-2477. Int J cohort study in Mumbai (Bombay), India. Petursson H, Sigurdsson JA, Bengtsson C, Nilsen 47. cumference and all-cause mortality in a large US cohort. Epidemiol . 2008;37(3):524-535. TI, Getz L. Body configuration as a predictor of mor- Arch Intern Med . 2010;170(15):1293-1301. 29. Simpson JA, MacInnis RJ, Peeters A, Hopper JL, tality: comparison of five anthropometric measures in 10. Jee SH, Sull JW, Park J, et al. Body-mass index Giles GG, English DR. A comparison of adiposity mea- a 12 year follow-up of the Norwegian HUNT 2 Study. and mortality in Korean men and women. N Engl J sures as predictors of all-cause mortality: the Mel- . 2011;6(10):e26621. PLoS One Med . 2006;355(8):779-787. Obesity (Silver bourne Collaborative Cohort Study. Walter S, Kunst A, Mackenbach J, Hofman A, Klenk J, Nagel G, Ulmer H, et al; VHM&PP Study 11. 48. . 2007;15(4):994-1003. Spring) Tiemeier H. Mortality and disability: the effect of over- Group. Body mass index and mortality: results of a co- 30. Flegal KM, Graubard BI, Williamson DF, Gail MH. . 2009; hort of 184,697 adults in Austria. Eur J Epidemiol . 2009;33 weight and obesity. Int J Obes (Lond) Impact of smoking and preexisting illness on esti- (12):1410-1418. 24(2):83-91. mates of the fractions of deaths associated with un- Pischon T, Boeing H, Hoffmann K, et al. General 49. Farrell SW, Braun L, Barlow CE, Cheng YJ, Blair 12. derweight, overweight, and obesity in the US SN. The relation of body mass index, cardiorespira- and abdominal adiposity and risk of death in Europe. population. Am J Epidemiol . 2007;166(8):975- N Engl J Med . 2008;359(20):2105-2120. tory fitness, and all-cause mortality in women. Obes 982. Res Whitlock G, Lewington S, Sherliker P, et al; Pro- 13. . 2002;10(6):417-423. Hotchkiss JW, Leyland AH. The relationship be- 31. spective Studies Collaboration. Body-mass index and ̈ 50. Ferrie JE, Singh-Manoux A, Kivima ki M, et al. Car- tween body size and mortality in the linked Scottish cause-specific mortality in 900 000 adults: collabora- diorespiratory risk factors as predictors of 40-year mor- Health Surveys: cross-sectional surveys with follow-up. Lancet Heart . 2009;95(15): . 2009; tive analyses of 57 prospective studies. tality in women and men. Int J Obes (Lond) . 2011;35(6):838-851. 1250-1257. 373(9669):1083-1096. Mehta NK, Chang VW. Mortality attributable to 32. 14. Zheng W, McLerran DF, Rolland B, et al. Asso- 51. Lakoski SG, Le AH, Muntner P, et al. Adiposity, obesity among middle-aged adults in the United States. ciation between body-mass index and risk of death inflammation, and risk for death in black and white . 2009;46(4):851-872. Demography N Engl J Med men and women in the United States: the Reasons for . 2011; in more than 1 million Asians. Lisko I, Tiainen K, Stenholm S, Luukkaala T, 33. Geographic and Racial Differences in Stroke (RE- 364(8):719-729. Hervonen A, Jylha M. Body mass index, waist circum- ̈ 15. Al Snih S, Ottenbacher KJ, Markides KS, Kuo YF, . 2011; J Clin Endocrinol Metab GARDS) Study. ference, and waist-to-hip ratio as predictors of mor- 96(6):1805-1814. Eschbach K, Goodwin JS. The effect of obesity on dis- tality in nonagenarians: the Vitality 90  Study. J Geron- Arch Intern Med 52. Fang X, Tang Z, Xiang M, et al. The relationship ability vs mortality in older Americans. . tol A Biol Sci Med Sci . 