Why Scientists Can’t Agree on Whether It’s Unhealthy to Be Overweight

Some studies show being overweight leads to a greater risk of death; others show it doesn’t. Here’s what’s really going on.

A person stands on a scale displaying a frowning face.
Nico De Pasquale Photography / Getty

Is being a little bit overweight bad for you? Could it lead to an untimely death?

It’s a question with real consequences. Many overweight people feel locked in a fruitless battle with their size. If they do slim down, the process might distort their metabolisms forever. But if they remain overweight, non-thin people may face intense prejudice and stigma, as the writer Taffy Brodesser-Akner poignantly described in The New York Times Magazine recently:

I was in Iceland, for a story assignment, and the man who owned my hotel took me fishing and said, ‘‘I’m not going to insist you wear a life jacket, since I think you’d float, if you know what I mean.’’ I ignored him, and then afterward, back on land, after I fished cod like a Viking, he said, ‘‘I call that survival of the fattest.’’

The “health at every size” movement, though, has its own pitfalls, and not just because it can come off as oddly objectifying. American life expectancy recently dipped slightly, and obesity might be part of the cause. Telling people it’s perfectly fine to be dozens of pounds overweight would be terrible advice—if it’s wrong.

Most researchers agree that it’s unhealthy for the average person to be, say, 300 pounds. They don’t really know why being very overweight is bad for you, but the thinking is that all those fat cells disrupt how the body produces and uses insulin, leading to elevated glucose in the blood and, eventually, diabetes. Extra weight also increases blood pressure, which can ultimately damage the heart.

But whether just a few extra pounds raise the risk of death is a surprisingly controversial and polarizing issue. Usually, nutrition scientists tell journalists hedgy things like, “this is just what my study shows,” followed by the dreaded disclaimer: “Further research is needed.” But on this question, the researchers involved are entrenched, having reached opposite conclusions and not budging an inch. Like many internecine wars, the dispute mostly comes down to one small thing: how you define the “overweight” population in the study.

Over the years, myriad side controversies—personal attacks, money from the Coca-Cola Company, and a debate over who is truly “overweight”—have deepened the divide. But they haven’t clarified things.

* * *

It all started in 2004, when the Centers for Disease Control and Prevention scientists published a study suggesting obesity was responsible for 400,000 deaths a year, making it almost as deadly as smoking. It turned out to be a false alarm: The authors made methodological errors that skewed their number too high.

But a CDC senior scientist named Katherine Flegal was already working with a small group of her colleagues to write a different obesity paper using better data and better methods. In 2005, they published their results, and their estimate was substantially lower: Obesity was only responsible for about 112,000 excess deaths. They also found something peculiar. Being “overweight,” but not obese, was not associated with an increased risk of death at all.

Millions of despairing dieters likely sighed with relief, perhaps celebratorily pouring a SlimFast down the drain. But while Flegal’s study was praised by some researchers, others were skeptical, saying past research had already shown that the heavier you are, the greater your risk of dying. "We can't afford to be complacent about the epidemic of obesity," JoAnn Manson, the chief of preventive medicine at Brigham and Women's Hospital in Boston, told The New York Times after Flegal’s study came out.

Flegal pressed on, and in 2013 she and colleagues published a meta-analysis—a study of studies—that replicated her earlier findings. Even when adjusting for smoking, age, and sex, overweight people—those with a body mass index of between 25 and 30—had a 6 percent lower risk of dying than normal-weight individuals. Body mass index, or BMI, is a measure of a person’s weight divided by their height. Her paper found that in terms of mortality, it’s better for this number to be slightly elevated than to be normal. A 5-foot-6-inch woman, in other words, would be better off weighing 180 pounds than 120.

A “pile of rubbish” is what Walter Willett, a Harvard University professor of epidemiology and nutrition, deemed that paper. Willett has co-authored studies finding the opposite effect. He and Andrew Stokes, a demographer at Boston University, say Flegal’s work suffers from a problem they call “reverse causality.” They think that because she didn’t examine her subjects’ entire weight history, her study didn’t control for people who used to be overweight, but became normal-weight because they got sick before they died. They argue her study conflates normal-weight, healthy people with formerly overweight people who lost weight due to liver disease, cancer, or some other illness. Having those individuals in the pool of normal-weight people makes the normal-weight people seem sicker, and the overweight people seem healthier, than they actually are.

“I think Kathy Flegal just doesn’t get it that people often lose weight before they die,” Willett told me.

In 2016, Willett and dozens of other researchers from around the world published a paper in The Lancet analyzing 239 studies and millions of study subjects. Their takeaway was clear: Above the normal weight range, the fatter you are, the higher your risk of premature death. “On average, overweight people lose about one year of life expectancy, and moderately obese people lose about three years of life expectancy,” the paper’s lead author, Emanuele Di Angelantonio, told The Guardian.

Flegal takes issue with how Willett and his colleagues selected the studies for their review. “It seems like they took studies they already knew about and that gave the answers that they preferred,” said Flegal, who is now a consulting professor at Stanford.

