Overweight, Obesity, and Mortality
http://www.100md.com
《新英格兰医药杂志》
To the Editor: Adams et al. (Aug. 24 issue)1 state that overweight, defined by a body-mass index (BMI) (the weight in kilograms divided by the square of the height in meters) of 25.0 to 29.9 during midlife, is associated with an increased risk of death. The National Institutes of Health (NIH)–AARP Diet and Health Study cohort consists mainly of white, highly educated people. The response rate to the study questionnaire was only 18%, and even fewer people responded to a second questionnaire about midlife weight. The midlife BMI was calculated on the basis of recalled weight at 50 years of age; this calculation results in an "immortality bias," since respondents could not have died between 50 years of age and enrollment. This phenomenon, together with the very high rate of nonresponse of AARP members, leads to multiple selection biases. The results are different from those in previous studies because, we would speculate, of the use of a BMI of 23.5 to 24.9 for men and women in the reference group. The use of a broader BMI range for the reference group, as in previous studies with higher response rates,2 would show that overweight has little or no effect on mortality. Because of these shortcomings of the NIH–AARP study, it is too early to accept that overweight with a BMI of less than 30.0 carries an appreciable risk of death.
Cathelijne W.Y. Appels, M.D.
Jan P. Vandenbroucke, M.D.
Leiden University Medical Center
2300 RC Leiden, the Netherlands
c.w.y.appels@lumc.nl
References
Adams KF, Schatzkin A, Harris TB, et al. Overweight, obesity, and mortality in a large prospective cohort of persons 50 to 71 years old. N Engl J Med 2006;355:763-778.
Flegal KM, Graubard BI, Williamson DF, Gail MH. Excess deaths associated with underweight, overweight, and obesity. JAMA 2005;293:1861-1867.
To the Editor: Adams et al. analyzed 10-year mortality rates in a large cohort according to the BMI as calculated on the basis of self-reported weights, and they conclude that overweight during midlife is associated with an increased risk of death. This conclusion, however, is based on a post hoc analysis of the subgroup of persons who had never smoked cigarettes — about 30% of the total. In the entire cohort, the lowest relative risk of death was among persons with a BMI between 21.0 and 29.9, and the relative risk of death among persons with a BMI of 18.5 to 20.9 (lean normal) was similar to that among persons with a BMI of 35.0 to 39.9 (severe obesity). The unique findings in the "never smoked" group are unexplained. Furthermore, the use of relative risks and death rates per 100,000 person-years gives an inflated picture of the differences: the absolute increase in the risk of death over a period of 10 years among men who never smoked was about 0.06%. A more appropriate conclusion is that excess body weight is associated with a minimally increased risk of death among overweight persons who have never smoked.
Jay H. Hoofnagle, M.D.
National Institutes of Health
Bethesda, MD 20892
hoofnaglej@extra.niddk.nih.gov
To the Editor: Adams et al. present a general observation that makes intuitive sense. It is surprising, however, that there is no validation of the basic data points — self-reported height and weight — used to calculate the BMI in this massive study. The general clinical experience is that people overestimate their height and underestimate their weight, resulting in some uncertainty in the validity of the actual data.
Furthermore, the BMI is a statistical calculation for the analysis of obesity, but even when the BMI is determined on the basis of actual measurements of height and weight,1 it does not account for variability in the relative amounts of muscle and fat among individual subjects. Our patients are never statistical averages. Clinicians have seen compulsive exercisers with a BMI of 33.0 who do not have any subcutaneous or abdominal fat and who do not fit the statistical mold. Therefore, when treating individual patients, clinicians should correlate their subcutaneous and abdominal fat with the BMI, keeping in mind these limitations. In his accompanying Perspective article, Byers's description of his own BMI is not informative with regard to the ratio of muscle and fat.2
Uriel S. Barzel, M.D.
Montefiore Medical Center
Bronx, NY 10467
ubarzel@montefiore.org
References
Jee SH, Sull JW, Park J, et al. Body-mass index and mortality in Korean men and women. N Engl J Med 2006;355:779-787.
Byers T. Overweight and mortality among baby boomers -- now we're getting personal. N Engl J Med 2006;355:758-760.
To the Editor: Adams et al. assert in their concluding paragraph that many of the study participants were "from the baby-boomer generation." In the same paragraph, the authors state that the participants were 50 to 71 years old at baseline, which was 1995–1996. Thus, these participants must have been born between 1924 and 1945. According to the Merriam-Webster Dictionary, baby boomers were born after the end of World War II — that is, after August 1945.1 It would appear, then, that few of the study participants were baby boomers.
The misrepresentation becomes the headline in an accompanying Perspective article by Byers: "Overweight and Mortality among Baby Boomers — Now We're Getting Personal." Well, not that personal. According to his article, Byers is 57 years old; thus, he was born in 1948 or 1949. Boomer, yes. But he's younger than anyone in the study.