2011;66(11):1244-1250. 2007;167(8):774-780. between body mass index, hypertension and all- 34. Takata Y, Ansai T, Soh I, et al. Association be- cause mortality in the populaton over 55 year old in 16. Bellocco R, Jia C, Ye W, Lagerros YT. Effects of tween body mass index and mortality in an 80-year- physical activity, body mass index, waist-to-hip ratio . 2002; Zhonghua Liu Xing Bing Xue Za Zhi Beijing. old population. . 2007;55(6):913- J Am Geriatr Soc and waist circumference on total mortality risk in the 23(1):28-31. 917. Eur J Epidemiol . 2010; 53. Blain H, Carriere I, Sourial N, et al. Balance and Swedish National March Cohort. 35. Stessman J, Jacobs JM, Ein-Mor E, Bursztyn M. walking speed predict subsequent 8-year mortality in- 25(11):777-788. Normal body mass index rather than obesity predicts 17. Carlsson S, Andersson T, de Faire U, Lichtenstein dependently of current and intermediate events in well- greater mortality in elderly people: the Jerusalem Lon- ̈ functioning women aged 75 years and older. J Nutr lsson K, Ahlbom A. Body mass index and P, Michae gitudinal Study. J Am Geriatr Soc . 2009;57(12): mortality: is the association explained by genetic . 2010;14(7):595-600. Health Aging 2232-2238. . 2011;22(1):98-103. 54. factors? Greenberg JA, Fontaine K, Allison DB. Putative Epidemiology Uretsky S, Supariwala A, Singh P, et al. Impact 36. 18. Flicker L, McCaul KA, Hankey GJ, et al. Body mass biases in estimating mortality attributable to obesity of weight on long-term survival among patients with- in the US population. index and survival in men and women aged 70 to 75. . 2007; Int J Obes (Lond) out known coronary artery disease and a normal stress . 2010;58(2):234-241. J Am Geriatr Soc 31(9):1449-1455. SPECT MPI. J Nucl Cardiol . 2010;17(3):390-397. 19. Hjarta ̊ ker A, Adami HO, Lund E, Weiderpass E. Janssen I. Morbidity and mortality risk associ- 55. stbye T. Body mass index (BMI), BMI 37. Ø Keller HH, Body mass index and mortality in a prospectively stud- ated with an overweight BMI in older men and change and mortality in community-dwelling seniors Obesity (Silver Spring) ied cohort of Scandinavian women: the women’s life- women. . 2007;15(7):1827- . 2005; without dementia. J Nutr Health Aging 1840. . 2005; Eur J Epidemiol style and health cohort study. 9(5):316-320. ndell PE, Carlsson AC, Theobald H. The as- Wa 56. ̈ 20(9):747-754. Visscher TL, Rissanen A, Seidell JC, et al. Obesity 38. sociation between BMI value and long-term mortality. Hu G, Tuomilehto J, Silventoinen K, Barengo NC, 20. and unhealthy life-years in adult Finns: an empirical Int J Obes (Lond) Peltonen M, Jousilahti P. The effects of physical ac- . 2009;33(5):577-582. approach. Arch Intern Med . 2004;164(13):1413- 57. Bevilacqua MR, Gimeno SG. Abdominal obesity tivity and body mass index on cardiovascular, cancer 1420. in Japanese-Brazilians: which measure is best for pre- and all-cause mortality among 47 212 middle-aged 39. Arndt V, Rothenbacher D, Zschenderlein B, dicting all-cause and cardiovascular mortality? . 