Obesity Health Risks: Up, Down, or Sideways?

Speculation has been bubbling for years now about trends in the health risks associated with obesity. Maybe, according to this speculation, the risk of death and other bad outcomes from obesity are going down as obesity and its complications are becoming more common. Maybe better medical care is reducing those health risks.

In a very smart new analysis, Tapan Mehta and colleagues have shot a hole in that speculation.

They looked closely at the analyses behind that speculation and found a serious problem. By lumping people with different BMI values into just a few categories, those analyses became confounded and potentially misleading. The authors conclude:

Some of the apparent diminution of the association between obesity and mortality may be an artifact of treating BMI as a categorical variable.

We advise caution in interpreting the apparent reduction in the obesity-mortality association as being driven by treatment advances.

This is a very challenging subject. The statistical methods are complicated. But the problem comes down to the fact that BMI is a continuous variable. And the analyses in question took a problematic shortcut. They squeezed many different BMI values from many thousands of people into just a few categories. That shortcut was a particular problem because the distribution of BMI values within the categories was changing over time.

The theory is intuitively appealing. Maybe we’ll get better at dealing with obesity and its complications as they become more common. But right now, we don’t have compelling evidence that holds up to scrutiny. All we have is speculation.

Thank goodness for smart biostatisticians.

Click here to read the study and here for more background. Click here for the study that started this speculation.

Up and Down, photograph © Antonio Cinotti / flickr

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

September 26, 2016