Body-Mass Index and Mortality in Korean Men and Women
http://www.100md.com
《新英格兰医药杂志》
ABSTRACT
Background Obesity is associated with diverse health risks, but the role of body weight as a risk factor for death remains controversial.
Methods We examined the association between body weight and the risk of death in a 12-year prospective cohort study of 1,213,829 Koreans between the ages of 30 and 95 years. We examined 82,372 deaths from any cause and 48,731 deaths from specific diseases (including 29,123 from cancer, 16,426 from atherosclerotic cardiovascular disease, and 3362 from respiratory disease) in relation to the body-mass index (BMI) (the weight in kilograms divided by the square of the height in meters).
Results In both sexes, the average baseline BMI was 23.2, and the rate of death from any cause had a J-shaped association with the BMI, regardless of cigarette-smoking history. The risk of death from any cause was lowest among patients with a BMI of 23.0 to 24.9. In all groups, the risk of death from respiratory causes was higher among subjects with a lower BMI, and the risk of death from atherosclerotic cardiovascular disease or cancer was higher among subjects with a higher BMI. The relative risk of death associated with BMI declined with increasing age.
Conclusions Underweight, overweight, and obese men and women had higher rates of death than men and women of normal weight. The association of BMI with death varied according to the cause of death and was modified by age, sex, and smoking history.
Although obesity is widely accepted as an important health risk, the optimal body-mass index (BMI) (the weight in kilograms divided by the square of the height in meters) and the effects of being either underweight or overweight on the risk of death are controversial. In the Cancer Prevention Study (CPS) II,1 sponsored by the American Cancer Society, the rate of death was lowest among men with a BMI of 23.5 to 24.9 and among women with a BMI of 22.0 to 23.4; above and below these levels, the risk of death increased. However, being overweight was not associated with an increased risk of death in the National Health and Nutrition Examination Survey (NHANES) I, II, or III.2 The results of other studies have been mixed3,4 and may reflect differences in age, the number or extent of coexisting illnesses, and BMI distributions among subjects, as well as in analytic approaches.1,5
Since studies of the association between BMI and death have been conducted primarily in Western populations, it is uncertain whether the findings of these studies can be applied to other groups. Continental Asian populations have a higher percentage of body fat for a given BMI than do whites,6 and a World Health Organization (WHO) Expert Consultation proposed a new BMI cutoff of 23.0 for public health action in Asia.7 The use of this cutoff, however, was not directly supported by data on mortality.8,9 Indeed, deaths from any cause were lowest among men with a BMI of 24.0 to 24.9 and women with a BMI of 25.0 to 26.9 in a representative group of Chinese subjects.10,11
We conducted a prospective cohort study of BMI and the risk of death from any cause and from specific diseases in more than 1 million Koreans in the Korean Cancer Prevention Study (KCPS).12,13
Methods
Study Population
We enrolled 1,329,525 Koreans between the ages of 30 and 95 years who had undergone one biennial medical evaluation through the National Health Insurance Corporation between 1992 and 1995.12,13 Of the subjects, 784,870 (59.0 percent) were enrolled in 1992, 367,903 (27.7 percent) in 1993, 98,417 (7.4 percent) in 1994, and 78,335 (5.9 percent) in 1995.
To avoid confounding of the association between BMI and the risk of death by preexisting disease,14,15 904 subjects who died before January 1, 1993, were excluded from the study, as were 87,911 subjects who reported having atherosclerotic cardiovascular disease, cancer, liver disease, diabetes, or a respiratory disease at or before the initial study visit. In addition, 26,881 subjects with missing information about BMI or alcohol intake and those with an extremely low BMI (less than 16.0) or short stature (1.30 m or less) were excluded. The final sample included 1,213,829 subjects.
Because the study involved data that were routinely collected, consent was not specifically obtained. The institutional review boards of Yonsei University and the Johns Hopkins Bloomberg School of Public Health approved the study.
Data Collection
Enrollees in the National Health Insurance Corporation undergo standardized examinations every two years at local hospitals. During visits between 1992 and 1995, subjects reported on their smoking habits and alcohol consumption, and weight and height measurements were recorded while subjects were wearing light clothing. Subjects were seated for blood-pressure measurement. Blood samples were obtained after an overnight fast for white-cell counts and clinical chemical analysis. Quality-control procedures were performed in accordance with the Korean Association of Laboratory Quality Control.
Follow-up and Outcome Classification
Deaths among subjects through December 31, 2004, were confirmed by matching the information to death records. Death certificates from the National Statistical Office were identified with the use of identification numbers assigned to subjects at birth. Abstractors coded causes of death according to the International Classification of Diseases, 10th Revision.16 We used underlying causes as reported.
We attempted to minimize the effect of existing medical conditions on the baseline BMI by excluding all events that occurred among subjects during the first two years of follow-up. Thus, general follow-up began on January 1 of the third year after the baseline visit. In a sensitivity analysis, we excluded the first five years of follow-up.
Statistical Analysis
Proportional-hazards models were used to evaluate the association between the baseline BMI and death.17 The BMI was categorized as less than 18.5, 18.5 to 19.9, 20.0 to 21.4, 21.5 to 22.9, 23.0 to 24.9, 25.0 to 26.4, 26.5 to 27.9, 28.0 to 29.9, 30 to 31.9, or 32.0 or more. Analyses were performed separately in men and women and were adjusted for the following covariates: age at enrollment (continuous variable), alcohol intake (five categories based on grams consumed per day: 0, 1 to 24 g, 25 to 49 g, 50 to 99 g, and 100 g or more), and participation in regular physical activity (yes or no). Because the proportion of women who reported having smoked cigarettes was small, analyses in women were restricted to those who reported never having smoked. Analyses of men who reported having smoked were adjusted for smoking status (never smoked, former smoker, or current smoker) and the number of cigarettes smoked daily among current smokers (1 to 9, 10 to 19, and 20 or more).
Modification of the effect of BMI was assessed by the inclusion of interaction terms of BMI category indicators with indicator variables for sex, age (three categories), and smoking history (two categories, one consisting of current and former smokers and one of lifetime nonsmokers). All analyses were conducted with the use of SAS software, version 9 (SAS Institute).
Results
The average BMI was 23.2 for both sexes (Table 1), and the majority of subjects had a BMI below 25.0. The BMI was below 18.5 in 2.2 percent of men and 4.7 percent of women; above 25.0 in 23.8 percent and 26.8 percent, respectively; and above 30.0 in 0.8 percent and 2.4 percent, respectively. Systolic blood pressure, total serum cholesterol, fasting serum glucose level, and white-cell count had strong, progressive associations with increasing BMIs (Table 2).
Table 1. Baseline Characteristics of the Study Population.
Table 2. Differences in Baseline Systolic Blood Pressure and Clinical Chemical Analyses, According to BMI.
