I recently wrote about the obesity paradox. There are diseases in which patients who have a disease and are overweight live longer than patients with the same disease who are thinner –  hence the paradox. The December 2 New England Journal of Medicine examines Body-Mass Index and Mortality among 1.46 Million White Adults.The authors measured the relationship between BMI (weight in kg divided by the square of height in meters) and all-cause mortality in a pooled analysis of 19 prospective studies, predominantly designed to study cancer, which included 1.46 million white (non-Hispanic) adults and 160,087 deaths. The figure below presents the data obtained from white men. The data from women was virtually identical. The hazard ratio for death (+/- 95% confidence interval) is plotted on the ordinate while BMI is on the abscissa.

The relationship of death to BMI is similar between the healthy subjects and all subjects which includes sick people who were in studies becasue of the underlying disease they were afflicted with.  The optimal BMI seems to be between 20 and 25. This paper, which is a retrospective look at old prospective studies, shows a strong association; it does not define causality. In other words it does not tell us whether people with BMIs above 25 gain increase longevity by permanently losing  losing weight and thus moving to a new BMI. It would take a new prospective study to pin this down. Such a study would obviously be costly and take a very long time to perform.

Finally, while the mortality data are statistically significant for BMIs greater than 25 the difference is very small and apt to be invisible on an individual basis. A clinically important effect of death rate is likely evident only in the morbidly obese. Nevertheless, this study does provide support for the intuitive belief that too many pounds are a drag on longevity.