John Spitzer, Ph.D.
San Francisco Conservatory of Music
San Francisco, CA 94102
jspitzer@sfcm.edu
References
Merriam-Webster online dictionary. (Accessed November 30, 2006, at http://www.m-w.com/.)
The authors reply: As Appels and Vandenbroucke imply, our cohort does not constitute a random sample of AARP members or of the population in the United States. However, the association between the BMI and mortality among those choosing not to complete the initial questionnaire probably did not differ qualitatively from that observed in our cohort. Moreover, our findings are generalizable to populations other than our mainly "white, highly educated" cohort — in the same sense that the smoking–lung cancer association among white, male, British physicians1 is virtually universally applicable. We acknowledge that some AARP members would have died after 50 years of age and before they could have entered the study in 1995–1996, and we of course did not include their "immortal time"2 in our analysis. There is no compelling reason why the association between BMI and mortality among these "early dying" persons would have differed enough to offset our observed association between adiposity and mortality. Our reference group reflects the nadir of the BMI–mortality curve, avoids the inclusion of persons with preexisting disease who have a lower BMI, and facilitates comparison with previous research.3
Our analyses were not post hoc, as Hoofnagle suggests. They were conceived a priori to address the potentially confounding effects of smoking and preexisting disease and thereby best capture the underlying biology of adiposity and mortality. Among the overweight men who had never smoked, the fraction of premature deaths that could have been prevented by elimination of overweight (the etiologic fraction)4 was a nontrivial 19.2%.
Regarding Barzel's concerns, we note that the correlation between BMI based on self-reported height and weight and that based on measured height and weight is high, typically exceeding 0.9. Therefore, the effect of measurement error in our study is probably negligible. We agree that studies incorporating measures distinguishing between fat and lean mass as well as reflecting body-fat distribution would complement investigations based on BMI. There may be unusually physically active persons with a BMI of 33.0, but they are uncommon. In our study, only 2.7% of men and women both were highly physically active and had high BMIs.
Spitzer correctly states that one common definition has the baby-boomer generation starting in 1945. Alternatively, some authors define baby boomers as those born as early as 1943,5 giving our cohort more than 35,000 boomers. The main point, though, is that our investigation is eminently relevant to baby boomers, whatever their present age, as Byers concluded in his Perspective article.
Kenneth F. Adams, Ph.D.
Arthur Schatzkin, M.D.
Michael F. Leitzmann, M.D.
National Cancer Institute
Rockville, MD 20850
adamske@mail.nih.gov
References
Doll R, Hill AB. Mortality in relation to smoking: ten years' observations of British doctors. BMJ 1964;1:1399-410, 1460.
Rothman KJ. Modern epidemiology. Boston: Little, Brown, 1986.
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.
Miettinen OS. Proportion of disease caused or prevented by a given exposure, trait or intervention. Am J Epidemiol 1974;99:325-332.
Strauss W, Howe N. Generations: the history of America's future, 1584 to 2069. New York: Morrow, 1991.
Cathelijne W.Y. Appels, M.D.
Jan P. Vandenbroucke, M.D.
Leiden University Medical Center
2300 RC Leiden, the Netherlands
c.w.y.appels@lumc.nl
References
Adams KF, Schatzkin A, Harris TB, et al. Overweight, obesity, and mortality in a large prospective cohort of persons 50 to 71 years old. N Engl J Med 2006;355:763-778.
Flegal KM, Graubard BI, Williamson DF, Gail MH. Excess deaths associated with underweight, overweight, and obesity. JAMA 2005;293:1861-1867.
To the Editor: Adams et al. analyzed 10-year mortality rates in a large cohort according to the BMI as calculated on the basis of self-reported weights, and they conclude that overweight during midlife is associated with an increased risk of death. This conclusion, however, is based on a post hoc analysis of the subgroup of persons who had never smoked cigarettes — about 30% of the total. In the entire cohort, the lowest relative risk of death was among persons with a BMI between 21.0 and 29.9, and the relative risk of death among persons with a BMI of 18.5 to 20.9 (lean normal) was similar to that among persons with a BMI of 35.0 to 39.9 (severe obesity). The unique findings in the "never smoked" group are unexplained. Furthermore, the use of relative risks and death rates per 100,000 person-years gives an inflated picture of the differences: the absolute increase in the risk of death over a period of 10 years among men who never smoked was about 0.06%. A more appropriate conclusion is that excess body weight is associated with a minimally increased risk of death among overweight persons who have never smoked.
Jay H. Hoofnagle, M.D.
National Institutes of Health
Bethesda, MD 20892
hoofnaglej@extra.niddk.nih.gov
To the Editor: Adams et al. present a general observation that makes intuitive sense. It is surprising, however, that there is no validation of the basic data points — self-reported height and weight — used to calculate the BMI in this massive study. The general clinical experience is that people overestimate their height and underestimate their weight, resulting in some uncertainty in the validity of the actual data.
