Rethinking a “healthy weight”: New study finds that being overweight is not associated with increased mortality from cardiovascular causes
November 9, 2007 By Benjamin A. Olenchock, M.D. Ph.D. , Bethesda, MD
A new study has examined disease-specific mortality rates based on body mass index (BMI) using data from the National Health and Nutrition Examination Survey (NHANES) databases. The findings, published in the Journal of the American Medical Association, suggest that individuals defined as overweight (BMI between 25 and 30) do not have increased cardiovascular mortality. Obesity (BMI > 30), however, was associated with excess cardiovascular mortality, although to a much lesser extent than in previous years. Under-weight (BMI<18.5) was associated with higher non-CVD/non-cancer mortality. This research is a continuation of work published in 2005 that examined all-cause mortality. The previous study reported that over-weight individuals had decreased all-cause mortality compared to normal weight individuals, while obesity and under-weight were found to have increased all-cause mortality.
The statistical analyses required to do such a study were quite complex. The study used the NHANES databases to gather baseline data in representative cross-sectional samples of the United States population. Cause of death was divided into three categories: cardiovascular, cancer, and all other. They used a Cox proportional hazard model to calculate the relative disease-specific risk for the different BMI classes, separating the analysis by age for statistical reasons. The model included sex, smoking status, race, and alcohol consumptions as other covariates. To estimate the excess disease-specific mortality in 2004 attributable to BMI class, they applied their relative risk data to mortality data from the United States vital statistics in 2004, estimating the current distribution of covariates using the NHANES 1999-2002 cross-sectional data set.
Obesity was associated with 81,072 [CI 51,433 to 110,710] excess deaths from cardiovascular disease but no statistically significant increase in cancer deaths (14,930, [CI -13,721 to 43,582]). Obesity was, however, associated with increased obesity-related cancers, defined as colon, breast, esophageal, uterine, ovarian, kidney and pancreatic (13,839 excess deaths, CI 1920 to 25,758). The authors analyzed longitudinal changes in attributable deaths by using relative risk estimates for individual NHANES databases. They found that using NHANES I (1971-1975) relative risk estimates, obesity was associated with 161,290 excess deaths from cardiovascular causes, much more than the current estimate of 81,072.
Overweight was associated with no excess deaths from cardiovascular causes (-17,074, CI -50,407 to 16,259) or cancer (-13,533, CI -44364 to 17298), but significantly decreased deaths from all-other causes (-107674, CI -148738 to -66610). Underweight, in contrast, was associated with excess deaths from all-other causes (23,455, CI 11,848 to 35,061). Looking closer at the over-weight category, the authors did find that both overweight and obese individuals had excess deaths from kidney disease and diabetes.
This data adds significantly to their previous work, which demonstrated differences in all-cause mortality based on BMI. One interesting finding is that over-weight is associated with less all-cause mortality and no difference in cardiovascular mortality. Also, the data are notable for the apparent change in the excess cardiovascular deaths associated with obesity over time, perhaps reflecting improvements in medical care in recent years. These data are sure to bring commentaries on the statistical methods used, changes in BMI class over time as a consequence rather than cause of disease, concerns about cause of death reporting bias, etc. Nonetheless, this work is an important contribution to our current understanding of healthy body weight and the relationship between cardiovascular disease mortality and excess body weight.
- PMID 17986696