A literature review published today in The Journal of the American Medical Association confirms earlier research indicating that the government's notion of how much you should weigh does not correspond to mortality risks for people with different body mass indexes (BMIs). Looking at nearly 100 studies, Katherine M. Flegal of the National Center for Health Statistics and three other researchers find that people in the "normal" BMI range of 18.5 through 24.9 (which is not in fact normal, since most Americans exceed it) were more likely to die during the study period than people in the "overweight" range of 25 through 29.9. Even people in the BMI range corresponding to "grade 1 obesity" (between 30 and 35) were no more likely to die during the study period than people in the government-recommended range, although people with "grades 2 and 3 obesity" (BMIs of 35 or more) were. The New York Times notes that "the report, although not the first to suggest this relationship between B.M.I. and mortality, is by far the largest and most carefully done." Similar patterns can be seen in studies of medical spending associated with various BMI ranges, where extremely obese people are the ones with substantially higher annual costs.
Do thinner people have a higher mortality rate because they are more likely to smoke or because they have lost weight due to illness? Flegal and her co-authors say the data "provide little support" for those explanations. "Our findings are consistent with observations of lower mortality among overweight and moderately obese patients," they write. "Possible explanations have included earlier presentation of heavier patients, greater likelihood of receiving optimal medical treatment, cardioprotective metabolic effects of increased body fat, and benefits of higher metabolic reserves."
Another possible factor: BMI, which is your weight in kilograms divided by the square of your height in meters, is not a very good measure of obesity, since it does not differentiate between muscle and fat. (If you are five feet, nine inches tall, for example, you are considered "overweight" at 169 pounds and "obese" at 203, whether you are a bodybuilder or a flabby office worker.) Furthermore, extra fat in itself is not necessarily a health problem. "Fat per se is not as bad as we thought," one expert tells the Times. "What is bad is a type of fat that is inside your belly. Nonbelly fat, underneath your skin in your thigh and your butt area—these are not necessarily bad."
Whatever factors explain these mortality rates, it is increasingly clear that the definition of obesity as problem, let alone an "epidemic" requiring government intervention, hinges on official standards with little basis in reality. If the share of American adults whose weight poses a life-threatening danger is closer to 6 percent (the share classified as extremely obese) than to 69 percent (the share deemed "overweight"), that makes a huge difference, whether or not you think trying to move those numbers is an appropriate function of government.
In case you are not confused enough by the idea of a "normal" and "heathy" weight range that is neither, last month the Times reported that "17 percent of children under 20" have "a body mass index at or above the 95th percentile." If that sounds impossible to you, you are right: A few days later, the paper clarified that the definition of obesity in children is "based on a comparison of a child's body mass index to growth charts from a reference population" of children in the 1960s and '70s, rather than "a comparison of a child's body mass index to measurements of the current population of children." The mistake illustrates how the press often tosses around obesity numbers without much thought about what they actually mean. So does the correction, which raises the question of why those weights from 40 to 50 years ago are automatically considered better.