Monday, July 19, 2010

Not everything's caused by obesity: Brain and memory function edition


Good lord.

I know people like to blame obesity for everything. Every disease, every problem - everything.

You know what I'm blaming on it today? Authors' and peer reviewers' attitudes about their studies and results.

Last Tuesday I was asked by CTV to read a study that was pending publication in the Journal of the American Geriatrics Society so that I could comment on it for the national news. The study looked at 8,745 women between the ages of 65-79 free of dementia and evaluated their weight and waist to hip ratios in relation to their scores on the 3MSE, a modified mini-mental state examination validated to give an overview of cognitive function.

So I read the study.

Want to know what I found out?

That after controlling for age and education the test scores of folks whose BMI's were <25 were 95.2 while those whose BMIs were >40 were 94.1. I also discovered that the authors failed to provide the p value (the number whereby you'd see what the likelihood simple chance would lead to the result) for that comparison but they did provide a p value for the larger difference when age and education weren't controlled for and guess what, that p value, the one that looked at an even larger variance in test scores, wasn't even close to significant (at 0.10 there was a 10% likelihood the result occurred solely due to chance).

Another odd result? Women with abdominal weight distributions (apples) scored better than those with truncal distributions (pears) suggesting that unlike pretty much everything else weight related, abdominal obesity was protective against this supposed negative impact on cognition.

So basically in a best case scenario the authors could conclude that obesity may lead people to score a single percentage point lower on a test of global cognition but they'll point out at least that the difference in scores could easily have occurred due to chance. Worst case? They'll make it sound important so that they could get published.

My take? I took it to read that obesity doesn't lead to any statistically significant differences in a test of global cognition and that consequently it would seem that obesity and cognition aren't too tightly linked - a result that perhaps is bolstered by the fact that abdominal obesity appeared to be beneficial (though I should note, it may simply be due to the fact the researchers didn't measure waist circumference properly as rather than use a consistent bony landmark they used the floating umbilicus).

So what did the authors conclude?

"Higher BMI was associated with poorer cognitive function in women with smaller WHR....Further research is needed to clarify the mechanism for this interaction".
What I wrote back to CTV was the following,
"It would have to be an unbelievably slow news day for this to make the rotation."
And yet it was all over the news.

Sigh.

The media? I can forgive them, they're just trying to sell stories.

The authors? I can almost forgive them too as certainly negative publication bias might have precluded this piece and it's a publish or perish world.

The peer reviewers? I got nothin'.

Kerwin, D., Zhang, Y., Kotchen, J., Espeland, M., Van Horn, L., McTigue, K., Robinson, J., Powell, L., Kooperberg, C., Coker, L., & Hoffmann, R. (2010). The Cross-Sectional Relationship Between Body Mass Index, Waist-Hip Ratio, and Cognitive Performance in Postmenopausal Women Enrolled in the Women's Health Initiative Journal of the American Geriatrics Society DOI: 10.1111/j.1532-5415.2010.02969.x