This well-done meta-analysis looked at 97 studies that examined the link between BMI and mortality, and found that, relative to those with a normal BMI, those with BMIs in the 25-30 range had a lower all-cause mortality. The weight of evidence, it seemed, confirmed an "obesity paradox" whereby people with higher BMIs lived longer. Cue the lay press with articles like this one from Time magazine, speculating on the benefits of body fat:
And this one from Prevention, identifying four situations where you're better off if you are overweight.
Now, to be pedantic. This was never an obesity paradox, as people in the "obese" categories (BMI greater than 30) always had higher mortalities. It would be better labeled an overweight paradox but that isn't as catchy I suppose.
Speculation as to the mechanism here was far-ranging, but as Homer Simpson said in the famous episode where he deliberately gained 61 pounds to be placed on disability: "I'm drought and famine resistant".
But a persistent chorus of researchers asked if maybe this finding was an artifact of poor measurement. Maybe people who are sick, even if they don't know it, lose weight, inflating the risk of mortality in the normal weight group.
Well that theory got a king-sized helping of facts from this study, published in the Annals of Internal Medicine:
Using data from three large cohort studies, comprising over 225,000 individuals, the researchers demonstrated that, no there was no protective effect of being overweight. In fact, there was a small, but significant, added risk for all-cause mortality.
What did this study do that the 97 in that meta-analysis didn’t? They looked at maximum weight achieved over the past 16 years as opposed to current weight.
Imagine an individual with a BMI in the normal range. We follow him for some period of time (mean follow-up in this study was 12.3 years) and see if he dies or not. If he does, most studies would assign that death to the "normal" BMI category that he had when we started following him.
But what if you found out that he had been overweight 2 years prior to that baseline visit? Would that change your opinion? Might his weight loss have been associated with some underlying disease? When the researchers used the maximum weight achieved in the 16 years prior to the baseline weight, they found this relationship between BMI and mortality:
In the dashed line, you see the hazard ratio for mortality associated with the various BMI categories based on baseline BMI. Note the slightly reduced risk among those in the overweight category compared to the normal weight category.
But when maximum BMI is used, the paradox disappears. Overweight individuals have slightly increased risk compared to those in the normal range.
To drive home the point, the authors looked at change in weight, and the results are just what you'd expect. People who lose weight were at much higher risk of death than those who did not, regardless of what their BMI was to start with. In fact, those who had a normal BMI at the most recent weight check, but who had been overweight in the past had a 19% higher risk of death than those who stayed overweight the whole time. This doesn't mean that weight loss is bad for you. It suggests that weight loss is a sign of underlying disease.
And that's the big take home here. A single measurement of BMI at one timepoint may not be that useful, because being too heavy may cause disease, but disease can make you less heavy. Trends in weight are much more important. And unintentional weight loss is a major danger signal. After all, we all know how hard it is to lose weight when you are really trying to. If your patient is losing weight without putting effort into it, it's time to look a bit deeper.