2005; Finnish men and women. Int J Obes (Lond) Cad Schuberth S, Brenner H. Body mass index and pre- 29(8):894-902. Saude Publica . 2011;27(10):1986-1996. mature mortality in physically heavily working men—a 58. Lawlor DA, Hart CL, Hole DJ, Davey Smith G. Re- Krueger PM, Rogers RG, Hummer RA, Boardman 21. ten-year follow-up of 20,000 construction workers. verse causality and confounding and the associations JD. Body mass, smoking, and overall and cause- J Occup Environ Med . 2007;49(8):913-921. Obesity (Sil- . of overweight and obesity with mortality. specific mortality among older US adults. Res Aging Fontaine KR, McCubrey R, Mehta T, et al. Body 40. 2004;26(1):82-107. ver Spring) . 2006;14(12):2294-2304. mass index and mortality rate among Hispanic adults: Osler M, Heitmann BL, Høidrup S, Jørgensen LM, Lahmann PH, Lissner L, Gullberg B, Berglund G. 22. 59. a pooled analysis of multiple epidemiologic data sets. Schroll M. Food intake patterns, self rated health and A prospective study of adiposity and all-cause mor- Int J Obes (Lond) . 2012;36(8):1121-1126. ̈ . Diet and Cancer Study. mortality in Danish men and women: a prospective Obes Res tality: the Malmo 41. Sui X, LaMonte MJ, Laditka JN, et al. Cardiores- . J Epidemiol Community Health observational study. 2002;10(5):361-369. piratory fitness and adiposity as mortality predictors 2001;55(6):399-403. Lang IA, Llewellyn DJ, Alexander K, Melzer D. Obe- 23. . 2007;298(21):2507-2516. JAMA in older adults. sity, physical function, and mortality in older adults. 60. Crespo CJ, Palmieri MR, Perdomo RP, et al. The Lubin F, Lusky A, Chetrit A, Dankner R. Lifestyle 42. . 2008;56(8):1474-1478. relationship of physical activity and body weight with J Am Geriatr Soc J Nutr and ethnicity play a role in all-cause mortality. . all-cause mortality: results from the Puerto Rico Heart 24. Luchsinger JA, Patel B, Tang MX, Schupf N, 2003;133(4):1180-1185. . 2002;12(8):543- Ann Epidemiol Health Program. Mayeux R. Body mass index, dementia, and mortal- 43. Seccareccia F, Lanti M, Menotti A, Scanga M. Role 552. . 2008;12 J Nutr Health Aging ity in the elderly. of body mass index in the prediction of all cause mor- (2):127-131. Gu D, He J, Duan X, et al. Body weight and mor- 61. tality in over 62,000 men and women: the Italian RIFLE 25. tality among men and women in China. . 2006; McAuley PA, Kokkinos PF, Oliveira RB, Emerson JAMA J Epi- Pooling Project: Risk Factor and Life Expectancy. BT, Myers JN. Obesity paradox and cardiorespiratory 295(7):776-783. demiol Community Health . 1998;52(1):20-26. fitness in 12,417 male veterans aged 40 to 70 years. 62. Faeh D, Braun J, Tarnutzer S, Bopp M. Obesity 44. Hanson RL, McCance DR, Jacobsson LT, et al. The . 2010;85(2):115-121. Mayo Clin Proc but not overweight is associated with increased mor- U-shaped association between body mass index and Nechuta SJ, Shu XO, Li HL, et al. Combined im- 26. . 2011;26(8):647-655. tality risk. Eur J Epidemiol mortality: relationship with weight gain in a Native 63. Tsai SP, Donnelly RP, Wendt JK. Obesity and mor- pact of lifestyle-related factors on total and cause- 80 JAMA, January 2, 2013—Vol 309, No. 1 ©2013 American Medical Association. All rights reserved. Downloaded From: http://jama.jamanetwork.com/ on 01/05/2013