During follow-up, 58,312 men died (including 22,249 from cancer, 10,486 from atherosclerotic cardiovascular causes, and 2442 from respiratory causes) and 24,060 women died (including 6874 from cancer, 5940 from atherosclerotic cardiovascular causes, and 920 from respiratory causes). The shape of the curve showing the association between BMI and the risk of death from any cause was similar in men, regardless of their smoking history, and in women who reported never having smoked (Figure 1A). The hazard ratio was higher at the lowest and highest BMI values.
Figure 1. Hazard Ratios for Death from Any Cause and from Any Cause Except Respiratory, According to BMI and Smoking History.
Data are from the KCPS, 1993–2004.12,13 The reference category was a BMI of 23.0 to 24.9. Results for men who reported having smoked cigarettes were further adjusted for whether the subject was a former smoker or a current smoker and the number of cigarettes smoked per day (1 to 9, 10 to 19, and 20 or more). All hazard ratios were adjusted for age.
Men with a BMI of 23.0 to 24.9 who reported never having smoked had the lowest risk of death from any cause (Table 1 of the Supplementary Appendix, available with the full text of this article at www.nejm.org). As compared with men with a BMI of 23.0 to 24.9, men who reported never having smoked had a hazard ratio for death from any cause of 1.29 (95 percent confidence interval, 1.15 to 1.44) in association with a BMI of less than 18.5, a hazard ratio of 1.04 (95 percent confidence interval, 0.98 to 1.10) in association with a BMI of 25.0 to 29.9, and a hazard ratio of 1.71 (95 percent confidence interval, 1.44 to 2.03) in association with a BMI of 30.0 or more. As compared with men with a BMI of 23.0 to 24.9, men who reported having smoked had a hazard ratio for death from any cause of 1.36 (95 percent confidence interval, 1.30 to 1.42) in association with a BMI of less than 18.5, a hazard ratio of 0.98 (95 percent confidence interval, 0.95 to 1.01) in association with a BMI of 25.0 to 29.9, and a hazard ratio of 1.31 (95 percent confidence interval, 1.18 to 1.45) in association with a BMI of 30.0 or more.
Among women (with the analysis restricted to those who reported never having smoked), the risk of death from any cause was lowest for those with a BMI of 23.0 to 24.9 and similar for those with a BMI of 20.0 to 26.4 (Figure 1A, and Table 2 of the Supplementary Appendix). As compared with women with a BMI of 23.0 to 24.9, women with a BMI of less than 18.5 had a hazard ratio for death from any cause of 1.17 (95 percent confidence interval, 1.09 to 1.26), women with a BMI of 25.0 to 29.9 had a hazard ratio of 1.04 (95 percent confidence interval, 1.00 to 1.08), and women with a BMI of 30.0 or more had a hazard ratio of 1.20 (95 percent confidence interval, 1.10 to 1.30).
The association between BMI and the risk of death according to the cause of death had a similar pattern of variation for both sexes (Figure 2, and Table 1 and Table 2 of the Supplementary Appendix). The risk of death from respiratory causes decreased progressively with increasing BMI, whereas the risk of death from atherosclerotic cardiovascular causes increased steadily with increasing BMI. The risk of death from cancer increased at a BMI above 26.0 to 28.0. Deaths from respiratory causes explained some of the increase in the risk of death at a low BMI (Figure 1B). For deaths associated with lung disease, the association between BMI and the risk of death was similar for the major categories of pulmonary illnesses, including tuberculosis, chronic obstructive pulmonary disease (COPD), asthma, and pneumonia; the association persisted after the exclusion of the first five years of follow-up. With this exclusion, the hazard ratio for death from respiratory causes that was associated with a decrease in BMI of 1.0 was 1.26 (95 percent confidence interval, 1.20 to 1.31) among men who reported never having smoked, 1.25 (95 percent confidence interval, 1.22 to 1.27) among men who reported having smoked, and 1.08 (95 percent confidence interval, 1.05 to 1.12) among women.
Figure 2. Hazard Ratios for Death from Cancer and from Atherosclerotic Cardiovascular and Respiratory Causes, According to BMI and Smoking History.
Data are from the KCPS, 1993–2004.12,13 The reference category was a BMI of 23.0 to 24.9. Results for men who reported having smoked cigarettes were further adjusted for whether the subject was a former smoker or a current smoker and the number of cigarettes smoked per day (1 to 9, 10 to 19, and 20 or more). The number of deaths from respiratory causes among subjects with a BMI of 30.0 or more was too small to yield a reliable estimate of relative risks. All hazard ratios were adjusted for age. Panels A, B, and C have different scales for hazard ratios in the vertical axes.
The relative increase in the risk of death from any cause that was associated with a high BMI was dependent on age (Figure 3). For both sexes, the highest relative risks associated with a high BMI were observed among subjects younger than 50 years of age. An increase in BMI to more than 25.0 was not associated with an increased risk of death from any cause among men or women who were 65 years or older at baseline. The interaction between BMI and age was significant (P<0.001), as were interactions between BMI and sex and BMI and smoking history (P<0.001 for both).
Figure 3. Hazard Ratios for Death from Any Cause among Men and Women with No History of Smoking, According to Age Group and BMI.
Data are from the KCPS, 1993–2004.12,13 The reference category was a BMI of 23.0 to 24.9. All hazard ratios were adjusted for age. Panel A and Panel B have different scales for hazard ratios.
We explored whether the association between BMI and the risk of death from atherosclerotic cardiovascular disease could be explained by systolic blood pressure or levels of blood glucose or cholesterol. As expected, adjustment for these factors attenuated this association (Table 3). Analyses of death from any cause, from cancer, and from respiratory causes adjusted for risk factors are shown in Tables 3, 4, and 5, respectively, of the Supplementary Appendix.
Table 3. Hazard Ratios for Death from Atherosclerotic Cardiovascular Causes, According to BMI.
Discussion
Our study confirms the findings of previous studies demonstrating that the relationship between death from any cause and BMI follows a J-shaped pattern.18 This curve reflects the association between BMI and the risk of death from all the major diseases. Among subjects with a low BMI, the increased risk was driven by respiratory and other causes, whereas among those with a high BMI, it was associated with cancer and cardiovascular diseases. Similar patterns were observed in smokers and those who reported never having smoked, implying that confounding by a history of smoking cannot explain the J-shaped relationship. This J-shaped risk relationship has been documented in several of the largest, but not all, cohorts.2,11 The patterns in the Korean, Chinese, and Western1,4 cohorts appear to be similar, suggesting that the risk of death associated with obesity among Asians is not apparent at lower BMI values, as compared with that among Western populations.
