The report by scientists from the U.S. Centers for Disease Control and Prevention, the National Institutes of Health and the University of Ottawa combined the results of 97 prior studies to see how risk of death varied by body size. The researchers used a measure known as Body Mass Index — weight in pounds divided by the square of height in inches, multiplied by 703 for metric adjustment — to categorize 2.9 million people from those studies, then measured how likely they were to die while being tracked in the studies. Where possible, researchers controlled for factors such as age, sex and smoking.
The result appeared to undermine orthodoxy on the benefits of weight loss: People classified as overweight, which are those with a BMI greater than or equal to 25 and less than 30, had a 6% lower chance of dying than those with a BMI greater than or equal to 18.5 and less than 25, considered the normal range.
The finding, which compiled and reinforced similar ones from prior studies, says less about the health risks of being overweight, and more about BMI’s flaws, according to some researchers. "I suspect it’s primarily an issue with BMI being unable to measure the things it’s intended to measure, or what really matters," said Francisco Lopez-Jimenez, a cardiologist and professor of medicine at the Mayo Clinic in Rochester, Minn.
Based on work by a 19th-century Belgian scientist, BMI has been the standard way to define obesity in the U.S. since 1998 and was a clear improvement over its predecessors, researchers say. Those included reading an insurance company’s mortality tables by height and weight, which was cumbersome and didn’t boil down the numbers to a single indicator.
The problem with BMI, many scientists say, is that it lumps together all body mass, including bone, muscle and beneficial fat, rather than singling out the more dangerous abdominal fat, which most researchers see as the real threat to health.
Prof. Lopez-Jimenez and other researchers are hunting for alternatives to BMI, ranging from a simple waist-circumference measurement to a type of X-ray that measures body-fat content. The hold of BMI is powerful, however, and even if a low-cost alternative that better reflects people’s health is devised, it could take a major investment to change practice.
Most people with BMIs above 35 — for example, a 6-footer weighing more than 258 pounds — are dangerously obese, researchers agree. It is at lower levels, particularly with BMIs in the high 20s, that a wide range of body types are painted with the same brush.
It is hard to say how many people are misclassified by BMI. Steven Heymsfield, executive director of the Pennington Biomedical Research Center, an NIH-funded institution in Baton Rouge, La., who co-wrote a commentary accompanying the JAMA study, estimated about 5% to 10% of the U.S. population may be ill-served by BMI.
Prof. Lopez-Jimenez thinks far more people are misclassified, pointing to a 2008 study he co-wrote in which he calculated body-fat percentages for 54 men and 54 women who were found by a CDC study to have a BMI of 25 — the threshold between normal weight and overweight. Their percentage of body fat ranged from 13.8% to 35.3% in men, and from 26.4% to 42.8% in women.
Other researchers defend BMI and say the JAMA study’s counterintuitive findings are the result of the researchers’ inability to control for all diseases and to isolate the study to young, healthy subjects. "BMI isn’t the problem," said Walter Willett, chair of the department of nutrition at the Harvard School of Public Health. "This study lumps together everybody."
Dr. Graubard said the study’s findings hold up among studies that control for more factors.
Even if other measures catch on, researchers likely will continue to use BMI, if only because that is what is most readily available from past studies, and it allows for comparability. "It doesn’t take much — just measuring height and weight," said Prof. Lopez-Jimenez. "Any research study will measure that."
Journal of the American Medical Association, Jan 2013
Diabetes In Control. A free weekly diabetes newsletter for Medical Professionals. News and information for Medical Professionals.