11 ASSOCIATION OF ALL-CAUSE MORTALITY WITH OVERWEIGHT AND OBESITY tality in a prospective study of a middle-aged indus- Yates LB, Djousse ́ L, Kurth T, Buring JE, Gaziano 98. 81. Locher JL, Roth DL, Ritchie CS, et al. Body mass index, weight loss, and mortality in community- J Occup Environ Med . 2006;48 trial population. JM. Exceptional longevity in men: modifiable . J Gerontol A Biol Sci Med Sci dwelling older adults. (1):22-27. factors associated with survival and function to 2007;62(12):1389-1392. Atlantis E, Browning C, Kendig H. Body mass in- 64. age 90 years. Arch Intern Med . 2008;168(3):284- 82. Taylor DH Jr, Ostbye T. The effect of middle- and dex and unintentional weight change associated with 290. old-age body mass index on short-term mortality in all-cause mortality in older Australians: the Mel- Iversen L, Hannaford PC, Lee AJ, Elliott AM, 99. J Am Geriatr Soc older people. . 2001;49(10):1319- bourne Longitudinal Studies on Healthy Ageing Fielding S. Impact of lifestyle in middle-aged women 1326. . 2010;39(5):643-646. (MELSHA). Age Ageing on mortality: evidence from the Royal College of Gen- 83. Monteverde M, Noronha K, Palloni A, Novak B. 65. Cabrera MA, Wajngarten M, Gebara OC, Diament eral Practitioners’ Oral Contraception Study. BrJGen Obesity and excess mortality among the elderly in the J. Relationship between body mass index, waist cir- Pract . 2010;60(577):563-569. Leitzmann MF, Moore SC, Koster A, et al. Waist . 2010; United States and Mexico. Demography cumference, and waist-to-hip ratio and mortality in 100. circumference as compared with body-mass index in 47(1):79-96. Cad Saude elderly women: a 5-year follow-up study. . PLoS One predicting mortality from specific causes. 84. Freedman DM, Sigurdson AJ, Rajaraman P, Doody . 2005;21(3):767-775. Publica 2011;6(4):e18582. MM, Linet MS, Ron E. The mortality risk of smoking Janssen I, Bacon E. Effect of current and midlife 66. ck Weitoft G, Eliasson M, Rose ́ nM.Un- 101. Ringba ̈ and obesity combined. Am J Prev Med . 2006;31 Obesity obesity status on mortality risk in the elderly. derweight, overweight and obesity as risk factors for (5):355-362. (Silver Spring) . 2008;16(11):2504-2509. Scand J Public Health . mortality and hospitalization. 85. Majer IM, Nusselder WJ, Mackenbach JP, Kunst m G, Nilsson P, ̈ 67. Jonsson S, Hedblad B, Engstro 2008;36(2):169-176. AE. Life expectancy and life expectancy with disabil- Berglund G, Janzon L. Influence of obesity on cardio- Bessonova L, Marshall SF, Ziogas A, et al. The 102. ity of normal weight, overweight, and obese smok- vascular risk: twenty-three-year follow-up of 22,025 association of body mass index with mortality in the . ers and nonsmokers in Europe. Obesity (Silver Spring) Int J Obes men from an urban Swedish population. Int J Cancer California Teachers Study. . 2011;129 2011;19(7):1451-1459. . 2002;26(8):1046-1053. Relat Metab Disord (10):2492-2501. 86. Jerant A, Franks P. Body mass index, diabetes, hy- 68. McTigue K, Larson JC, Valoski A, et al. Mortality Gray L, Hart CL, Smith GD, Batty GD. What is 103. pertension, and short-term mortality: a population- and cardiac and vascular outcomes in extremely obese the predictive value of established risk factors for total J Am Board based observational study, 2000-2006. women. JAMA . 2006;296(1):79-86. and cardiovascular disease mortality when measured Fam Med . 2012;25(4):422-431. 