Besides, other studies have since implied there’s a health benefit to heaviness. Last year researchers in Copenhagen looked at three cohorts of Danes during the 1970s, ’90s, and between 2003 and 2013. In the 1970s, the BMI that was associated with the lowest risk of death was 23.7—so-called normal weight. Surprisingly, by the 2000s, the “healthiest” BMI had shifted up to 27, or technically overweight.

Børge G. Nordestgaard, a clinical professor at the University of Copenhagen and an author of that study, speculated that this could be because over time, doctors have gotten better at treating some of the side effects of excess weight, like high blood pressure and high triglycerides.

Or, “it could just be that as the population has become more overweight and obese, the people who are in the middle of the BMI distribution, these are the most ‘normal’ people, they are the ones who do all the most normal things,” Nordestgaard said. “They are the ones who survive the best.”

What’s more, in 2014, New Orleans cardiologist Carl Lavie published the book The Obesity Paradox: When Thinner Means Sicker and Heavier Means Healthier, based in part on his research showing that overweight and mildly obese patients with cardiovascular disease have a better prognosis than their leaner counterparts.

But when reporters found that Lavie had received money from the Coca-Cola Company for speaking and consulting on obesity, it fueled speculation that junk-food companies are promoting the supposed benefits of obesity in order to evade blame for causing it. (In an email, Lavie said Coca-Cola only funded a few of his lectures, of which he gives more than 100 a year.)

Andrew Stokes, the demographer at Boston University, says some of most vocal supporters of the “obesity paradox” are activists and people with vested interests. He’s found that the paradox disappears when “normal weight” is defined as only those people who have remained thin over time, as opposed to those who entered the normal-weight category after losing weight due to an illness. In a paper published this April, Stokes, Willett, and others found being overweight was associated with mortality—but only if you looked at a person’s maximum weight over the past 16 years. According to their findings, it’s having ever been overweight that’s risky.

* * *

That’s not the end of the methodological gripes, though. Flegal and others say the self-report data that Willett and Stokes use in some of their studies is not reliable. “It is well-known that underreporting of body weight along with underreporting for females and overreporting for males of height can result in biased BMI’s,” said Barry Graubard, a senior investigator with the National Cancer Institute, which is part of the National Institutes of Health.

Stokes counters that not only has self-report data been found to correspond closely with measured weight, not all of the data refuting the obesity paradox is self-reported. Flegal, meanwhile, thinks Stokes and others haven’t demonstrated that the weight loss was the result of a sickness, or that the sickness-induced weight loss is a big enough problem to taint an entire study. She also thinks his results are consistent with her 2013 meta-analysis, falling “pretty much in the middle of the other studies that we found.” Stokes disputes this. He also began one of our phone calls by asking me if I was regretting doing this story yet.

If a little extra pudge is somehow good for you, it’s not clear why. Some researchers suggest overweight people might be better equipped to fight off certain diseases, with fat serving as a last-ditch fuel for the ailing body. And they point to studies that failed to show that losing weight led to less heart disease in overweight people. Stokes, meanwhile, thinks that explanation is speculative, and it pales compared to the many ways obesity harms health. Even a BMI of 25, for example—just barely “overweight”—has been associated with an increased risk of diabetes.

There’s also the idea that some people we now consider “overweight”—say, a 6-foot, 1-inch man who weighs 200 pounds—don’t actually have too much fat. For one thing, athletes and other very muscular people might be wrongly categorized as overweight, and some scientists now think it’s stomach fat, not hip fat, that’s the dangerous kind. What’s more, in 1998 the NIH revised down its BMI threshold for “overweight” to 25, from 27.8 for men and 27.3 for women, in order to better align with the rest of the world.

“I think it was the French who pushed [the lower cutoff],” said Judy Stern, an emeritus professor of nutrition at the University of California, Davis, and a member of the advisory panel that worked on the new guidelines. “The French always push.” She thinks it might have had something to do with different standards of beauty around the world. “In general, in Europe, it’s better to weigh less. When Americans go to Europe and we weigh more, we’re viewed as not as beautiful.” (She voted against the new guidelines.)

The new standard means that “if you showed someone with a 26 [BMI] had no excess mortality in 1996—there would be no question,” Flegal said. She speculates the change was made to emphasize the seriousness of the obesity epidemic, and she notes that her critics have expressed fears her results might lull the public into complacency around obesity. “The problem with my research is apparently just that I did it,” she said. “That’s not science.”

But there’s a big caveat to this theory. Medical advice urging heavy people to lose weight is based on the premise that being overweight is unhealthy. If Flegal and Nordestgaard are right, and being overweight is linked to less mortality, then should people whose BMIs fall in the normal range gain weight? Should they be guzzling milkshakes in hopes of staving off death? Both Flegal and Nordestgaard said “no.”

“Weight is just one risk factor for most of these conditions, it’s not the risk factor,” Flegal said. She points out that some studies show people with doctorate degrees live longer than those with bachelor’s degrees. “If someone tells me, ‘I have a bachelor’s degree, but I know the risk is lower if I have a doctoral degree,’ should I tell them they should go get a Ph.D.?”

She reiterated something—perhaps the only thing—that epidemiologists who work on this issue can still agree on: “It’s associated. The causality is unclear.”