TCS Daily

Slaughtering the Fatted Calf

By Iain Murray - August 19, 2002 12:00 AM

The current obsession in the public health arena is obesity. Ever since Surgeon-General David Hatcher announced in December last year that obesity was a major public health hazard, killing 300,000 people every year, the fat has been sizzling in the fire. The 61 percent of Americans who are classified as overweight or obese have been subject to more and more claims that their lifestyle is putting them at severe risk, while the food and drink industries braced themselves for the inevitable onslaught of litigation.

There is, however, a significant gray area in this debate that the sweeping statements of the former Surgeon General and the trial lawyers overlook. The official classification of so many Americans as overweight relies on the Body Mass Index (BMI), which is derived by dividing weight in kilograms by the square of the height in meters, thus getting a good indicator of weight adjusted for height. Researchers generally agree that a BMI up to 24 is an indicator of little or no risk to health. A BMI of 27 or above is linked to higher risk of heart disease, diabetes, stroke and other weight-related problems. The risks increase especially when the BMI is greater than 30. The problem is how to categorize those whose BMI is 24 to 27. The official standards introduced in 1998 start the category of "overweight" at a BMI of 25, well below the level clearly associated with health risks.

This categorization is controversial. "I'd rather see a caution zone from 25 to 26.9, where you tell people, 'Don't gain more weight, become physically active,'" Judith Stern, professor of Nutrition at the University of California, Davis, told Newsweek in 1998. She went on, "There's no evidence that BMIs lower than 27 are associated with significant increases in mortality."

Until recently, it seemed. A study published in the New England Journal of Medicine (NEJM) by the Framingham Heart Study suggested that the overweight were in danger of heart failure. The Washington Post thought the study provided conclusive proof that "being even moderately overweight ... increases a person's risk of developing heart failure, a serious condition in which the heart is unable to pump enough blood to supply the body's needs" ("Study Links Excess Weight to Risk of Heart Failure," Aug.1). In fact, it did not, but that didn't stop the researchers, the journal and the media from acting as if it did.

The study found a relative risk for the overweight of 1.5:1 in women and 1.2:1 in men, meaning that overweight women were 50 percent and overweight men 20 percent more likely to suffer heart failure than those of normal weight. Yet, as we have argued here before, on their own these figures should not be enough to convince scientists of anything. Because any study, no matter how well designed, may fail to account for other, confounding factors and the random effects of chance, hazard ratios of less than 2:1 are not recognized as being indicative of anything. In fact, NEJM told Science magazine in 1995 that, "As a general rule of thumb, we are looking at a relative risk of 3 or more (before accepting a paper for publication)."

So why did they accept this paper, with its inconsequential ratios, which, in the case of overweight men, were not even statistically significant? The study could not find a consequential risk even when it looked at obese men, whose relative risk also did not exceed 2:1. The only weight and sex combination that showed any risk ratio to warrant significant attention was that of obese women.

The study also looked at risk in a different way, abandoning the weight categories and looking at weight as a "continuous variable," thereby lumping together all weight categories with a body mass index over 25. This did produce a statistically significant risk ratio for men, but at the miniscule level of a mere five percent increased risk. The risks for women and the overall population were similarly tiny using this method.

The study therefore proves very little, and certainly not what the Post claimed. Women may be at some risk of heart failure from being obese, and men might be, but people of either sex have no real reason to fear heart failure if they are merely overweight. Another question arises from this research. If being overweight is bad in itself, why is it worse for women than for men? This very finding suggests that something else may be at work that the researchers have not considered.

What is most interesting about this particular story is the way in which the media, in reporting the study, ignored the hazard ratios and simply accepted the claims of the researchers at face value. It is one thing to claim that a study has found a causal link if it is well designed and discovers a relative risk of 50 percent. It is another to do so if it has failed to find statistical significance or, equally, a statistically significant but tiny relative risk of five percent. Researchers who over-inflate their claims, journals that publish them, and journalists who fail to ask the basic questions are failing the public. Policy decisions taken on the basis of such weak science can expose the public to wasteful public expenditure and higher costs. We have already seen the first lawsuit aimed at fast food restaurants for failing to warn people that eating too much might make them fat. The trial lawyers are probably licking their lips at being presented with such a tasty morsel in the shape of this study.



TCS Daily Archives