69. Miller MD, Crotty M, Giles LC, et al. Corrected before middle age? pooled analyses of two prospec- 87. Strawbridge WJ, Wallhagen MI, Shema SJ. New arm muscle area: an independent predictor of long- tive cohort studies from Scotland. Eur J Cardiovasc Prev NHLBI clinical guidelines for obesity and overweight: term mortality in community-dwelling older adults? . 2010;17(1):106-112. Rehabil will they promote health? Am J Public Health . 2000; J Am Geriatr Soc . 2002;50(7):1272-1277. Stevens J, Cai J, Juhaeri, Thun MJ, Wood JL. Evalu- 104. 90(3):340-343. Batty GD, Shipley MJ, Jarrett RJ, Breeze E, Marmot 70. ation of WHO and NHANES II standards for over- 88. Ford J, Spallek M, Dobson A. Self-rated health and MG, Davey Smith G. Obesity and overweight in re- . 2000; weight using mortality rates. JAmDietAssoc a healthy lifestyle are the most important predictors lation to disease-specific mortality in men with and 100(7):825-827. of survival in elderly women. Age Ageing . 2008; without existing coronary heart disease in London: the van Dam RM, Li T, Spiegelman D, Franco OH, 105. 37(2):194-200. Heart . 2006;92(7):886- original Whitehall Study. Hu FB. Combined impact of lifestyle factors on mor- Niedhammer I, Bourgkard E, Chau N; Lorhandi- 89. 892. tality: prospective cohort study in US women. BMJ . cap Study Group. Occupational and behavioural fac- 71. Gale CR, Martyn CN, Cooper C, Sayer AA. Grip 2008;337:a1440. tors in the explanation of social inequalities in prema- Int J strength, body composition, and mortality. Gelber RP, Kurth T, Manson JE, Buring JE, Gaziano 106. ture and total mortality: a 12.5-year follow-up in the Epidemiol . 2007;36(1):228-235. JM. Body mass index and mortality in men: evaluat- Lorhandicap Study. Eur J Epidemiol . 2011;26(1): Visscher TL, Seidell JC, Menotti A, et al. Under- 72. ing the shape of the association. . Int J Obes (Lond) 1-12. weight and overweight in relation to mortality among 2007;31(8):1240-1247. Haapanen-Niemi N, Miilunpalo S, Pasanen M, 90. men aged 40-59 and 50-69 years: the Seven Coun- Wannamethee SG, Shaper AG, Lennon L, 107. Vuori I, Oja P, Malmberg J. Body mass index, physi- . 2000;151(7):660- Am J Epidemiol tries Study. Whincup PH. Decreased muscle mass and increased cal inactivity and low level of physical fitness as de- 666. central adiposity are independently related to mor- terminants of all-cause and cardiovascular disease mor- Baldinger B, Schwarz C, Jaggy C. Cardiovascular 73. tality in older men. Am J Clin Nutr . 2007;86(5): tality—16 y follow-up of middle-aged and elderly men risk factors, BMI and mortality in a cohort of Swiss males 1339-1346. and women. . 2000; Int J Obes Relat Metab Disord (1976-2001) with high-sum-assured life insurance Suadicani P, Hein HO, von Eyben FE, Gyntelberg 108. 24(11):1465-1474. . 2006;38(1):44-53. J Insur Med cover. F. Metabolic and lifestyle predictors of ischemic heart 91. Boggs DA, Rosenberg L, Cozier YC, et al. Gen- Katzmarzyk PT, Reeder BA, Elliott S, et al. Body 74. disease and all-cause mortality among normal weight, eral and abdominal obesity and risk of death among mass index and risk of cardiovascular disease, cancer overweight, and obese men: a 16-year follow-up in N Engl J Med black women. . 2011;365(10):901- . 2012; and all-cause mortality. Can J Public Health the Copenhagen Male Study. Metab Syndr Relat 908. 103(2):147-151. . 2009;7(2):97-104. Disord Lantz PM, Golberstein E, House JS, Morenoff J. 92. Ioachimescu AG, Brennan DM, Hoar BM, 75. 