Confound It!

“There is another type of medical-study fallacy that borders on residual confounding. Consider the recently released meta-analysis (analysis of other analyses) of the relationship between obesity and mortality. It allegedly found that being slightly overweight, or even modestly obese, was associated with lower mortality than being so-called normal (having a body-mass index of under 25). Morbid obesity (BMI over 35) was, as would be expected, associated with higher mortality.” Sanford Rose

Dolors & Sense

By Sanford Rose

KISSIMMEE Florida—(Weekly Hubris)—4/8/2013—That’s residual confounding, a condition epidemic in medical research.

It’s when independent explanatory factors that are not taken into account in a study are correlated with either those that are taken into account or with the dependent conditions that the study purports to explain.

Thus, in a prominent study, sleeping pills are associated with increased mortality and a higher incidence of cancer.

But depression, which is not considered or measured in the study, is associated with all three—use of sleeping pills, cancer and increased mortality. So one doesn’t know whether it’s the pills that kill or cause cancer, or whether it’s the depression. And, of course, there might be reverse causation—i.e., cancer leads to increased use of the pills.

To be sure, sleeping pills, whether Ambien or the more traditional benzodiazepines, are not terribly good for one. At best, they merely shorten the time it takes to fall asleep (improved sleep latency). At worst . . . . Well, we simply do not know. Some studies suggest that, while they contribute to anterograde amnesia, the effects are scarcely noticeable—in most individuals.

If truth be told, it is not easy to measure with much exactitude the role of sleeping pills in causing either illness or death.

Thus, the sensationalistic newspaper headlines that invariably accompany the release of this type of study are rarely warranted, although one can hardly ask an editor to title his story: “Associational Link Between Morbidity and Hypnotics Reaffirmed.”

There is another type of medical-study fallacy that borders on residual confounding. Consider the recently released meta-analysis (analysis of other analyses) of the relationship between obesity and mortality. It allegedly found that being slightly overweight, or even modestly obese, was associated with lower mortality than being so-called normal (having a body-mass index of under 25). Morbid obesity (BMI over 35) was, as would be expected, associated with higher mortality.

Quite apart from the weakness of using BMI as a proxy for obesity (heavily muscled types may have high BMIs, but still not be obese), this review of the literature does not, as it probably could not, take into account the effect of increased medical care on mortality. An argument can be made that the overweight and the modestly obese are more apt to get sick than the less corpulent. Therefore, they are more likely to seek medical help, which keeps them alive longer than those who are not in the habit of visiting doctors because they don’t feel they need to.

The overweight are not healthier; they are just more vigilant.

Note: The image used to illustrate this column was taken from http://www.flickr.com/photos/puuikibeach/8178021004/.