Apparently conflicting results on obesity and mortality

The Washington Post has an article claiming that being just a few pounds overweight can lead to premature death:

The 10-year study of more than 500,000 U.S. adults found that those who were just moderately overweight in their fifties were 20 percent to 40 percent more likely to die in the next decade. Another study involving more than 1 million Korean adults, also being published in this week's New England Journal of Medicine, produced similar results.

Meanwhile, Jake at Pure Pedantry has uncovered a report claiming just the opposite:

In a review of the studies published on the subject, Romero-Corral et al. show that BMI does not reflect an increased risk for Coronary Artery Disease (CAD) unless you are either very thin or severely obese.

What's up with that? I thought BMI was supposed to correlate with increased mortality. Jake has an excellent explanation:

Just to be clear, the take home from this article is not that obesity is not a risk factor for cardiovascular disease. We know from the Framington Study that not only is it associated with a variety of other risk factors, obesity is also in itself a risk factor. In addition, there are numerous other ways that obesity increases morbidity and mortality.

Jake goes on to point out that, overall, BMI is a blunt instrument for measuring health problems. Just as there are different types of people, there are different types of over- and underweight people, many of whom may be perfectly healthy. The WaPo article gets at some of these issues, but not as clearly as Jake does:

But other researchers were not convinced, saying the findings are questionable for a number of reasons, including the fact that the weight data relied on the participants' recollections, which are notoriously unreliable, instead of direct measurements. Also, the sample was not necessarily representative of the general population, they said.

"I feel like the researchers were trying to manipulate their data to match their conclusion," said Linda Bacon of the University of California at Davis. "I think it's very threatening to people to be open to the idea that overweight may not be as bad as we think."

This gets back to the notion that BMI is a blunt instrument. What I'd like to see is more research exploring other ways to address the underlying health problems that appear to be related to BMI.