Because of the weight distribution of the subjects, the KCPS probably contains substantially more information about people with lower BMI values than do studies with Western cohorts. At a BMI of less than 18.5, hazard ratios for death were significantly increased, with the excess due, in part, to respiratory causes. In the Nurses' Health Study, Hu et al.5 showed that an increased risk in the leanest group was primarily due to an increase in COPD and cirrhosis. Other studies have also noted a substantially increased risk of death among subjects with a low BMI,1,2 although they did not provide information about the cause of death. He et al.10 identified an increased risk of death among underweight Chinese subjects, which persisted after the exclusion of subjects with baseline cardiovascular disease, cancer, renal disease, or COPD, and an increased risk of death during the first three years of follow-up. In the CPS II study, Calle et al.1 showed a greater increase in the risk of death associated with being underweight among those with a history of disease at enrollment than among those without such a history; relative risks in CPS II were similar to estimates in the KCPS for BMI values of less than 18.5. In NHANES,2 relative risks were higher overall (values were as high as 3.0 across the age and smoking strata), but analyses included subjects with coexisting illnesses and included deaths during the entire follow-up period.
To reduce the potential for attributing an excess risk of being underweight to weight loss associated with illness, we excluded from all analyses subjects reporting diagnoses of certain chronic diseases at enrollment, as well as during the first two years of follow-up, and we conducted sensitivity analyses that excluded the first five years of follow-up. Although this analytic strategy may be effective if the illness causing rapid weight loss leads to death, reverse causation may influence risk estimates if the disease course is lengthy and accompanied by weight loss.19 COPD has these characteristics, and subjects with more severe disease had greater weight loss over time.20 In several studies, after adjusting for lung function, the BMI remained a significant predictor of death.21,22 Thus, for COPD, the relationship between BMI and the risk of death may represent both reverse causation and a true causal role for body weight in determining prognosis. For tuberculosis, wasting at the time of diagnosis is a feature of the disease, and body weight predicts the short-term risk of death.23,24,25 The exclusion of the first two years of follow-up should address any acute contribution of active tuberculosis to body weight.
Since the distribution of respiratory causes of death may differ between Koreans and inhabitants of Western countries, it may not be possible to generalize our findings to other populations. Deaths from respiratory causes were due to tuberculosis in 19.3 percent of subjects, to pneumonia in 27.9 percent, to COPD in 27.8 percent, and to asthma in 24.7 percent.
For deaths from atherosclerotic cardiovascular disease, the hazard ratio increased steadily with increasing BMI, similar to the findings in a smaller cohort study of insured Koreans.26 Information on selected cardiovascular risk factors showed an increasingly unfavorable profile with increasing BMI, but these risk factors alone did not explain the excess risk of death from atherosclerotic cardiovascular causes associated with obesity. Although misclassification of cardiovascular risk factors, particularly those that vary during follow-up, could partially explain the persistent risk, the metabolic syndrome, sleep-disordered breathing, and other consequences of increased BMI are also likely to contribute to the risk associated with cardiovascular disease. The association of BMI with the risk of death from atherosclerotic cardiovascular causes was substantial, as has been shown in many other studies.27 A few large cohorts provided reasonably precise estimates of the risk of death from cardiovascular causes according to BMI. Similar, progressive increases in risk associated with BMI were seen in some studies, including the Nurses' Health Study,5,14 nonsmokers in the Physicians' Health Study,4 and the CPS I,28 but not in others, including CPS II,1 a U.K. cohort of particularly lean people,29 men with cardiovascular disease in the Physicians' Health Study,30 and a representative sample of Chinese men and women.11
The risk of death from cancer increased slightly among overweight men and women and more substantially among subjects with a BMI above 30.0 at enrollment; we observed no excess risk among subjects who had a low BMI at enrollment. In a meta-analysis by McGee,27 pooled estimates of the risk of death from cancer that compared obese subjects with those of normal weight were much smaller — 1.10 for women and 1.06 for men — than in our study. The Nurses' Health Study reported risks similar to those in our study, but risks in similar BMI strata were lower in CPS I and II.1,31 Distributions of deaths according to the type of the primary tumor differ between Korean and U.S. populations.32 Nonetheless, our findings indicate that obesity does contribute to Korea's cancer burden.
As has been reported in other populations, we found that the association between BMI and the risk of death varied according to age, with little evidence of increased risk among obese subjects over the age of 65 years. This effect modification has been the subject of controversy because BMI is less well correlated with adiposity in the elderly and because of the increased probability of undiagnosed diseases and survivor effects in this age group.33 Substantial interest exists, however, in conducting an estimation of the future burden of obesity as today's obese children and young adults grow older.2,34,35 In other studies, investigators showed that an older age at enrollment (the variable used in our analysis) attenuated the risk associated with obesity.28,36 In fact, in CPS I, overweight and obesity were not associated with an increased risk of death among subjects older than 85 years.28 In a recent report by Flegal et al.2 of follow-up data from NHANES I, II, and III, subjects in the oldest age group (70 years or older) who had a BMI of more than 25 were not at increased risk for death. Our evidence provides support for this modification of effect by age, but this effect has not been observed in all populations.4
The association of BMI in the overweight and obese range with an increased risk of death from atherosclerotic cardiovascular causes and cancer suggests that control of excess adiposity may reduce the two most important causes of death among Koreans. The inverse association between BMI and the risk of death from respiratory causes partly explains the J-shaped relationship between BMI and the risk of death from any cause, but further research is warranted to examine the extent of reverse causation and to consider the role of other causes of an increased risk of death in association with a low BMI.
In considering whether the findings of the KCPS can be applied to other populations, we recognize that Asian populations generally have a higher percentage of body fat than do Western populations at the same BMI level.7 In a meta-analysis of the predictive ability of BMI to estimate the percentage of body fat among various ethnic groups, Deurenberg et al.6 found that for the same percentage of body fat, BMI among subjects from various East Asian countries was lower by 1.9 to 3.2 than that of white subjects. Although contributing factors are not completely understood,7 Asians generally have a slighter body build than do whites, and slighter people tend to have less muscle mass and connective tissue.37 Consequently, the WHO recommends that cutoff values in the definition of overweight and obesity should be lower for Asian populations than for Western populations.7,38 Our observations may prove to be useful in the evaluation of this recommendation.
Supported by a grant (1R03 CA94771-02) from the National Cancer Institute, National Institutes of Health, Department of Health and Human Services, and by a grant (10526) from the Korean Seoul City Research and Business Development Program.
No potential conflict of interest relevant to this article was reported.
We are indebted to the staff of the Korean National Health Insurance Corporation, to Dr. Michael Thun for his helpful comments, and to Charlotte Gerczak for her editorial assistance.
Source Information
From the Department of Epidemiology and Health Promotion, Graduate School of Public Health, Yonsei University, Seoul (S.H.J., J.P.); the National Health Insurance Corporation, Seoul (S.-Y.L.); the Department of Health Policy and Management, College of Medicine, Cheju National University, Jeju (S.-Y.L.); the Department of Preventive Medicine and Public Health, Yonsei University College of Medicine, Seoul (H.O.) — all in Korea; and the Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore (S.H.J., J.W.S., E.G., J.M.S.).
Address reprint requests to Dr. Jee at the Department of Epidemiology and Health Promotion, Graduate School of Public Health, Yonsei University, Seoul, Korea, or at jsunha@yumc.yonsei.ac.kr.
