A paper in the current issue of the Journal of the American medical Association presents some interesting data on the association of obesity with overall mortality. The authors of the study reviewed mortality data from almost 3 million subjects. They classified mortality against body mass index (BMI). Their reference group were people with BMIs of 18.5 to 25 which was defined as normal. The table below summarizes their findings.
Concentrate on the first group of data – the one labeled Self Reported or Measured Height and Weight. HR stands for hazard ratio. If it is less than 1 the group has a lower death rate than the controls. If it is more than 1 the mortality rate is greater than controls. If the range given in parentheses overlaps 1 the change is not statistically significant. The first thing that jumps off the page is that those subjects with BMIs 25 to 30, defined as overweight, have a statistically significantly lower death rate than the controls. Those with BMIs 30 to 35 defined as obese have a death rate no different from the controls. Their HR straddles 1. Those with BMIs greater than 35 – very obese – had an increased death rate compared to controls but only if they were under age 65. The HR for the 65 year olds was on both sides of 1 and thus not statistically different from the controls.
So what does all this mean? First of all the study is retrospective and subject to all the caveats attached to this type of analysis. Next it just shows associations. Causality cannot be imputed to these data. Finally, mortality is a very complex mix of an almost infinite series of events any one of which, though important, may be minor and not have the same impact as a bullet in the brain or getting hit by a truck. This said, the study suggests several tentative conclusions.
First, if we believe that you can never be too thin or too rich this paper and the recent “fiscal cliff’ fiasco suggest that we should change our view on both states. A “normal” BMI of 18.5 to 25 may be too low. If this study is correct a BMI range of 25 to 30 is optimal. To put this in perspective, a five foot six inch woman who weighs 170 pounds has a BMI of 27.4 – smack dab in the middle of the seemingly optimal range. A five foot 10 inch man weighing 190 has a BMI of 27.3 – again in the middle of the “right” range.
Mortality rates for obese subjects (BMI 30-35) were no different from controls though they might have been different from the overweight individuals; but this comparison wasn’t made. A higher normal range of BMI should be considered. Finally, BMI at any level was not associated with increased mortality if the subjects were over 65. Perhaps all those who were condemned to death by fat had died by 65 leaving only those indifferent to obesity. Regardless, this study and others that have shown a beneficial effect associated with obesity in patients with certain diseases suggest that we may be focusing a little too intensely on weight. You might want to reconsider one of your New Year’s resolutions.