Diabetes. Cancer. Heart disease. Stroke. Extra pounds raise the risk of nearly every health threat facing Americans. Yet, the Journal of the American Medical Association (JAMA) published—and publicized—a study suggesting that overweight people live longer.
“It’s extremely frustrating,” says Michael Thun, vice president emeritus for surveillance and epidemiology research at the American Cancer Society. “It perpetuates a myth.” Here’s how.
At first glance, the JAMA study seems impressive.
“When we assembled all 97 studies, we had almost three million participants,” explained the lead author of the meta-analysis, Katherine Flegal, senior research scientist at the National Center for Health Statistics of the Centers for Disease Control and Prevention in Hyattsville, Maryland, in a video on the JAMA Web site.
“When people talk about things being controversial, they tend to cite only one or two studies,” noted Flegal. “This is the first study that assembles all the literature together in one place.”
How could such a large, comprehensive meta-analysis be wrong?
Flaws in the data
The meta-analysis should have excluded current or former smokers.
To say that overweight people live longer ignores their higher risk of diabetes, heart disease, and cancers of the breast, colon and rectum, esophagus, kidney, pancreas, and uterus. So why would the normal-weight people in the meta-analysis be more likely to die than the overweight?
For starters, any group of normal- weight—or underweight—people is likely to include many smokers, who tend to be thinner. They can make the people in those groups appear more likely to die.
To address the problem, most of the studies in Flegal’s analysis tried to statistically “adjust” for smoking.
“But adjusting for smoking just doesn’t take care of the problem,” says JoAnn Manson, chief of preventive medicine at Brigham and Women’s Hospital in Boston. Why?
To fully adjust for smoking, researchers need to compare the risk of dying among people with equal exposure to cigarette smoke. But few studies ask enough about smoking habits to do that.
“Most studies have limited information on how many cigarettes people smoke, how many years they’ve smoked, and how deeply they inhale, and few studies update the information over time,” explains Manson, who is also professor of medicine at Harvard Medical School.
What’s more, “some people who say that they’re former smokers may have re-lapsed and are actually current smokers,” she adds. “Many people who use smoking as a weight-control tool keep going back and forth.”
Solution: look only at people who have never smoked.
Once you look at never-smokers, it becomes clear that suggesting overweight people live longer is just not true, notes Manson.
Flegal argues that some of the studies in her meta-analysis did look at never-smokers, and it didn’t matter. “The results adjusted for smoking and the results for never-smokers showed almost no difference,” she says.
But most of those studies were too small to see a difference, counters Manson, who notes that larger studies do show a difference.
“Also, it’s almost a moot point to ask, ‘What’s the best body weight for a smoker?’” she points out. “Smoking is such a hazardous, powerful risk factor that it dwarfs the effect of anything else. Smokers should focus on quitting, not on their weight.”
Additional flaws in the data
The meta-analysis should have excluded the sick.
Smokers are one problem. People who are thinner because they have hidden cancer, emphysema, dementia, or other illnesses can also end up in the normal-weight group.
“Although the new meta-analysis pulls together many studies, it has the same problem of including people who are sick and have weight loss caused by disease,” says the American Cancer Society’s Michael Thun.
That moves high-risk people into the normal-weight range, and consequently it makes it look like overweight people live longer. But that is essentially a methodologic artifact.”
The meta-analysis should have looked at different ages.
“This meta-analysis had no ability to look at younger and older age groups separately,” says Harvard’s Walter Willett.
That matters because many older people lose weight before they die, but the weight loss didn’t cause them to die.
“Clinicians have known for hundreds of years that it’s a really bad sign when older people start to lose weight,” says Willett.
Better studies were ignored
What frustrates many experts isn’t just that the meta-analysis was flawed. It’s that better studies didn’t make a splash.
One of the best, a huge collaboration led by the National Cancer Institute (NCI), was released in 2010 in the New England Journal of Medicine.
“The investigators from 19 different studies submitted their original data to the NCI,” says Willett, one of 33 co-authors. “The NCI had the active involvement of many of the top epidemiologists around the country. The collaboration provided a very powerful and very detailed look at this issue.”
The NCI collaboration included 1.46 million people who were tracked for roughly 10 years. “It showed very clearly that people who are overweight have higher risks of dying than those in the lean group,” notes Willett.
Other large studies have come up with the same results. For example, in 2006, the National Cancer Institute published an analysis on 527,000 men and women in the National Institutes of Health-AARP study. And in 1999, the American Cancer Society (ACS) published data on one million
participants in its Cancer Prevention Study. Both found that normal-weight people were least likely to die.
Living longer isn’t living well
If something raises your risk of dying, that may sound like the whole ball of wax, but it’s not.
“All around the world, life expectancies in developed countries are going up, but health expectancy is lagging way behind,” says Yale’s David Katz, referring to a recent report funded by the Bill & Melinda Gates Foundation called “The Global Burden of Disease Study 2010.”
“Modern medicine and modern technology are remarkably good at staving off death, and as we rely more and more on interventions and technology, we have this tendency to look at forestalling death as the goal,” adds Katz. “But if you’re not living well, you’re not really living.”
And who wants to get cancer, even if it doesn’t kill you? Even if it were true that overweight people live longer, they also have a higher risk of cancers of the breast, colon and rectum, esophagus, kidney, pancreas, and uterus. Evidence suggests that that may also be true for cancers of the gallbladder, liver, cervix, and ovary, as well as multiple myeloma, non-Hodgkin lymphoma, and aggressive forms of prostate cancer.
Waist matters more as we age
Could extra pounds be unhealthy when you’re younger, but healthy when you’re older?
Not if the excess weight is fat. “Even in older age groups, people with excess body fat have a higher risk of diabetes, heart disease, stroke, and cancer,” says Harvard’s JoAnn Manson.
But older people may have excess body fat even if they’re normal weight.
“In older individuals, weight doesn’t reflect fatness as reliably because you lose muscle as you age,” says Manson. In other words, a normal-weight 50-year-old may be muscular, while a normal-weight 80-year-old may have a big belly.
And belly fat matters. “A higher waist circumference still predicts a higher risk of dying,” notes Manson.
If you’re interested in your own health and well-being, then you should be interested in the effect of weight on disease, not the effect of disease on weight.”
The bottom line
Even if there were no flaws in the JAMA meta-analysis, and it is accurate that overweight people live longer, how could health experts encourage people to gain weight but stop before they cross the line from overweight to obese? “Being overweight puts you at risk of becoming obese,” says Thun.
What’s more, flip-flops lead the public to start to wonder if scientists get anything right.
“We’re doing a lot of harm by treating science like a ping-pong ball,” says Katz. “I feel very strongly that that needs to stop.”
Sources: JAMA 309: 71, 2013. N. Engl. J. Med. 363: 2211, 2010. N. Engl. J. Med. 355: 763, 2006. N. Engl. J. Med. 341: 1097, 1999. N. Engl. J. Med. 333: 677, 1995. Lancet 380: 2224, 2012. N. Engl. J. Med. 355: 763, 2006. 4 N. Engl. J. Med. 341: 1097, 1999. www.iom.edu/Reports/2013/US-Health-in-International-Perspective-Shorter-Lives-Poorer-Health.aspx. www.cancer.org/research/cancerfactsfigures/cancerpreventionearlydetectionfactsfigures/index. N. Engl. J. Med. 348: 1625, 2003. www.cdc.gov/obesity/adult/causes/index.html. www.iom.edu/Reports/2012/The-Role-of-Obesity-in-Cancer-Survival-and-Recurrence.aspx.