109. Seidell JC, Verschuren WM, van Leer EM, Socioeconomic and behavioral risk factors for Hoogwerf BJ. The lipid accumulation product and all- Kromhout D. Overweight, underweight, and mortality in a national 19-year prospective study cause mortality in patients at high cardiovascular risk: mortality: a prospective study of 48,287 men and of US adults. . 2010;70(10):1558- Soc Sci Med Obesity (Silver Spring) a PreCIS database study. . 2010; Arch Intern Med . 1996;156(9):958- women. 1566. 18(9):1836-1844. 963. Ma J, Flanders WD, Ward EM, Jemal A. Body mass 93. Heir T, Erikssen J, Sandvik L. Overweight as pre- 76. ̈ lt E, Sonestedt E, Gullberg B, Ericson U, Wirfa 110. index in young adulthood and premature death: analy- dictor of long-term mortality among healthy, middle- Hedblad B, Orho-Melander M. Association between ses of the US National Health Interview Survey linked aged men: a prospective cohort study. Prev Med . 2011; fat intake, physical activity and mortality depending . 2011;174(8):934- Am J Epidemiol mortality files. 52(3-4):223-226. on genetic variation in FTO. Int J Obes (Lond) . 2011; 944. Katzmarzyk PT, Craig CL, Bouchard C. Original 77. 35(8):1041-1049. 94. Zunzunegui MV, Sanchez MT, Garcia A, Casado article underweight, overweight and obesity: relation- Flegal KM, Carroll MD, Kit BK, Ogden CL. Preva- 111. JM, Otero A. Body mass index and long-term mor- ships with mortality in the 13-year follow-up of the lence of obesity and trends in the distribution of body J Ag- tality in an elderly Mediterranean population. . 2001; Canada Fitness Survey. J Clin Epidemiol mass index among US adults, 1999-2010. JAMA . 2012; . 2012;24(1):29-47. ing Health 54(9):916-920. 307(5):491-497. 95. Nagai M, Kuriyama S, Kakizaki M, et al. Impact 78. Cesari M, Pahor M, Lauretani F, et al. Skeletal Public Health Agency of Canada and the Cana- 112. of obesity, overweight and underweight on life ex- muscle and mortality results from the InCHIANTI Study. dian Institute for Health Information. Obesity in pectancy and lifetime medical expenditures: the Oh- . 2009;64(3):377- J Gerontol A Biol Sci Med Sci Canada: a joint report from the Public Health Agency saki Cohort Study. BMJ Open . 2012;2(3). 384. of Canada and the Canadian Institute for Health Fujino Y; Japan Collaborative Cohort Study 96. ̈ 79. mJ,LindL.Im- ̈ Arnlo v J, Ingelsson E, Sundstro Information. http://www.phac-aspc.gc.ca/hp-ps for Evaluation of Cancer. Anthropometry, develop- pact of body mass index and the metabolic syn- /hl-mvs/oic-oac/assets/pdf/oic-oac-eng.pdf. Acces- ment history and mortality in the Japan Collaborative drome on the risk of cardiovascular disease and death sibility verified December 6, 2012. Cohort Study for Evaluation of Cancer (JACC). Circulation . 2010;121(2): in middle-aged men. 113. National Health Service Information Centre. Asian Pac J Cancer Prev . 2007;8(suppl):105- 230-236. Health survey for England: 2010 trend tables. http: 112. 80. Tsai SP, Wen CP, Chan HT, Chiang PH, Tsai MK, //www.ic.nhs.uk/pubs/hse10trends. Accessed April Corrada MM, Kawas CH, Mozaffar F, Paganini-Hill 97. Cheng TY. The effects of pre-disease risk factors within 9, 2012. A. Association of body mass index and weight change metabolic syndrome on all-cause and cardiovascular 114. McGee DL; Diverse Populations Collaboration. Am J Epidemiol . with all-cause mortality in the elderly. . 2008; disease mortality. Diabetes Res Clin Pract Body mass index and mortality: a meta-analysis based 2006;163(10):938-949. 82(1):148-156. ©2013 American Medical Association. All rights reserved. JAMA, January 2, 2013—Vol 309, No. 1 81 Downloaded From: http://jama.jamanetwork.com/ on 01/05/2013

