Skip to main content.

For the healthiest weight, the past is as important as the present

For the healthiest weight, the past is as important as the present

Picture of William Heisel

Being overweight is bad for you, right? Except when it isn’t. A recent study by Katherine Flegal and colleagues in JAMA found that being a little bit overweight is actually the healthiest of all weight categories, and being mildly obese does not appear to raise your mortality risk.

A new analysis of weight data by Andrew Stokes arrives at a different conclusion, though, and the findings are a good reminder to reporters that there are often many ways to slice a data pie. (Disclosure: the new study was published in Population Health Metrics, where I am a co-managing editor.) Stokes, an assistant professor in the Department of Global Health at Boston University, published his analysis in March while still a graduate student at the University of Pennsylvania.

When researchers are studying and writing about weight, they often are relying on surveys where people self-report their weight or data from Medicare or other populations where weight has been collected at a single point in time. Healthy weight is generally measured as the relationship between height and weight, or the body-mass index (BMI).

What Flegal found was that people in the overweight category had a lower all-around mortality level than people in the normal weight category. She also found that people in “obese class one” did not have a higher mortality risk than people of normal weight. Immediately, critics of the study argued that the findings should be viewed with caution because the group studied included sick people and smokers. Illness and smoking can confound the results because they lead to both lower weight and higher mortality.

Stokes then backed that critique up with his own analysis. He wrote, “Using BMI at the time of the survey to assess the mortality risks of overweight and obesity is problematic, especially in older populations, because slimness can be a marker of illness.”

Even if you try to address the bias by eliminating sick people from the sample, you can end up with mixed results because weight loss often occurs before an illness has been diagnosed.

What to do?

Stokes argues that the solution lies in incorporating weight histories when studying the link between weight and mortality. This makes it possible to distinguish people who have always been slim from those who were formerly overweight or obese and lost weight due to an illness.

Stokes’ study used data on maximum lifetime weight from the CDC’s National Health and Nutrition Examination Surveys, conducted between 1988 and 1994 and between 1999 and 2004. Then he linked those findings to the death records of U.S. adults ages 50 to 84.  

With this method, Stokes found that 33 percent of deaths among adults who never smoked could be tied to people being overweight or obese. So why did Flegal’s study and others find a totally different pattern? It goes back to how much people weighed the day they were surveyed versus what they weighed in the past. People who were overweight or obese in the past but normal weight at the time of the survey had much higher mortality rates than those who had stayed at a normal weight throughout adulthood. 

In Flegal’s study, “normal weight” included everybody who had a normal weight on the day they were surveyed or measured, regardless of how much they weighed in the past. In Stokes’ analysis, he only classified people as “normal weight” if they had a stable normal weight over time. As Stokes said, slimness can be a marker of illness. 

Walter Willett, who chairs the nutrition department at the Harvard School of Public Health, praised the approach Stokes took. He called it “simple and intuitive.” He wasn’t so kind when referring to studies such as Flegal’s, which he called "complete rubbish." He told Don Sapatkin at the Philadelphia Inquirer that when trying to figure out the best weight for long-term health, “you should not have that answer be influenced by data from someone who has lost 20 pounds and is dying of cancer."

Dr. Yoni Freedhoff of the Bariatric Medicine Institute and the Weighty Matters blog told Stokes on Twitter, “Loved your piece on obesity paradox and use of max weight. Smack-your-forehead smart.”

But veteran science writer Wayt Gibbs – the editor in chief behind the lab-meets-kitchen Modernist Cuisine books – had some doubts. He wrote on Twitter: “But removing this ‘bias’ produces results that are only relevant in an ideal world where no one dies of wasting diseases!”

Stokes responded: “That's the beauty of using weight histories. People with illnesses are not excluded.”

So before you write that being a little bit heavy might make you a little bit healthier, consider the evidence a little more carefully. 

Photo by u-o-0024 via Flickr.


Picture of

Do they know if those who had lost weight lost it through smoking or illness? Without that information, how do they know if it isn't weight loss (diets, restriction, yoyo dieting, ect...) that causes bad health? Is the percentage of people who have wasting diseases (even if not diagnosed) significant enough to account for this? Could maintaining a stable weight, no matter what that weight is, be the most important part?

It seems to me (of course, I have not read the study) that this study isn't better than the other one- it seems like it's all just correlations and people are guessing at the causes...


The Center for Health Journalism’s two-day symposium on domestic violence will provide reporters with a roadmap for covering this public health epidemic with nuance and sensitivity. The first day will take place on the USC campus on Friday, March 17. The Center has a limited number of $300 travel stipends for California journalists coming from outside Southern California and a limited number of $500 travel stipends for those coming from out of state. Journalists attending the symposium will be eligible to apply for a reporting grant of $2,000 to $10,000 from our Domestic Violence Impact Reporting Fund. Find more info here!


Follow Us



CHJ Icon