Thursday, January 14, 2010

Obesity trends flat?

Today's Times reports a plateau in U.S. obesity rates, based on new studies of NHANES data published in JAMA. The main paper by Flegal et al. reveals differential trends among adults primarily by sex, and another paper by Ogden et al. shows essentially no trends in obesity among children since 1999--2000.

But when I look at the data, I'm much less sanguine about the supposed plateau in adult obesity rates. Based on NHANES data analyzed in a 2006 paper by Ogden et al. and in a 2002 paper by Flegal et al., I see a fairly uninterrupted upward trend in obesity prevalence (BMI >= 30) of about 0.5 percentage point per year. That's including the most recent data point.

The issue, and I agree that it could be important, is that the confidence intervals around the point estimates are apparently overlapping this decade. So one cannot reject with statistical tests the hypothesis that there has been no change in obesity rates. But failure to reject is not the same as finding no increase.

I would agree that there seem to be interesting differences by sex, with women displaying more stability in their distribution of BMI and men showing a clear rightward march. The Times article quotes folks as wondering whether this is indicative of good things to come, since women are typically the food preparers in families. They probably are, but are they also the fast food purchasers?