References
Calle EE, Thun MJ, Petrelli JM, Rodriguez C, Heath CW Jr. Body-mass index and mortality in a prospective cohort of U.S. adults. N Engl J Med 1999;341:1097-1105.
Flegal KM, Graubard BI, Williamson DF, Gail MH. Excess deaths associated with underweight, overweight, and obesity. JAMA 2005;293:1861-1867.
Dorn JM, Schisterman EF, Winkelstein W Jr, Trevisan M. Body mass index and mortality in a general population sample of men and women. Am J Epidemiol 1997;146:919-931.
Ajani UA, Lotufo PA, Gaziano JM, et al. Body mass index and mortality among US male physicians. Ann Epidemiol 2004;14:731-739.
Hu FB, Willett WC, Li T, Stampfer MJ, Colditz GA, Manson JE. Adiposity as compared with physical activity in predicting mortality among women. N Engl J Med 2004;351:2694-2703.
Deurenberg P, Yap M, Van Staveren WA. Body mass index and percent body fat: a meta analysis among different ethnic groups. Int J Obes Relat Metab Disord 1998;22:1164-1171.
WHO Expert Consultation. Appropriate body-mass index for Asian populations and its implications for policy and intervention strategies. Lancet 2004;363:157-163.
Stevens J. Ethnic-specific revisions of body mass index cutoffs to define overweight and obesity in Asians are not warranted. Int J Obes Relat Metab Disord 2003;27:1297-1299.
Stevens J. BMI cutoffs for obesity should not vary by ethnic group. In: Madeiros-Neto G, Halpern A, Bouchard C, eds. Progress in obesity research. Monograph 9. Montrouge, France: John Libbey Eurotext, 2003:554-7.
He J, Gu D, Wu X, et al. Major causes of death among men and women in China. N Engl J Med 2005;353:1124-1134.
Gu D, He J, Duan X, et al. Body weight and mortality among men and women in China. JAMA 2006;295:776-783.
Jee SH, Ohrr H, Sull JW, Samet JM. Cigarette smoking, alcohol drinking, hepatitis B, and risk for hepatocellular carcinoma in Korea. J Natl Cancer Inst 2004;96:1851-1856.
Jee SH, Ohrr H, Sull JW, Yun JE, Ji M, Samet JM. Fasting serum glucose level and cancer risk in Korean men and women. JAMA 2005;293:194-202.
Manson JE, Willett WC, Stampfer MJ, et al. Body weight and mortality among women. N Engl J Med 1995;333:677-685.
Manson JE, Stampfer MJ, Hennekens CH, Willett WC. Body weight and longevity: a reassessment. JAMA 1987;257:353-358.
International classification of diseases and related health problems: manual of the International Statistical Classification of Diseases, Injuries, and Causes of Death. 10th ed. Geneva: World Health Organization, 1992.
Cox DR. Regression models and life-tables. J R Stat Soc 1972;34:187-202.
Katzmarzyk PT, Janssen I, Ardern CI. Physical inactivity, excess adiposity and premature mortality. Obes Rev 2003;4:257-290.
Allison DB, Heo M, Flanders DW, Faith MS, Williamson DF. Examination of "early mortality exclusion" as an approach to control for confounding by occult disease in epidemiologic studies of mortality risk factors. Am J Epidemiol 1997;146:672-680.
Prescott E, Almdal T, Mikkelsen KL, Tofteng CL, Vestbo J, Lange P. Prognostic value of weight change in chronic obstructive pulmonary disease: results from the Copenhagen City Heart Study. Eur Respir J 2002;20:539-544.
Wilson DO, Rogers RM, Wright EC, Anthonisen NR. Body weight in chronic obstructive pulmonary disease: the National Institutes of Health Intermittent Positive-Pressure Breathing Trial. Am Rev Respir Dis 1989;139:1435-1438.
Landbo C, Prescott E, Lange P, Vestbo J, Almdal TP. Prognostic value of nutritional status in chronic obstructive pulmonary disease. Am J Respir Crit Care Med 1999;160:1856-1861.
Macallan DC. Malnutrition in tuberculosis. Diagn Microbiol Infect Dis 1999;34:153-157.
Sacks LV, Pendle S. Factors related to in-hospital deaths in patients with tuberculosis. Arch Intern Med 1998;158:1916-1922.
Garcia-Garcia ML, Ponce-De-Leon A, Garcia-Sancho MC, et al. Tuberculosis-related deaths within a well-functioning DOTS control program. Emerg Infect Dis 2002;8:1327-1333.
Jee SH, Pastor-Barriuso R, Appel LJ, Suh I, Miller ER III, Guallar E. Body mass index and incident ischemic heart disease in South Korean men and women. Am J Epidemiol 2005;162:42-48.
McGee DL. Body mass index and mortality: a meta-analysis based on person-level data from twenty-six observational studies. Ann Epidemiol 2005;15:87-97.
Stevens J, Cai J, Pamuk ER, Williamson DF, Thun MJ, Wood JL. The effect of age on the association between body-mass index and mortality. N Engl J Med 1998;338:1-7.
Thorogood M, Appleby PN, Key TJ, Mann J. Relation between body mass index and mortality in an unusually slim cohort. J Epidemiol Community Health 2003;57:130-133.
Widlansky ME, Sesso HD, Rexrode KM, Manson JE, Gaziano JM. Body mass index and total and cardiovascular mortality in men with a history of cardiovascular disease. Arch Intern Med 2004;164:2326-2332.
Calle EE, Rodriguez C, Walker-Thurmond K, Thun MJ. Overweight, obesity, and mortality from cancer in a prospectively studied cohort of U.S. adults. N Engl J Med 2003;348:1625-1638.
Gomez SL, Le GM, Clarke CA, Glaser SL, France AM, West DW. Cancer incidence patterns in Koreans in the US and in Kangwha, South Korea. Cancer Causes Control 2003;14:167-174.
Villareal DT, Apovian CM, Kushner RF, Klein S. Obesity in older adults: technical review and position statement of the American Society for Nutrition and NAASO, The Obesity Society. Obes Res 2005;13:1849-1863.
Olshansky SJ, Passaro DJ, Hershow RC, et al. A potential decline in life expectancy in the United States in the 21st century. N Engl J Med 2005;352:1138-1145.
Institute of Medicine. Preventing childhood obesity: health in the balance. Washington, D.C.: National Academies Press, 2005.
Bender R, Jockel KH, Trautner C, Spraul M, Berger M. Effect of age on excess mortality in obesity. JAMA 1999;281:1498-1504.
Deurenberg P, Deurenberg YM, Wang J, Lin FP, Schmidt G. The impact of body build on the relationship between body mass index and percent body fat. Int J Obes Relat Metab Disord 1999;23:537-542.
World Health Organization, International Association for the Study of Obesity, International Obesity Task Force. The Asia-Pacific perspective: redefining obesity and its treatment. Sydney: Health Communications Australia, 2000.(Sun Ha Jee, Ph.D., Jae Wo)
Background Obesity is associated with diverse health risks, but the role of body weight as a risk factor for death remains controversial.
