The “obesity paradox” dominated last week’s cardiology news cycle after the British Medical Journal reported that obese patients with atrial fibrillation fared better than patients in a normal weight range.
The authors used a registry of more than 52k patients recently diagnosed with AFib and at least one stroke risk factor and followed them for over two years. The primary endpoints were death, stroke, bleeding, and heart failure.
The authors found a U-shaped relationship between BMI and hard endpoints; both low and high BMI were positively associated with events. For example:
- Mortality was lowest for people in the “obese” range (BMI between 30-35 kg/m2) compared even with “normal-weight” or “overweight” BMI.
- Mortality risk was highest for those who were “under-weight” and “severely obese.”
- For every 5 unit increase above 30 BMI, mortality risk increased by 16%.
- For every 5 unit decrease below 30 BMI, mortality risk increased by 32% (so 2x greater).
The findings suggest it is worse to be normal weight than to have a BMI in the obese range. Does this track? John Mandrola, MD (host of This Week in Cardiology) warns of collider bias—a subtype of selection bias—within the study.
To illustrate his point, he offers an example:
Say you are interested in researching the association between IQ and athletic ability. You decide to study students at an elite school, but to get into an elite school, you must be either extraordinarily smart or athletic. Following that logic, you will likely discover that athletes at elite schools have, on average, lower IQs than non-athletes. Consequently, you will find a (false) negative association between IQ and athleticism.
Mandrola argues that the recent BMJ study is also vulnerable to this bias. AFib patients with lower BMI must have had some AFib-causing condition other than obesity (e.g., cancer, pulmonary embolism, alcohol use disorder, etc.), and it’s likely that the AFib-causing comorbidity led to the worse outcomes observed in the lower weight group.
When the authors of this study investigated the relationship between body mass index and AFib patient outcomes, they found an inverse relationship between BMI and all-cause mortality for patients with BMI under 30. Others pointed out that this study may suffer from collider bias, switching an unbiased harmful effect of obesity on mortality into a biased protective effect.