Obesity and Diabetes: Parsing the Proposition of a Paradox

Hero Images / Getty Images

A study published in the Annals of Internal Medicine in 2015 provocatively refers to the “obesity paradox” in its title. That alleged paradox refers to any measure of better health with higher BMI. In this study, type 2 diabetic adults who were overweight had a lower rate of mortality than did those with a lower weight, although obesity was associated with increased mortality. That, then, was the purported paradox.


But there really was no paradox.

1) Overweight did increase the likelihood of cardiac events, just not mortality. Modern medicine is far better at keeping people alive than keeping them healthy, so this finding may merely indicate the expected impact of modern medical treatment among those intact enough to respond to it. Although attempts were made to control for such factors, there is a good chance that adults with type 2 diabetes who had lost, or were losing, the excess weight that caused (or contributed to) diabetes in the first place were sick or failing. People who are sick or failing obviously tend to do less well than those who are not.

2) In the general population, being 'lean' may indicate particular health and fitness. But that is unlikely in a population of people with type 2 diabetes, because if they were especially fit, they would be unlikely to have diabetes in the first place.

So that begs the question: Who is lean/thin in a population with type 2 diabetes? The group might include some people working hard to lose weight and get fit in an effort to control or reverse their diabetes. But it certainly includes people who are sick, frail, depressed, isolated, and so on. Weight loss in such a population, if not intentional and associated with fitness, is likely to indicate failing health for any such reason.

So the results of the study are then a bit misleading. It's not that a higher BMI protects health; it's that the factors responsible for a lower or falling BMI in this population tend to be rather ominous.

3) Both the highest and lowest BMIs in the study were associated with bad outcomes. This is far from paradoxical, and if anything, quite concordant with sense, as well as prior science. Being obese is bad for health, clearly. So is the combination of having diabetes and being quite thin. 

4) The measures in this study were static, at one point in time.  But weight can change over time, and that is an important consideration.

So, for instance, if someone develops type 2 diabetes partly or wholly because they are overweight—and then starts to lose weight because of failing health at some point—they will pass through 'normal' weight on their way to 'below normal.' This invites a testable hypothesis: Is the apparent 'advantage' of overweight ads compared to normal weight an actual advantage of the extra body weight, or an inadvertent comparison of 'stable' weight to 'falling' weight?

This study cannot answer the question; it merely invites it.

There may be some survival advantage linked directly to extra body mass. Overall, though, I think the issue is one of reverse causation. It's not that overweight confers a survival advantage, but that salient survival advantages related to overall health and social circumstance are associated with keeping on the extra weight that caused the type 2 diabetes in the first place, whereas adverse social and health circumstances are apt to be the reasons that weight starts to fall off. A higher rate of mortality in a sicker population is the furthest thing from a paradox.

The Obesity Paradox in Type 2 Diabetes Mellitus: Relationship of Body Mass Index to Prognosis: A Cohort StudyAnn Intern Med. 2015;162(9):610-618

Continue Reading