Updated

The "obesity paradox" - the controversial notion that being overweight might actually be healthier for some people with diabetes - seems to be a myth, researchers report. A major study finds there's no survival advantage to being large, and a disadvantage to being very large.

More than 24 million Americans have diabetes, mostly Type 2, the kind that is on the rise because of obesity. About two-thirds of U.S. adults are overweight, including one-third who are obese.

Weighing too much increases the chances of heart disease, cancer and premature death. But some small studies have suggested this might not be true for everyone, and that Type 2 diabetics might even benefit from a few extra pounds - a "metabolic reserve" to help get them through sickness.

The new research - which looked at deaths according to how much people weighed when they were diagnosed with diabetes - dispels that idea.

"We didn't see this protective effect at all," said one study leader, Deirdre Tobias of the Harvard School of Public Health. "The lowest risk was seen in the normal-weight category."

The National Institutes of Health and the American Diabetes Association paid for the work. Results are in this week's New England Journal of Medicine.

"It's a very convincing study" and large enough to give a clear answer, said one independent expert, Dr. Patrick Remington, associate dean for public health at the University of Wisconsin-Madison.

It involved 11,427 female nurses and male health professionals diagnosed with diabetes sometime after enrolling in two long-running health studies. They were grouped according to body mass index, a measure of height and weight. People with a BMI over 25 are considered overweight, and 30 or higher, obese. A 5-foot-8-inch person would be overweight at 164 pounds and obese at 197.

During more than 15 years of follow-up, there were 3,083 deaths. The lowest risk was among those in the normal range - BMIs of 22.5 to 25.

For the rest, researchers saw a J-shaped curve - deaths trended higher at both extremes. Being just a little overweight did not substantially raise the risk of death, but the trend was in that direction.

The study was big enough that researchers could look at subgroups. For those under 65 when they were diagnosed with diabetes, the risk of death rose directly in relation to BMI. The same was true of people who had never smoked.

Trends for smokers and people over 65 were less uniform. Smokers had higher death rates in general. Smoking suppresses appetite and contributes to lower weight, but contributes so strongly to many diseases that it can overshadow and complicate efforts to measure the effect of weight alone, Tobias said. Older people have many other health conditions that also make it hard to see the effect of BMI.

The results support guidelines urging people to keep a healthy weight, said Dr. Donna Ryan of Louisiana State University's Pennington Biomedical Research Center. Even though being modestly overweight did not raise the risk of dying very much in this study, many others show that it impairs quality of life by contributing to high blood pressure, high cholesterol and other problems, she said.

Ryan said she doubted that overweight people would think it's OK to be a little pudgy.

"Are you kidding?" she said. "Everybody is trying to lose weight."