Weight and health: no simple answers

When we're worried about our weight, is it for reasons of appearance, or health? And what if -- shades of Sleeper -- it were discovered that being fat increased longevity? I'd bet on fashion over health.

In the event, it seems the relationship of weight to longevity is more complex than once believed, as today's NY Times reports:

About two years ago, a group of federal researchers reported that overweight people have a lower death rate than people who are normal weight, underweight or obese. Now, investigating further, they found out which diseases are more likely to lead to death in each weight group.

Linking, for the first time, causes of death to specific weights, they report that overweight people have a lower death rate because they are much less likely to die from a grab bag of diseases that includes Alzheimer's and Parkinson's, infections and lung disease. And that lower risk is not counteracted by increased risks of dying from any other disease, including cancer, diabetes or heart disease. . .

"If we use the criteria of mortality, then the term 'overweight' is a misnomer," said Daniel McGee, professor of statistics at Florida State University.

"I believe the data," said Dr. Elizabeth Barrett-Connor, a professor of family and preventive medicine at the University of California, San Diego. A body mass index of 25 to 30, the so-called overweight range, "may be optimal," she said.

I haven't looked at the study myself, and there may be flaws in the methodology. Importantly, the article notes:

The researchers caution that a study like theirs cannot speak to cause and effect. They do not yet know, precisely, what it is about being underweight, for instance, that increases the death rate from everything except heart disease and cancer. Researchers tried to rule out those who were thin, because they might have been already sick. They also ruled out smokers, and the results did not change.

The reference to "ruling out" (that is, excluding from the study sample) "those who were thin" makes no sense; if no thin subjects, then no underweight subjects. I wonder what correction was being made that the NYT reporter so misconstrued? I could see there being real problems with the composition of the group classified as "underweight", with some just being lean, while others might be obviously sickly -- even if not smokers or suffering from chronic disease. The problem for study designers is that sickliness is easy to recognize but difficult to define.

Posted by David on November 7, 2007 8:54 AM