Methods We examined the association between body weight and the risk of death in a 12-year prospective cohort study of 1,213,829 Koreans between the ages of 30 and 95 years. We examined 82,372 deaths from any cause and 48,731 deaths from specific diseases (including 29,123 from cancer, 16,426 from atherosclerotic cardiovascular disease, and 3362 from respiratory disease) in relation to the body-mass index (BMI) (the weight in kilograms divided by the square of the height in meters).
Results In both sexes, the average baseline BMI was 23.2, and the rate of death from any cause had a J-shaped association with the BMI, regardless of cigarette-smoking history. The risk of death from any cause was lowest among patients with a BMI of 23.0 to 24.9. In all groups, the risk of death from respiratory causes was higher among subjects with a lower BMI, and the risk of death from atherosclerotic cardiovascular disease or cancer was higher among subjects with a higher BMI. The relative risk of death associated with BMI declined with increasing age.
Conclusions Underweight, overweight, and obese men and women had higher rates of death than men and women of normal weight. The association of BMI with death varied according to the cause of death and was modified by age, sex, and smoking history.
Although obesity is widely accepted as an important health risk, the optimal body-mass index (BMI) (the weight in kilograms divided by the square of the height in meters) and the effects of being either underweight or overweight on the risk of death are controversial. In the Cancer Prevention Study (CPS) II,1 sponsored by the American Cancer Society, the rate of death was lowest among men with a BMI of 23.5 to 24.9 and among women with a BMI of 22.0 to 23.4; above and below these levels, the risk of death increased. However, being overweight was not associated with an increased risk of death in the National Health and Nutrition Examination Survey (NHANES) I, II, or III.2 The results of other studies have been mixed3,4 and may reflect differences in age, the number or extent of coexisting illnesses, and BMI distributions among subjects, as well as in analytic approaches.1,5
Since studies of the association between BMI and death have been conducted primarily in Western populations, it is uncertain whether the findings of these studies can be applied to other groups. Continental Asian populations have a higher percentage of body fat for a given BMI than do whites,6 and a World Health Organization (WHO) Expert Consultation proposed a new BMI cutoff of 23.0 for public health action in Asia.7 The use of this cutoff, however, was not directly supported by data on mortality.8,9 Indeed, deaths from any cause were lowest among men with a BMI of 24.0 to 24.9 and women with a BMI of 25.0 to 26.9 in a representative group of Chinese subjects.10,11
We conducted a prospective cohort study of BMI and the risk of death from any cause and from specific diseases in more than 1 million Koreans in the Korean Cancer Prevention Study (KCPS).12,13
Methods
Study Population
We enrolled 1,329,525 Koreans between the ages of 30 and 95 years who had undergone one biennial medical evaluation through the National Health Insurance Corporation between 1992 and 1995.12,13 Of the subjects, 784,870 (59.0 percent) were enrolled in 1992, 367,903 (27.7 percent) in 1993, 98,417 (7.4 percent) in 1994, and 78,335 (5.9 percent) in 1995.
To avoid confounding of the association between BMI and the risk of death by preexisting disease,14,15 904 subjects who died before January 1, 1993, were excluded from the study, as were 87,911 subjects who reported having atherosclerotic cardiovascular disease, cancer, liver disease, diabetes, or a respiratory disease at or before the initial study visit. In addition, 26,881 subjects with missing information about BMI or alcohol intake and those with an extremely low BMI (less than 16.0) or short stature (1.30 m or less) were excluded. The final sample included 1,213,829 subjects.
Because the study involved data that were routinely collected, consent was not specifically obtained. The institutional review boards of Yonsei University and the Johns Hopkins Bloomberg School of Public Health approved the study.
Data Collection
Enrollees in the National Health Insurance Corporation undergo standardized examinations every two years at local hospitals. During visits between 1992 and 1995, subjects reported on their smoking habits and alcohol consumption, and weight and height measurements were recorded while subjects were wearing light clothing. Subjects were seated for blood-pressure measurement. Blood samples were obtained after an overnight fast for white-cell counts and clinical chemical analysis. Quality-control procedures were performed in accordance with the Korean Association of Laboratory Quality Control.
Follow-up and Outcome Classification
Deaths among subjects through December 31, 2004, were confirmed by matching the information to death records. Death certificates from the National Statistical Office were identified with the use of identification numbers assigned to subjects at birth. Abstractors coded causes of death according to the International Classification of Diseases, 10th Revision.16 We used underlying causes as reported.
We attempted to minimize the effect of existing medical conditions on the baseline BMI by excluding all events that occurred among subjects during the first two years of follow-up. Thus, general follow-up began on January 1 of the third year after the baseline visit. In a sensitivity analysis, we excluded the first five years of follow-up.
Statistical Analysis
Proportional-hazards models were used to evaluate the association between the baseline BMI and death.17 The BMI was categorized as less than 18.5, 18.5 to 19.9, 20.0 to 21.4, 21.5 to 22.9, 23.0 to 24.9, 25.0 to 26.4, 26.5 to 27.9, 28.0 to 29.9, 30 to 31.9, or 32.0 or more. Analyses were performed separately in men and women and were adjusted for the following covariates: age at enrollment (continuous variable), alcohol intake (five categories based on grams consumed per day: 0, 1 to 24 g, 25 to 49 g, 50 to 99 g, and 100 g or more), and participation in regular physical activity (yes or no). Because the proportion of women who reported having smoked cigarettes was small, analyses in women were restricted to those who reported never having smoked. Analyses of men who reported having smoked were adjusted for smoking status (never smoked, former smoker, or current smoker) and the number of cigarettes smoked daily among current smokers (1 to 9, 10 to 19, and 20 or more).
Modification of the effect of BMI was assessed by the inclusion of interaction terms of BMI category indicators with indicator variables for sex, age (three categories), and smoking history (two categories, one consisting of current and former smokers and one of lifetime nonsmokers). All analyses were conducted with the use of SAS software, version 9 (SAS Institute).
Results
The average BMI was 23.2 for both sexes (Table 1), and the majority of subjects had a BMI below 25.0. The BMI was below 18.5 in 2.2 percent of men and 4.7 percent of women; above 25.0 in 23.8 percent and 26.8 percent, respectively; and above 30.0 in 0.8 percent and 2.4 percent, respectively. Systolic blood pressure, total serum cholesterol, fasting serum glucose level, and white-cell count had strong, progressive associations with increasing BMIs (Table 2).
Table 1. Baseline Characteristics of the Study Population.
Table 2. Differences in Baseline Systolic Blood Pressure and Clinical Chemical Analyses, According to BMI.
