Resolving the obesity paradox

A new study in The BMJ1 has taken a big step toward resolving the so-called ‘obesity paradox’.

The paradox is not that some people can be overweight and still be metabolically healthy (as I’ve written before, that is not surprising); the paradox is that when considering the risk of death over some fixed period of time (for example, 5 years) for all people, the risk is lower for overweight people than for lean people2.

That is surprising – after all, excess body fat clearly increases the risk of developing many diseases that should increase the risk of death (diabetes, cardiovascular disease, hypertension, cancer, etc.).

So what’s going on? Does excess body fat somehow increase the risk of developing many diseases while simultaneously providing protection against dying from them?

Obviously not. The results of the new study clearly demonstrate that the apparent paradox is simply the result of confounding factors that were not controlled for in previous studies.
Performing well-controlled experiments to test hypotheses about human health is extremely difficult because it’s impossible to even observe, let alone control, people’s behavior for any extended period of time. As a result, it’s never really clear whether a relationship between two experimental variables is directly causal or is, in fact, due to some other variable that is either unobserved or not correctly accounted for.

The previous studies that reported the apparent protective effect of excess body fat had some pretty obvious potential confounds. For starters, they didn’t account for smoking. Smokers are, of course, more likely to die than non-smokers, and they are also more likely to be lean. So the increased risk of death for lean people in those studies may have had nothing to do with their leanness per se, but may have simply reflected the fact that they were more likely to smoke.

The previous studies also failed to account for illness. It’s well known that many illnesses are accompanied by weight loss, so some fraction of the lean people in previous studies may have actually been lean because they were dying, rather than dying because they were lean.

The researchers who conducted the new study started by reproducing the results of the previous studies. When they looked at the risk of death for all people over a period of 5-10 years after the start of the study, they found the same protective effect of excess body fat as in previous studies; as shown by the red line in the graph below, the risk of death was lowest for people in the overweight range.


But when they considered only non-smokers that were healthy at the start of the study (i.e. they did not have any known diseases) and looked at their risk of death over a much longer period (20-25 years), they found something very different. In fact, as shown in the blue line in graph above, the protective effect of excess body fat went away completely. There was still an increased risk of death for people that were extremely underweight, but the risk was lowest for people on the low end of the normal range.

So, in the end, the paradox was really nothing more than a distraction resulting from poor experimental design. Let’s hope we can all remember this lesson the next time we’re trying to resolve a ‘paradox’.
(1) Aune D, Sen A, Prasad M, Norat T, Janszky I, Tonstad S, Romundstad P, Vatten LJ. BMI and all cause mortality: systematic review and non-linear dose-response meta-analysis of 230 cohort studies with 3.74 million deaths among 30.3 million participants. BMJ. 2016.

(2) Lavie CJ, De Schutter A, Milani RV. Healthy obese versus unhealthy lean: the obesity paradox. Nat Rev Endocrinol. 2015.