Furthermore, the BMI is a statistical calculation for the analysis of obesity, but even when the BMI is determined on the basis of actual measurements of height and weight,1 it does not account for variability in the relative amounts of muscle and fat among individual subjects. Our patients are never statistical averages. Clinicians have seen compulsive exercisers with a BMI of 33.0 who do not have any subcutaneous or abdominal fat and who do not fit the statistical mold. Therefore, when treating individual patients, clinicians should correlate their subcutaneous and abdominal fat with the BMI, keeping in mind these limitations. In his accompanying Perspective article, Byers's description of his own BMI is not informative with regard to the ratio of muscle and fat.2
Uriel S. Barzel, M.D.
Montefiore Medical Center
Bronx, NY 10467
ubarzel@montefiore.org
References
Jee SH, Sull JW, Park J, et al. Body-mass index and mortality in Korean men and women. N Engl J Med 2006;355:779-787.
Byers T. Overweight and mortality among baby boomers -- now we're getting personal. N Engl J Med 2006;355:758-760.
To the Editor: Adams et al. assert in their concluding paragraph that many of the study participants were "from the baby-boomer generation." In the same paragraph, the authors state that the participants were 50 to 71 years old at baseline, which was 1995–1996. Thus, these participants must have been born between 1924 and 1945. According to the Merriam-Webster Dictionary, baby boomers were born after the end of World War II — that is, after August 1945.1 It would appear, then, that few of the study participants were baby boomers.
The misrepresentation becomes the headline in an accompanying Perspective article by Byers: "Overweight and Mortality among Baby Boomers — Now We're Getting Personal." Well, not that personal. According to his article, Byers is 57 years old; thus, he was born in 1948 or 1949. Boomer, yes. But he's younger than anyone in the study.
John Spitzer, Ph.D.
San Francisco Conservatory of Music
San Francisco, CA 94102
jspitzer@sfcm.edu
References
Merriam-Webster online dictionary. (Accessed November 30, 2006, at http://www.m-w.com/.)
The authors reply: As Appels and Vandenbroucke imply, our cohort does not constitute a random sample of AARP members or of the population in the United States. However, the association between the BMI and mortality among those choosing not to complete the initial questionnaire probably did not differ qualitatively from that observed in our cohort. Moreover, our findings are generalizable to populations other than our mainly "white, highly educated" cohort — in the same sense that the smoking–lung cancer association among white, male, British physicians1 is virtually universally applicable. We acknowledge that some AARP members would have died after 50 years of age and before they could have entered the study in 1995–1996, and we of course did not include their "immortal time"2 in our analysis. There is no compelling reason why the association between BMI and mortality among these "early dying" persons would have differed enough to offset our observed association between adiposity and mortality. Our reference group reflects the nadir of the BMI–mortality curve, avoids the inclusion of persons with preexisting disease who have a lower BMI, and facilitates comparison with previous research.3
Our analyses were not post hoc, as Hoofnagle suggests. They were conceived a priori to address the potentially confounding effects of smoking and preexisting disease and thereby best capture the underlying biology of adiposity and mortality. Among the overweight men who had never smoked, the fraction of premature deaths that could have been prevented by elimination of overweight (the etiologic fraction)4 was a nontrivial 19.2%.
Regarding Barzel's concerns, we note that the correlation between BMI based on self-reported height and weight and that based on measured height and weight is high, typically exceeding 0.9. Therefore, the effect of measurement error in our study is probably negligible. We agree that studies incorporating measures distinguishing between fat and lean mass as well as reflecting body-fat distribution would complement investigations based on BMI. There may be unusually physically active persons with a BMI of 33.0, but they are uncommon. In our study, only 2.7% of men and women both were highly physically active and had high BMIs.
Spitzer correctly states that one common definition has the baby-boomer generation starting in 1945. Alternatively, some authors define baby boomers as those born as early as 1943,5 giving our cohort more than 35,000 boomers. The main point, though, is that our investigation is eminently relevant to baby boomers, whatever their present age, as Byers concluded in his Perspective article.
Kenneth F. Adams, Ph.D.
Arthur Schatzkin, M.D.
Michael F. Leitzmann, M.D.
National Cancer Institute
Rockville, MD 20850
adamske@mail.nih.gov
References
Doll R, Hill AB. Mortality in relation to smoking: ten years' observations of British doctors. BMJ 1964;1:1399-410, 1460.
Rothman KJ. Modern epidemiology. Boston: Little, Brown, 1986.
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.
Miettinen OS. Proportion of disease caused or prevented by a given exposure, trait or intervention. Am J Epidemiol 1974;99:325-332.
Strauss W, Howe N. Generations: the history of America's future, 1584 to 2069. New York: Morrow, 1991.