During follow-up, 58,312 men died (including 22,249 from cancer, 10,486 from atherosclerotic cardiovascular causes, and 2442 from respiratory causes) and 24,060 women died (including 6874 from cancer, 5940 from atherosclerotic cardiovascular causes, and 920 from respiratory causes). The shape of the curve showing the association between BMI and the risk of death from any cause was similar in men, regardless of their smoking history, and in women who reported never having smoked (Figure 1A). The hazard ratio was higher at the lowest and highest BMI values.
Figure 1. Hazard Ratios for Death from Any Cause and from Any Cause Except Respiratory, According to BMI and Smoking History.
Data are from the KCPS, 1993–2004.12,13 The reference category was a BMI of 23.0 to 24.9. Results for men who reported having smoked cigarettes were further adjusted for whether the subject was a former smoker or a current smoker and the number of cigarettes smoked per day (1 to 9, 10 to 19, and 20 or more). All hazard ratios were adjusted for age.
Men with a BMI of 23.0 to 24.9 who reported never having smoked had the lowest risk of death from any cause (Table 1 of the Supplementary Appendix, available with the full text of this article at www.nejm.org). As compared with men with a BMI of 23.0 to 24.9, men who reported never having smoked had a hazard ratio for death from any cause of 1.29 (95 percent confidence interval, 1.15 to 1.44) in association with a BMI of less than 18.5, a hazard ratio of 1.04 (95 percent confidence interval, 0.98 to 1.10) in association with a BMI of 25.0 to 29.9, and a hazard ratio of 1.71 (95 percent confidence interval, 1.44 to 2.03) in association with a BMI of 30.0 or more. As compared with men with a BMI of 23.0 to 24.9, men who reported having smoked had a hazard ratio for death from any cause of 1.36 (95 percent confidence interval, 1.30 to 1.42) in association with a BMI of less than 18.5, a hazard ratio of 0.98 (95 percent confidence interval, 0.95 to 1.01) in association with a BMI of 25.0 to 29.9, and a hazard ratio of 1.31 (95 percent confidence interval, 1.18 to 1.45) in association with a BMI of 30.0 or more.
Among women (with the analysis restricted to those who reported never having smoked), the risk of death from any cause was lowest for those with a BMI of 23.0 to 24.9 and similar for those with a BMI of 20.0 to 26.4 (Figure 1A, and Table 2 of the Supplementary Appendix). As compared with women with a BMI of 23.0 to 24.9, women with a BMI of less than 18.5 had a hazard ratio for death from any cause of 1.17 (95 percent confidence interval, 1.09 to 1.26), women with a BMI of 25.0 to 29.9 had a hazard ratio of 1.04 (95 percent confidence interval, 1.00 to 1.08), and women with a BMI of 30.0 or more had a hazard ratio of 1.20 (95 percent confidence interval, 1.10 to 1.30).
The association between BMI and the risk of death according to the cause of death had a similar pattern of variation for both sexes (Figure 2, and Table 1 and Table 2 of the Supplementary Appendix). The risk of death from respiratory causes decreased progressively with increasing BMI, whereas the risk of death from atherosclerotic cardiovascular causes increased steadily with increasing BMI. The risk of death from cancer increased at a BMI above 26.0 to 28.0. Deaths from respiratory causes explained some of the increase in the risk of death at a low BMI (Figure 1B). For deaths associated with lung disease, the association between BMI and the risk of death was similar for the major categories of pulmonary illnesses, including tuberculosis, chronic obstructive pulmonary disease (COPD), asthma, and pneumonia; the association persisted after the exclusion of the first five years of follow-up. With this exclusion, the hazard ratio for death from respiratory causes that was associated with a decrease in BMI of 1.0 was 1.26 (95 percent confidence interval, 1.20 to 1.31) among men who reported never having smoked, 1.25 (95 percent confidence interval, 1.22 to 1.27) among men who reported having smoked, and 1.08 (95 percent confidence interval, 1.05 to 1.12) among women.
Figure 2. Hazard Ratios for Death from Cancer and from Atherosclerotic Cardiovascular and Respiratory Causes, According to BMI and Smoking History.
Data are from the KCPS, 1993–2004.12,13 The reference category was a BMI of 23.0 to 24.9. Results for men who reported having smoked cigarettes were further adjusted for whether the subject was a former smoker or a current smoker and the number of cigarettes smoked per day (1 to 9, 10 to 19, and 20 or more). The number of deaths from respiratory causes among subjects with a BMI of 30.0 or more was too small to yield a reliable estimate of relative risks. All hazard ratios were adjusted for age. Panels A, B, and C have different scales for hazard ratios in the vertical axes.
The relative increase in the risk of death from any cause that was associated with a high BMI was dependent on age (Figure 3). For both sexes, the highest relative risks associated with a high BMI were observed among subjects younger than 50 years of age. An increase in BMI to more than 25.0 was not associated with an increased risk of death from any cause among men or women who were 65 years or older at baseline. The interaction between BMI and age was significant (P<0.001), as were interactions between BMI and sex and BMI and smoking history (P<0.001 for both).
Figure 3. Hazard Ratios for Death from Any Cause among Men and Women with No History of Smoking, According to Age Group and BMI.
Data are from the KCPS, 1993–2004.12,13 The reference category was a BMI of 23.0 to 24.9. All hazard ratios were adjusted for age. Panel A and Panel B have different scales for hazard ratios.
We explored whether the association between BMI and the risk of death from atherosclerotic cardiovascular disease could be explained by systolic blood pressure or levels of blood glucose or cholesterol. As expected, adjustment for these factors attenuated this association (Table 3). Analyses of death from any cause, from cancer, and from respiratory causes adjusted for risk factors are shown in Tables 3, 4, and 5, respectively, of the Supplementary Appendix.
Table 3. Hazard Ratios for Death from Atherosclerotic Cardiovascular Causes, According to BMI.
Discussion
Our study confirms the findings of previous studies demonstrating that the relationship between death from any cause and BMI follows a J-shaped pattern.18 This curve reflects the association between BMI and the risk of death from all the major diseases. Among subjects with a low BMI, the increased risk was driven by respiratory and other causes, whereas among those with a high BMI, it was associated with cancer and cardiovascular diseases. Similar patterns were observed in smokers and those who reported never having smoked, implying that confounding by a history of smoking cannot explain the J-shaped relationship. This J-shaped risk relationship has been documented in several of the largest, but not all, cohorts.2,11 The patterns in the Korean, Chinese, and Western1,4 cohorts appear to be similar, suggesting that the risk of death associated with obesity among Asians is not apparent at lower BMI values, as compared with that among Western populations.
