Jigsaw Puzzle

Had Enough: Inconclusive Obesity Paradox Publications

“More research is needed.” We find those immortal words at the end of a large percentage of research papers. But Anna Peeters is calling for just the opposite. In the International Journal of Obesity (IJO), she says we’ve had enough of inconclusive obesity paradox publications.

How Much Health Risk Does Obesity Cause?

Simply asking this question serves to stir a controversy. But mostly, people use epidemiological evidence to raise the question. In the IJO recently, Hailey Banack and Andrew Stokes questioned the basis for continuing claims of an obesity paradox. The seeming paradox is that epidemiologists can find correlations between higher BMIs and better survival in some populations of people with an obesity-related disease.

Banack and Stokes suggest that the real problem lies with the methods in these studies. They say that issues with BMI, reverse causation, and selection bias should lead people to question claims of an obesity paradox:

The only “paradox” we can see here is why researchers continue to claim to have evidence of a paradox without careful consideration of potential methodological explanations.

Collider Bias Muddies the Water

Peeters focuses on the issue of selection bias in these studies, specifically in the form of collider stratification bias. The math behind this problem can be difficult, but the issue is not. When collider bias creeps into a study’s methods, you’re stuck with it. You can’t explain it away. You can’t exclude it from explaining your results. So you can’t draw conclusions about risks and benefits.

There’s only one way around it. Do a clinical trial.

We already have a number of such studies of obesity and outcomes for cardiovascular risk factors. For those factors, we have strong evidence that weight loss produces a benefit. The important gap is solid evidence for a survival benefit.

So publishing more epidemiology studies to suggest an obesity paradox is pointless. We’ve had enough. We need real studies of survival outcomes, says Peeters.

Click here for the letter by Peeters and here for the editorial by Banack and Stokes. If you want to dig deeper into this controversy, click here, here, and here for a debate on the role of collider bias in obesity paradox observations.

Jigsaw Puzzle, photograph © MIKI Yoshihito
/ flickr

Subscribe by email to follow the accumulating evidence and observations that shape our view of health, obesity, and policy.


October 22, 2017

2 Responses to “Had Enough: Inconclusive Obesity Paradox Publications”

  1. October 22, 2017 at 9:50 pm, John Dixon said:

    I agree that obesity paradox does not exist because the relationship between BMI and mortality is NOT a straight positive line but U-shaped. The NADIR (best BMI for mortality) is not and never will be fixed. It varies with age, ethnicity, and state of health. When BMI was introduced as a metric this was known. We have ignored this since.

    Our real problem is how this information can be translated to simplified public health messaging.

    Would it seem odd that in a competitive world carrying a little additional weight may provide some advantage to the elderly and ill at times of food scarcity. We were designed for earlier times.

    • October 22, 2017 at 9:57 pm, Ted said:

      Well said, John.