Katherine Flegal Discusses the Prevalence of Obesity in the US
Interview From the Special Topic of Obesity, September 2010
In our Special Topics analysis on Obesity, the work of Dr. Katherine Flegal ranks at #3 by total cites, based on 44 papers cited 13, 474 times over the past decade. Four of these papers appear on the top 20 list for the past decade. She also ranked among the top 20 researchers in our 2001 analysis of obesity research.
Her record in Essential Science IndicatorsSM from Thomson Reuters includes 54 papers, the majority of which are classified in the field of Clinical Medicine, cited a total of 16,327 times between January 1, 2000 and April 30, 2010.
Flegal is a Senior Research Scientist and Distinguished Consultant of the National Center for Health Statistics of the Centers for Disease Control and Prevention in Hyattsville, Maryland.
Your work is credited with revealing the existence of the obesity epidemic. How did you get started on this line of research and why?
I went to graduate school in nutrition at Cornell and I was interested in waist-height indices. I was interested in anthropometry. I started working in 1987 at the National Center for Health Statistics (NCHS), which had just become part of the Centers for Disease Control and Prevention.
This was just when the third National Health and Nutrition Examination Study began—NHANES III. The NHANES surveys involve interviews and health examinations in mobile units and are designed to give national estimates for a lot of health measures, including height and weight. The first half of NHANES III covered the years 1988 to 1991.
We realized that NCHS hadn’t published obesity prevalence journal articles from NHANES before. These numbers were kicking around here and there, but there was no actual journal article reporting what they were, so we thought we should write one for the NHANES III results. At that point, obesity prevalence had been pretty stable for about 20 years and we didn’t expect to see much, if any, change.
A picture of the NHANES trailers with one of the interview teams for NHANES.
But we ran the numbers, and said, "Oh look, it really went up a lot." That was a big surprise! We thought we probably had some methodological problem. We have three mobile units doing these examinations. We thought maybe something had gone wrong, maybe the scale was off in one of the mobile units, or we had some data processing problem.
But we couldn’t find anything wrong, so we published an article in 1994 in JAMA showing a fairly large increase in the prevalence of obesity. The increase was eight percentage points so it was not huge, but it was noticeable and surprising. Bob Kuczmarski was the first author and we used the old definitions at the time—a body mass index of 27.8 for men and 27.3 for women. (Kuczmarski RJ, et al., "Increasing prevalence of overweight among US adults—the National Health and Nutrition Examination Surveys, 1960 to 1991," JAMA 272: 205-11, 20 July 1994).
How was the paper received?
It caused quite a commotion. People were pretty surprised by the increase. We were very nervous when we published these results. We couldn't find any published data that confirmed the increase. We couldn't find anything that showed there was an increase in obesity. "Well," we thought, "hopefully something isn't really wrong here." We hoped that we weren’t publishing something that would embarrass us later.
You could find no other studies that covered that time period?
The only published information for the US that we could find was the Minnesota Heart Health Study, which was about the same time period and showed an increase. That was all we had to go on. Nobody had made any comments about this. It was not that long ago—only 16 years ago. People weren’t talking about increases in obesity. There was hardly anything out there that we could find.
After I knew what our findings were going to be but before we published them, I would go to meetings and casually ask people, "Oh, what do you think is going on with the prevalence of obesity," and no one ever said they thought it was climbing.
Of course, after we published this, lots of people told me that they had seen it from their data, but at the time nobody I had talked to admitted knowing this. So that launched my career into the prevalence of obesity.
Did you get locked into obesity research or was this just a part of your research?
No. I do a lot of other things, too. But we figured people are always interested in this topic, and if we didn't publish it, somebody else would. And we'd probably do a better job. We work in the survey program that produces the data, so we have the best grip on the design of the survey and the best understanding of the data. We're probably more responsible. We don't have any ax to grind. Basically our prevalence articles are all "just the facts, ma'am" articles.
In your 1998 article on the obesity epidemic (Flegal KM, et al., "Overweight and obesity in the United States: prevalence and trends, 1960-1994," Int. J. Obesity 22: 39-47, January 1998), the definitions of obesity changed and became what we're used to today—BMI over 30 for both men and women. Why was that?
There was a whole variety of reasons. There's a whole history to this. I myself went to a European conference in the early 1990s and realized the US was using cut-off points that nobody else was using. Other countries were using a BMI of 30 to define obesity.
Then, in 1995, there was a WHO expert committee report which defined different grades of overweight—grade 1 was BMI 25 to 30, grade 2 was 30 to 40, and then grade 3 was 40 and above. In 1997, there was another WHO report and they changed the terminology so overweight is a BMI of 25 and above and obesity is a BMI of 30 and above.
"We found that the excess mortality associated with obesity is primarily from cardiovascular disease..."
Related to that was an effort in the US by an expert committee of the NHLBI, and they basically used the same cutoff points, so these definitions became federal policy, and are really international policy at this point. We have a vested interest in trying to use definitions that agree with what other people are using, regardless of the other properties of those definitions, so we switched over.
Is there something particularly meaningful about a BMI of 30?
When you look at how these definitions come into being, you can never pin down exactly what the criteria are. I don't know of any research that says we're going to specify some criteria and show that a BMI of 29 doesn't meet our criteria, but a BMI of 30 does. It's just that this is a definition that we agree on.
Jeff Friedman of Rockfeller University has written several articles stating that the entire obesity epidemic is caused by a relatively modest average weight gain—maybe a dozen pounds—across the population. Is this a valid way of looking at it?
One of my projects is to write a paper about that. I have all the graphs. There are two effects at work. One, of course, is that if you have a normal distribution and you shift the distribution to the right by a fixed amount, the median is going to increase and the prevalence of obesity is going to increase.
But what you can see in the data is that both the median is moving and the distribution is becoming more skewed. The tails are longer and longer. It's not just a shift of the distribution. It's a shift and a change in the shape of the distribution. The median has gone up by a certain amount and the prevalence of obesity has gone up more than you'd expect from the change in the median.