Science, Schmience

I don't know what the etiquette is about cross-blogging, but here's the text of a post I just made on the Health At Every Size radio show blog, Didn't want anyone to miss it.

Another volley has been fired in the "war against obesity," in the form of articles in yesterday's edition of the The New England Journal of Medicine that claim "excess" weight during midlife is associated with "an increased risk of death."

The claims are already getting play in the media...which is, for the most part, missing the real story. Which is that the data published in these articles actually says almost the opposite of what the authors claim. A close inspection of their own data shows that either "excess weight" was actually associated with LOWER mortality, or that there was no increased mortality risk.

How could the authors come to the opposite conclusion? By jettisoning data that didn't fit with their preconceived notion....err, hypothesis...that fatness is a bad, bad thing which needs much money thrown at it for research and public health purposes.

But don't take my word for it. Here's what Linda Bacon, Ph.D., associate nutritionist with the University of California, Davis, has to say about these studies (her comments were originally posted on a Health At Every Size email list and are posted here with her permission). Bacon's doctorate, by the way, is in physiology with an emphasis on weight management. She also has two master's degrees, in exercise physiology and psychology. She's a researcher herself, and the lead author on a study published in the Journal of the American Dietetic Association last year that found a non-weight-loss focused Health At Every Size program was far better for middle-aged "obese" women than a traditional weight loss approach.

OK, here's what Bacon has to say about the NEJM articles [emphases mine]:

In April 2005, Flegal et al., using a sophisticated modeling method respected for accuracy and data that included actual measurements on people nationally representative of today's population, determined that overweight predicted lower mortality risk than “normal” weight. These results were consistent with a large body of previous literature.

The backlash to[Flegal et al.'s] JAMA publication is not surprising. Fear-mongering about weight is worth billions to industry and consistent with government policy. Few stand to gain from the news that overweight is benign, if not beneficial.

NEJM’s recently published article entitled "Overweight, Obesity, and Mortality in a Large Prospective Cohort of Persons 50 to 71 Years Old" is clearly part of that backlash, heavy on sensationalism and weak on facts. Buried in their results is a conclusion completely consistent with that reached by Flegal et al.:In their words, overweight was not associated with increased risk of death for men, and only very weakly for women. (Indeed, even that conclusion is generous: the relative risk at the lower end of the "normal" category is similar to that found at BMI [Body Mass Index] 40 for men and 35-40 for women.)

Yet this conclusion is not prominent in the paper, nor does it show up in the abstract. Midlife gave them the results they were looking for and that’s what they chose to focus on. This despite the fact that their midlife data was drawn from a subanalysis which had an even smaller sample (now whittled down to about 5 percent of the sample that was sent surveys) and was based on recalled weight from over a decade previous. (That 40% of the sample chose to leave the question on recalled weight from a decade ago blank should give some indication of the ability of people to accurately remember and report this information.)

Their paper is weak for many other reasons:

1. They had a very low response rate (18%) from a sample that is not nationally representative.
2. Their data is based on self-report, which is known to be inaccurate. One only needs to look to their physical activity data to recognize the implausibility.
3. Even after limiting data to never smokers, there's till no increased mortality associated with overweight. The researchers hide this by using only the top )23-24.9) of the "normal" weight category as their referent group, while comparing it to the entire overweight group. Among never smokers the mortality risk in the normal weight category as a whole is the same as among the overweight category as a whole.
4. Adjustments for potential confounders were also weakly conducted: there is no data on fitness, only crude data on physical activity, and no consideration of other pominent potential confounders such as nutrient intake (other than alcohol consumption), weight cycling, or socioeconomic class.

This paper also demonstrated that people that lost weight had higher mortality, which is again consistent with previous literature. Why was this important information left out of the analysis?

Bacon went on to say that Paul Campos, author of The Obesity Myth, wrote an excellent Rocky Mountain News column about this issue. Read  Campos's article here.