Because of the weight distribution of the subjects, the KCPS probably contains substantially more information about people with lower BMI values than do studies with Western cohorts. At a BMI of less than 18.5, hazard ratios for death were significantly increased, with the excess due, in part, to respiratory causes. In the Nurses' Health Study, Hu et al.5 showed that an increased risk in the leanest group was primarily due to an increase in COPD and cirrhosis. Other studies have also noted a substantially increased risk of death among subjects with a low BMI,1,2 although they did not provide information about the cause of death. He et al.10 identified an increased risk of death among underweight Chinese subjects, which persisted after the exclusion of subjects with baseline cardiovascular disease, cancer, renal disease, or COPD, and an increased risk of death during the first three years of follow-up. In the CPS II study, Calle et al.1 showed a greater increase in the risk of death associated with being underweight among those with a history of disease at enrollment than among those without such a history; relative risks in CPS II were similar to estimates in the KCPS for BMI values of less than 18.5. In NHANES,2 relative risks were higher overall (values were as high as 3.0 across the age and smoking strata), but analyses included subjects with coexisting illnesses and included deaths during the entire follow-up period.
To reduce the potential for attributing an excess risk of being underweight to weight loss associated with illness, we excluded from all analyses subjects reporting diagnoses of certain chronic diseases at enrollment, as well as during the first two years of follow-up, and we conducted sensitivity analyses that excluded the first five years of follow-up. Although this analytic strategy may be effective if the illness causing rapid weight loss leads to death, reverse causation may influence risk estimates if the disease course is lengthy and accompanied by weight loss.19 COPD has these characteristics, and subjects with more severe disease had greater weight loss over time.20 In several studies, after adjusting for lung function, the BMI remained a significant predictor of death.21,22 Thus, for COPD, the relationship between BMI and the risk of death may represent both reverse causation and a true causal role for body weight in determining prognosis. For tuberculosis, wasting at the time of diagnosis is a feature of the disease, and body weight predicts the short-term risk of death.23,24,25 The exclusion of the first two years of follow-up should address any acute contribution of active tuberculosis to body weight.
Since the distribution of respiratory causes of death may differ between Koreans and inhabitants of Western countries, it may not be possible to generalize our findings to other populations. Deaths from respiratory causes were due to tuberculosis in 19.3 percent of subjects, to pneumonia in 27.9 percent, to COPD in 27.8 percent, and to asthma in 24.7 percent.
For deaths from atherosclerotic cardiovascular disease, the hazard ratio increased steadily with increasing BMI, similar to the findings in a smaller cohort study of insured Koreans.26 Information on selected cardiovascular risk factors showed an increasingly unfavorable profile with increasing BMI, but these risk factors alone did not explain the excess risk of death from atherosclerotic cardiovascular causes associated with obesity. Although misclassification of cardiovascular risk factors, particularly those that vary during follow-up, could partially explain the persistent risk, the metabolic syndrome, sleep-disordered breathing, and other consequences of increased BMI are also likely to contribute to the risk associated with cardiovascular disease. The association of BMI with the risk of death from atherosclerotic cardiovascular causes was substantial, as has been shown in many other studies.27 A few large cohorts provided reasonably precise estimates of the risk of death from cardiovascular causes according to BMI. Similar, progressive increases in risk associated with BMI were seen in some studies, including the Nurses' Health Study,5,14 nonsmokers in the Physicians' Health Study,4 and the CPS I,28 but not in others, including CPS II,1 a U.K. cohort of particularly lean people,29 men with cardiovascular disease in the Physicians' Health Study,30 and a representative sample of Chinese men and women.11
The risk of death from cancer increased slightly among overweight men and women and more substantially among subjects with a BMI above 30.0 at enrollment; we observed no excess risk among subjects who had a low BMI at enrollment. In a meta-analysis by McGee,27 pooled estimates of the risk of death from cancer that compared obese subjects with those of normal weight were much smaller — 1.10 for women and 1.06 for men — than in our study. The Nurses' Health Study reported risks similar to those in our study, but risks in similar BMI strata were lower in CPS I and II.1,31 Distributions of deaths according to the type of the primary tumor differ between Korean and U.S. populations.32 Nonetheless, our findings indicate that obesity does contribute to Korea's cancer burden.
As has been reported in other populations, we found that the association between BMI and the risk of death varied according to age, with little evidence of increased risk among obese subjects over the age of 65 years. This effect modification has been the subject of controversy because BMI is less well correlated with adiposity in the elderly and because of the increased probability of undiagnosed diseases and survivor effects in this age group.33 Substantial interest exists, however, in conducting an estimation of the future burden of obesity as today's obese children and young adults grow older.2,34,35 In other studies, investigators showed that an older age at enrollment (the variable used in our analysis) attenuated the risk associated with obesity.28,36 In fact, in CPS I, overweight and obesity were not associated with an increased risk of death among subjects older than 85 years.28 In a recent report by Flegal et al.2 of follow-up data from NHANES I, II, and III, subjects in the oldest age group (70 years or older) who had a BMI of more than 25 were not at increased risk for death. Our evidence provides support for this modification of effect by age, but this effect has not been observed in all populations.4
The association of BMI in the overweight and obese range with an increased risk of death from atherosclerotic cardiovascular causes and cancer suggests that control of excess adiposity may reduce the two most important causes of death among Koreans. The inverse association between BMI and the risk of death from respiratory causes partly explains the J-shaped relationship between BMI and the risk of death from any cause, but further research is warranted to examine the extent of reverse causation and to consider the role of other causes of an increased risk of death in association with a low BMI.
In considering whether the findings of the KCPS can be applied to other populations, we recognize that Asian populations generally have a higher percentage of body fat than do Western populations at the same BMI level.7 In a meta-analysis of the predictive ability of BMI to estimate the percentage of body fat among various ethnic groups, Deurenberg et al.6 found that for the same percentage of body fat, BMI among subjects from various East Asian countries was lower by 1.9 to 3.2 than that of white subjects. Although contributing factors are not completely understood,7 Asians generally have a slighter body build than do whites, and slighter people tend to have less muscle mass and connective tissue.37 Consequently, the WHO recommends that cutoff values in the definition of overweight and obesity should be lower for Asian populations than for Western populations.7,38 Our observations may prove to be useful in the evaluation of this recommendation.
Supported by a grant (1R03 CA94771-02) from the National Cancer Institute, National Institutes of Health, Department of Health and Human Services, and by a grant (10526) from the Korean Seoul City Research and Business Development Program.
No potential conflict of interest relevant to this article was reported.
We are indebted to the staff of the Korean National Health Insurance Corporation, to Dr. Michael Thun for his helpful comments, and to Charlotte Gerczak for her editorial assistance.
Source Information
From the Department of Epidemiology and Health Promotion, Graduate School of Public Health, Yonsei University, Seoul (S.H.J., J.P.); the National Health Insurance Corporation, Seoul (S.-Y.L.); the Department of Health Policy and Management, College of Medicine, Cheju National University, Jeju (S.-Y.L.); the Department of Preventive Medicine and Public Health, Yonsei University College of Medicine, Seoul (H.O.) — all in Korea; and the Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore (S.H.J., J.W.S., E.G., J.M.S.).
Address reprint requests to Dr. Jee at the Department of Epidemiology and Health Promotion, Graduate School of Public Health, Yonsei University, Seoul, Korea, or at jsunha@yumc.yonsei.ac.kr.
