DB's Medical Rants: Debating Medpundit again

ACP-ASIM Observer
AMA news
NY Times Health
Washington Post Health
LA Times Health
Medscape
BBC Health News
Healthier US.Gov
No Free Lunch

Previous Entry | Main | Next Entry

August 01, 2002

Debating Medpundit again

We do not always agree. Hopefully, we are civil. Medpundit criticized the effect size in the obesity and risk of heart failure study - Sweep on, you fat and greasy citizens: -Shakespeare, As You Like It..

Or so says the result section of the not-so-helpful abstract of the paper, a statement which was duly copied in CNN’s report of the study. That sounds pretty impressive, but the abstract doesn’t tell us how many overweight subjects had heart failure compared to the non-obese. It only talks of rates of risk which can be misleading. In fact, the paper itself isn’t much clearer. It, too, never gives out the absolute numbers, preferring instead to deal with “person-year of follow-up” compared to numbers of heart failure cases. Even using these gymnastics, the final results aren’t all that impressive. For women of normal weight there was a 10-year cumulative age-adjusted incidence of heart failure of 3.4%. For overweight women (BMI 25.0- 29.9) it was 3.7%. For the obese (BMI > 30), it was 6.8%. Yes, it doubled, but the percentages are still pretty small. The figures for men are similar. Normal weight men had an incidence of 4.9% over ten years, overweight men had an incidence of 6.1% and the obese had an incidence of 10%.

Epidemiologic papers are tricky to interpret. I spent some time reading the article, and will give my interpretation.

Over a 10 year period, each BMI increase of 1 increases the heart failure risk by 5% (men) or 7% (women). Thus, having a BMI of 30 increases the heart failure risk by 30%. This makes sense (the more overweight the greater the risk), but the implication is difficult to understand. A better way to consider the data comes from the population attributable risk. This calculation estimates the percentage of heart failure that weight control would eliminate, or what percent of heart failure can we blame on weight alone (controlling for all other known variables). In this study,

The population attributable risk of heart failure due to overweight was 14.0 percent in women and 8.8 percent in men. The corresponding population attributable risks due to obesity were 13.9 percent in women and 10.9 percent in men.

These are impressive attributable risks. Given around 500,000 new patients with heart failure each year, over 50,000 are attributed to overweight and obesity. I do not believe the NEJM got this one wrong.

Posted by

on August 01, 2002 07:24 PM | TrackBack

Comments:

I should start reading this, as I do some other blogs, from the bottom up, I've already used my best quote about statistics.

"The population attributable risk of heart failure due to overweight was 14.0 percent in women and 8.8 percent in men. The corresponding population attributable risks due to obesity were 13.9 percent in women and 10.9 percent in men."

So overweight-vs-obese makes no difference for women? And a barely-significant 2% difference (in epidemiology, 1.5% is generally considered the lowest significant percentage) for men? I wonder which I am considered to be?

"Given around 500,000 new patients with heart failure each year, over 50,000 are attributed to overweight and obesity."

Ok, let's use THAT stat the way the conclusions (not the data or the report itself, just the summary) do. 10% were overweight, so 90% were not, thus overweight obviousy confers a benefit.

Keep those banana splits coming.

Sheesh.

PS., I shuld mention once in a while how grateful I am that your blog allows responses directly to the affected section. Those using EMail are cumbersome for me and the blogger alike (Me: How do I specify what I am responding to? Blogger: what is this person talking about? Blogger: I got HOW many EMails today?)

Posted by: John Anderson on August 4, 2002 10:56 PM

Ah, I was careful in explaining the statistics, but I must have not made the statistics clear. These data do NOT imply any protection from obesity. More than 10% of heart failure is associated with obesity. Obesity increases other risks - diabetes mellitus, hypertension, and hypercholesterolemia. These data state that even correcting for those other risks, obesity alone accounts for ~ 10% of heart failure.

Posted by: db on August 6, 2002 11:49 AM

"These data state ..."

I concur, the data do indeed show that. But the report's conclusion/summary implied quite differently in the same manner as the 50 of 500: it is that point I was trying to make, as it is the part which would be quoted in mainstream press.

Posted by: on August 7, 2002 12:31 AM

I take no responsibility for the irresponsibility of the mainstream press .

Posted by: db on August 7, 2002 05:54 AM

"I take no responsibility for the irresponsibility of the mainstream press"

LOL 8-)

Still, I think the summaries should be peer-reviewed with care approaching what is given to the reports themselves. Come to that, it is not unknown for the summary to contradict the report: the WHO is known for this in what they call "diplomatic summaries" meant for politicians instead of scientists.

Posted by: John Anderson on August 9, 2002 02:45 AM