12 ASSOCIATION OF ALL-CAUSE MORTALITY WITH OVERWEIGHT AND OBESITY on person-level data from twenty-six observational 123. Steinberg BA, Cannon CP, Hernandez AF, Pan . 1997; Obes Res self-reported body mass index. studies. Ann Epidemiol . 2005;15(2):87-97. W, Peterson ED, Fonarow GC. Medical therapies and 5(4):308-314. Janssen I, Mark AE. Elevated body mass index 115. invasive treatments for coronary artery disease by body Chiolero A, Peytremann-Bridevaux I, Paccaud 131. Obes Rev . 2007; and mortality risk in the elderly. mass: the “obesity paradox” in the Get With The F. Associations between obesity and health condi- 8(1):41-59. . 2007;100(9): Am J Cardiol Guidelines database. tions may be overestimated if self-reported body 116. Carnethon MR, De Chavez PJ, Biggs ML, et al. 1331-1335. Obes Rev . 2007;8(4):373- mass index is used. Association of weight status with mortality in adults Hastie CE, Padmanabhan S, Slack R, et al. Obe- 124. 374. JAMA . 2012;308(6):581- with incident diabetes. sity paradox in a cohort of 4880 consecutive patients He J, Gu D, Wu X, et al. Major causes of death 132. 590. Eur undergoing percutaneous coronary intervention. among men and women in China. N Engl J Med . 2005; 117. Romero-Corral A, Montori VM, Somers VK, et al. Heart J . 2010;31(2):222-226. 353(11):1124-1134. Association of bodyweight with total mortality and with Auyeung TW, Lee JS, Leung J, Kwok T, Leung 125. 133. Hahn S, Williamson PR, Hutton JL, Garner P, Flynn cardiovascular events in coronary artery disease: a sys- PC, Woo J. Survival in older men may benefit from EV. Assessing the potential for bias in meta-analysis Lancet tematic review of cohort studies. . 2006; being slightly overweight and centrally obese—a 5-year due to selective reporting of subgroup analyses within . 2000;19(24):3325-3336. 368(9536):666-678. follow-up study in 4,000 older adults using DXA. studies. Stat Med 134. Wang R, Lagakos SW, Ware JH, Hunter DJ, Doehner W, Clark A, Anker SD. The obesity para- 118. . 2010;65(1):99- J Gerontol A Biol Sci Med Sci Drazen JM. Statistics in medicine—reporting of sub- Eur Heart J dox: weighing the benefit. . 2010;31 104. . 2007; N Engl J Med group analyses in clinical trials. (2):146-148. Manson JE, Stampfer MJ, Hennekens CH, Willett 126. 357(21):2189-2194. Lainscak M, von Haehling S, Doehner W, Anker 119. WC. Body weight and longevity: a reassessment. 135. Cohen SS, Signorello LB, Cope EL, et al. Obe- SD. The obesity paradox in chronic disease: facts and . 1987;257(3):353-358. JAMA sity and all-cause mortality among black adults and numbers. J Cachexia Sarcopenia Muscle . 2012; 127. Gillum RF, Sempos CT. Ethnic variation in va- white adults. Am J Epidemiol . 2012;176(5):431- 3(1):1-4. lidity of classification of overweight and obesity using 442. 120. Oreopoulos A, McAlister FA, Kalantar-Zadeh self-reported weight and height in American women 136. Ioannidis JP. Why most discovered true asso- K, et al. The relationship between body mass index, and men: the Third National Health and Nutrition Ex- . 2008;19(5): Epidemiology ciations are inflated. treatment, and mortality in patients with established amination Survey. Nutr J . 2005;4:27. 640-648. coronary artery disease: a report from APPROACH. NyholmM,GullbergB,MerloJ,Lundqvist-Persson 128. Welch HG, Schwartz LM, Woloshin S. The ex- 137. . 2009;30(21):2584-2592. Eur Heart J C, Ra ̊ stam L, Lindblad U. The validity of obesity based aggerated relations between diet, body weight and 121. Chang VW, Asch DA, Werner RM. Quality of onself-reportedweightandheight:implicationsforpopu- mortality: the case for a categorical data approach. JAMA care among obese patients. . 2010;303(13): Obesity (Silver Spring) lation studies. . 2007;15(1): . 2005;172(7):891-895. CMAJ 1274-1281. 197-208. 138. Frøslie KF, Røislien J, Laake P, Henriksen T, Schenkeveld L, Magro M, Oemrawsingh RM, 122. 129. Spencer EA, Appleby PN, Davey GK, Key TJ. Va- Qvigstad E, Veierød MB. Categorisation of continu- et al. The influence of optimal medical treatment on lidity of self-reported height and weight in 4808 EPIC- ous exposure variables revisited: a response to the Hy- the “obesity paradox,” body mass index and long- Oxford participants. Public Health Nutr . 2002; perglycaemia and Adverse Pregnancy Outcome term mortality in patients treated with percutaneous 5(4):561-565. BMC Med Res Methodol (HAPO) Study. . 2010; BMJ coronary intervention: a prospective cohort study. Plankey MW, Stevens J, Flegal KM, Rust PF. Pre- 130. 10:103. . 2012;2:e000535. Open diction equations do not eliminate systematic error in The test of intelligence [is] not how much we know how to do, but how we behave when we don’t know what to do. Similarly, any situation, any activity, that puts before us real problems, that we have to solve for ourselves, problems for which there are no an- swers in any book, sharpens our intelligence. —John Holt (1923-1985) 82 JAMA, January 2, 2013—Vol 309, No. 1 ©2013 American Medical Association. All rights reserved. Downloaded From: http://jama.jamanetwork.com/ on 01/05/2013