References
Calle EE, Thun MJ, Petrelli JM, Rodriguez C, Heath CW Jr. Body-mass index and mortality in a prospective cohort of U.S. adults. N Engl J Med 1999;341:1097-1105.
Flegal KM, Graubard BI, Williamson DF, Gail MH. Excess deaths associated with underweight, overweight, and obesity. JAMA 2005;293:1861-1867.
Dorn JM, Schisterman EF, Winkelstein W Jr, Trevisan M. Body mass index and mortality in a general population sample of men and women. Am J Epidemiol 1997;146:919-931.
Ajani UA, Lotufo PA, Gaziano JM, et al. Body mass index and mortality among US male physicians. Ann Epidemiol 2004;14:731-739.
Hu FB, Willett WC, Li T, Stampfer MJ, Colditz GA, Manson JE. Adiposity as compared with physical activity in predicting mortality among women. N Engl J Med 2004;351:2694-2703.
Deurenberg P, Yap M, Van Staveren WA. Body mass index and percent body fat: a meta analysis among different ethnic groups. Int J Obes Relat Metab Disord 1998;22:1164-1171.
WHO Expert Consultation. Appropriate body-mass index for Asian populations and its implications for policy and intervention strategies. Lancet 2004;363:157-163.
Stevens J. Ethnic-specific revisions of body mass index cutoffs to define overweight and obesity in Asians are not warranted. Int J Obes Relat Metab Disord 2003;27:1297-1299.
Stevens J. BMI cutoffs for obesity should not vary by ethnic group. In: Madeiros-Neto G, Halpern A, Bouchard C, eds. Progress in obesity research. Monograph 9. Montrouge, France: John Libbey Eurotext, 2003:554-7.
He J, Gu D, Wu X, et al. Major causes of death among men and women in China. N Engl J Med 2005;353:1124-1134.
Gu D, He J, Duan X, et al. Body weight and mortality among men and women in China. JAMA 2006;295:776-783.
Jee SH, Ohrr H, Sull JW, Samet JM. Cigarette smoking, alcohol drinking, hepatitis B, and risk for hepatocellular carcinoma in Korea. J Natl Cancer Inst 2004;96:1851-1856.
Jee SH, Ohrr H, Sull JW, Yun JE, Ji M, Samet JM. Fasting serum glucose level and cancer risk in Korean men and women. JAMA 2005;293:194-202.
Manson JE, Willett WC, Stampfer MJ, et al. Body weight and mortality among women. N Engl J Med 1995;333:677-685.
Manson JE, Stampfer MJ, Hennekens CH, Willett WC. Body weight and longevity: a reassessment. JAMA 1987;257:353-358.
International classification of diseases and related health problems: manual of the International Statistical Classification of Diseases, Injuries, and Causes of Death. 10th ed. Geneva: World Health Organization, 1992.
Cox DR. Regression models and life-tables. J R Stat Soc 1972;34:187-202.
Katzmarzyk PT, Janssen I, Ardern CI. Physical inactivity, excess adiposity and premature mortality. Obes Rev 2003;4:257-290.
Allison DB, Heo M, Flanders DW, Faith MS, Williamson DF. Examination of "early mortality exclusion" as an approach to control for confounding by occult disease in epidemiologic studies of mortality risk factors. Am J Epidemiol 1997;146:672-680.
Prescott E, Almdal T, Mikkelsen KL, Tofteng CL, Vestbo J, Lange P. Prognostic value of weight change in chronic obstructive pulmonary disease: results from the Copenhagen City Heart Study. Eur Respir J 2002;20:539-544.
Wilson DO, Rogers RM, Wright EC, Anthonisen NR. Body weight in chronic obstructive pulmonary disease: the National Institutes of Health Intermittent Positive-Pressure Breathing Trial. Am Rev Respir Dis 1989;139:1435-1438.
Landbo C, Prescott E, Lange P, Vestbo J, Almdal TP. Prognostic value of nutritional status in chronic obstructive pulmonary disease. Am J Respir Crit Care Med 1999;160:1856-1861.
Macallan DC. Malnutrition in tuberculosis. Diagn Microbiol Infect Dis 1999;34:153-157.
Sacks LV, Pendle S. Factors related to in-hospital deaths in patients with tuberculosis. Arch Intern Med 1998;158:1916-1922.
Garcia-Garcia ML, Ponce-De-Leon A, Garcia-Sancho MC, et al. Tuberculosis-related deaths within a well-functioning DOTS control program. Emerg Infect Dis 2002;8:1327-1333.
Jee SH, Pastor-Barriuso R, Appel LJ, Suh I, Miller ER III, Guallar E. Body mass index and incident ischemic heart disease in South Korean men and women. Am J Epidemiol 2005;162:42-48.
McGee DL. Body mass index and mortality: a meta-analysis based on person-level data from twenty-six observational studies. Ann Epidemiol 2005;15:87-97.
Stevens J, Cai J, Pamuk ER, Williamson DF, Thun MJ, Wood JL. The effect of age on the association between body-mass index and mortality. N Engl J Med 1998;338:1-7.
Thorogood M, Appleby PN, Key TJ, Mann J. Relation between body mass index and mortality in an unusually slim cohort. J Epidemiol Community Health 2003;57:130-133.
Widlansky ME, Sesso HD, Rexrode KM, Manson JE, Gaziano JM. Body mass index and total and cardiovascular mortality in men with a history of cardiovascular disease. Arch Intern Med 2004;164:2326-2332.
Calle EE, Rodriguez C, Walker-Thurmond K, Thun MJ. Overweight, obesity, and mortality from cancer in a prospectively studied cohort of U.S. adults. N Engl J Med 2003;348:1625-1638.
Gomez SL, Le GM, Clarke CA, Glaser SL, France AM, West DW. Cancer incidence patterns in Koreans in the US and in Kangwha, South Korea. Cancer Causes Control 2003;14:167-174.
Villareal DT, Apovian CM, Kushner RF, Klein S. Obesity in older adults: technical review and position statement of the American Society for Nutrition and NAASO, The Obesity Society. Obes Res 2005;13:1849-1863.
Olshansky SJ, Passaro DJ, Hershow RC, et al. A potential decline in life expectancy in the United States in the 21st century. N Engl J Med 2005;352:1138-1145.
Institute of Medicine. Preventing childhood obesity: health in the balance. Washington, D.C.: National Academies Press, 2005.
Bender R, Jockel KH, Trautner C, Spraul M, Berger M. Effect of age on excess mortality in obesity. JAMA 1999;281:1498-1504.
Deurenberg P, Deurenberg YM, Wang J, Lin FP, Schmidt G. The impact of body build on the relationship between body mass index and percent body fat. Int J Obes Relat Metab Disord 1999;23:537-542.
World Health Organization, International Association for the Study of Obesity, International Obesity Task Force. The Asia-Pacific perspective: redefining obesity and its treatment. Sydney: Health Communications Australia, 2000.(Sun Ha Jee, Ph.D., Jae Wo)