Related documents

CalCOFI Atlas 33

CalCOFI Atlas 33

THE EARLY STAGES IN OF THE FISHES CALIFORNIA CURRENT REGION CALIFORNIA FISHERIES COOPERATIVE OCEANIC INVESTIGATIONS ATLAS NO. 33 BY THE SPONSORED STATES OF COMMERCE DEPARTMENT UNITED OCEANIC AND ATMOS...

More info »
vol9 organic ligands

vol9 organic ligands

C HERMODYNAMICS HEMICAL T OMPOUNDS AND C OMPLEXES OF OF C U, Np, Pu, Am, Tc, Se, Ni and Zr O ELECTED WITH RGANIC L IGANDS S Wolfgang Hummel (Chairman) Laboratory for Waste Management Paul Scherrer Ins...

More info »
oldnew 11.dvi

oldnew 11.dvi

C ́edric Villani O ptimal transport, old and new June 13, 2008 Springer Berlin Heidelberg NewYork Hong Kong London Milan Paris Tokyo

More info »
NEPA A Study of it's Effectiveness After 25 Years

NEPA A Study of it's Effectiveness After 25 Years

THE NATIONAL ENVIRONMENTAL POLICY ACT y of Its Effectiveness After Twent y -five Years A Stud y Council on Environmental Qualit Executive Office of the President y 1997 Januar

More info »
13128

13128

This PDF is available from The National Academies Press at http://www.nap.edu/catalog.php?record_id=13128 The Health of Lesbian, Gay, Bisexual, and Transgender People: Building a Foundation for Better...

More info »
c9nj00028c1

c9nj00028c1

Chemistry. of Journal New for (ESI) Material Supplementary Electronic 2019 is journal This The Recherche © Royal Society of Chemistry and the Centre National de la Scientifique Supporting Information ...

More info »
lmibook

lmibook

Linear Matrix Inequalities in System and Con trol Theory

More info »
fragstats.help.4.2

fragstats.help.4.2

FRAGSTAT S HELP 21 April 2015 Kevin McGariga l, onsulting Sole proprietor, LandEco C onservation essor Prof , Department of Environmental C University of Mas sachusetts , Amherst [email protected] mcgariga o.umas...

More info »
Microsoft Word   Table of Contents Full Report with ES.doc

Microsoft Word Table of Contents Full Report with ES.doc

Stern Review: The Economics of Climate Change PAGE TABLE OF CONTENTS i-xxvii Executive Summary i Preface & Acknowledgements iv Introduction to Review vi Summary of Conclusions Part I Climate change: o...

More info »
P:\TEX\BOOKS\fr18kelley2000reprintredo\optimtr

P:\TEX\BOOKS\fr18kelley2000reprintredo\optimtr

Kelley fm8_00.qxd 9/20/2004 2:56 PM Page 5 Iterative Methods for Optimization C.T. Kelley North Carolina State University Raleigh, North Carolina Society for Industrial and Applied Mathematics Philade...

More info »
Champion Report

Champion Report

C A M P I O N S H O F C H A N G E the i m p a c t of the arts on l e arning

More info »
UNICEF Birth_Registration_Handbook

UNICEF Birth_Registration_Handbook

A PASSPORT TO PRO TECTION A GUIDE TO BIRTH REGIST RATION PRO GRAMMING

More info »
The Health Consequences of Smoking   50 Years of Progress: A Report of the Surgeon General

The Health Consequences of Smoking 50 Years of Progress: A Report of the Surgeon General

The Health Consequences of Smoking—50 Years of Progress A Report of the Surgeon General U.S. Department of Health and Human Services

More info »
Campus Climate Survey Validation Study Final Technical Report

Campus Climate Survey Validation Study Final Technical Report

Bureau of Justice Statistics Research and Development Series Campus Climate Survey Validation Study Final Technical Report Christopher Krebs, Christine Lindquist, Marcus Berzofsky, Bonnie Shook-Sa, an...

More info »
Fourth National Report on Human Exposure to Environmental Chemicals Update

Fourth National Report on Human Exposure to Environmental Chemicals Update

201 8 Fourth National Report on Human Exposure to Environmental Chemicals U pdated Tables, March 2018 , Volume One

More info »
Microsoft PowerPoint   2019 05 RELX overview May 2019

Microsoft PowerPoint 2019 05 RELX overview May 2019

The global provider of information- based analytics and decision tools May 2019

More info »
nf2005

nf2005

N ATION AL C URRICUL UM F RAMEW ORK 2005

More info »
Then and Now

Then and Now

T H E N & N O W CHANGES IN CHILD CARE STAFFING, 1994-2000 T echnical Repor t PRINCIPAL INVESTIGATORS: Marcy Whitebook Laura Sakai Emily Gerber Carollee Howes A joint project of the: Center for the Chi...

More info »
tsa4

tsa4

i i “tsa4_trimmed” — 2017/12/8 — 15:01 — page 1 — #1 i i Springer Texts in Statistics Robert H. Shumway David S. Sto er Time Series Analysis and Its Applications With R Examples Fourth Edition i i i ...

More info »