Health beyond the headlines
fatchance

Fat Chance!

For Evidence, One Size Does Not Fit All in the Public Health War on Obesity

By Dana March

Published January 17, 2013

A study that casts doubt on whether being overweight is associated with a shorter life span has sparked 2013’s first public health controversy—and a fiery one at that.

Published in the Journal of the American Medical Association (JAMA) by researchers at the Center for Disease Control and Prevention’s National Center for Health Statistics, the study found that individuals who have a body mass index that classifies them as overweight lived about six percent longer than those considered to be normal weight, contradicting a widespread belief that those who are overweight have shorter lives. The study also found that the lives of individuals in the obese BMI category were 18 percent shorter compared to those of normal weight.

That obesity shortens life is consistent with the current campaign to trim the population’s collective waistline. That being overweight might lengthen life is not. And although the study is peer-reviewed and scientifically sound, several experts have taken issue with the methods as well as the findings, claiming they could undermine the current full-scale public health effort to combat the obesity epidemic. The debate has revealed some of the major fault lines over how health authorities communicate to individuals about weight loss, and represents more about dogmatic beliefs regarding health than the actual merits of the study. It also sets into relief some of the risks of resisting the introduction of new scientific knowledge in the context of a public health war.

The Skinny on the Study

BMI, a widely used measure around which standard international weight categories are constructed, is controversial. The measure, which is calculated by dividing weight in kilograms by height in meters squared, has been criticized for being an inadequate measure of health, particularly regarding who is considered overweight and obese.

So while it probably is not surprising that actor Marlon Brando is considered obese—not the Brando who sizzled in A Streetcar Named Desire but the Brando of Apocalypse Now—it would probably surprise most people that actor Tom Cruise, at 5 feet 7 inches tall and 166 pounds is considered overweight according to his BMI. Two less surprising comparisons: singer Beyonce Knowles at 5 feet 6 inches tall and 130 pounds is considered normal weight, and actress Angelina Jolie at 5 feet 8 inches tall and 115 pounds is considered underweight.

And although it may be imperfect, BMI does have its defenders. A recent study conducted by Columbia epidemiologists found that it is as good as any other measure when predicting many health conditions.

The JAMA study, which was a systematic review, or a study of studies, sought to address controversial inconsistencies in the scientific literature addressing BMI-based weight categories and risk of death.

Despite its rigor, sound methods, and transparent presentation, the study has elicited a spectrum of reactions from scientists and other commentators, many of whom take issue with the study for opposing reasons.

“This study is really a pile of rubbish and no one should waste their time reading it,” Dr. Walter Willett of the Harvard School of Public Health, told NPR. “We have a huge amount of other literature showing that people who gain weight or are overweight have increased risk of diabetes, heart disease, stroke, many cancers and many other conditions.”

Dr. Willett did not identify any methodological grounds for dismissal of the study. To Willett, it seems, the study was flawed because it did not address the association between overweight, obesity, and chronic disease.

However, the study focuses explicitly on risk of death from all causes, including chronic diseases, unintentional injuries, and suicides. Some causes of death, particularly diseases that lead to a form of severe weight loss known as wasting, may pose greater risks for normal weight individuals than those who are overweight.

Coming from the other end of the spectrum, Paul Campos, author of The Obesity Myth, seized onto the results of the study, writing in an op-ed piece in The New York Times that the fear of fat is “absurd” in the United States, and that the study’s results provide evidence that governmental claims of Americans’ expanding waistlines is “exaggerated and unscientific.”

Campos also claims that “the American obsession with fat… serves the economic interests of, among others, the multibillion-dollar weight-loss industry and large pharmaceutical companies,” failing to acknowledge also the powerful, symbiotic forces of the food industry.

Neither Willett nor Campos acknowledged that the 18 percent increase in mortality risk for obesity relative to normal BMI reported in the study is consistent with the rest of the literature, which gives some credence to the overweight finding, and provides further evidence that obesity has deleterious effects.

Dr. David Katz, director of Yale University’s Prevention Research Center reacted sharply to Campos’s op-ed, pointing to the perils of playing “ping-pong with science because of misguided bias or motivated self-interest.”

The study is an exhaustive analysis known as a systematic review. This kind of study addresses the inherent challenge in summarizing epidemiologic literature, which is dealing with a variety of comparison categories and methods that often obfuscate the overall pattern of findings.

“This literature has not been assembled in one place before in any kind of way…I think when people talk about things being controversial or questionable, they only tend to cite one or two studies,” Dr. Katherine Flegal, lead author of the study, told JAMA in an interview.

Scientists do not collect their own data for systematic reviews but rather assemble data from other studies on the subject. Because most studies have different measures and definitions, a systematic review provides an unbiased synthesis of the existing scientific literature. It is conducted with methodological rigor, using explicit inclusion and exclusion criteria for potential studies, which means that the review is reproducible.

One approach to the systematic review of quantitative studies, known as meta-analysis, statistically summarizes the data of the studies included in the review. The JAMA study is a meta-analysis, which included data from 97 longitudinal observational studies, known as prospective cohort studies, mostly from the U.S., Canada, and Europe, with a total of almost three million subjects and over 270,000 deaths from all causes.

According to Campos, the studies included in the meta-analysis cannot establish cause. “Observational studies merely record statistical correlations,” says Campos.

However, Campos does not account for the fact that observational studies like those included in the JAMA study can provide valuable, if exhaustive, information regarding whether and how risk factors ultimately influence outcomes, like disease or death. The statistics calculated in these studies are not just correlations—they provide estimates of the degree to which risk factors influence outcomes by holding other factors constant. And, observational studies can be conducted when experimental studies would be inappropriate or unethical.

As Dr. Katz notes, “A meta-analysis is never any better than the studies it is aggregating.” Thus, meta-analyses operate on a “garbage in, garbage out” principle. But the quality of the carefully selected observational studies in the JAMA study seems to have nothing to do with Dr. Willett calling it “rubbish.” After all, Dr. Willett leads a research group that is conducting three of the largest and longest running observational studies of women’s health, the Nurses’ Health Studies, which provided some data for the JAMA meta-analysis, helping to establish that obesity is associated with a shorter life span.

Whether the observational studies included in the meta-analysis demonstrate cause and effect or provide useful information regarding interventions is also a point of contention.

“…[T]here is no reason to believe that the trivial variations in mortality risk observed across an enormous weight range actually have anything to do with weight or that intentional weight gain or loss would affect that risk in a predictable way,” Campos says.

Indeed, some epidemiologists focused on the science of establishing cause and effect in observational studies, like Dr. Miguel Hernán of the Harvard School of Public Health, have questioned the validity of these studies that address BMI and mortality. In a 2008 paper, for example, Dr. Hernán notes,“BMI…is the potential result of many different types of interventions.” And different interventions may have different effects on risk of death. Therefore, Hernán argues, observational studies should account for as many factors as possible that influence both BMI and risk of death, such as diet, exercise, smoking, genetic predisposition, and illness. By accounting for these factors, which are often connected, these kinds of studies can better establish cause and effect and inform interventions.

The Language of War and Intervention in the Battle of the Bulge

Still, the current controversy and public debate sparked by the JAMA study seems to turn not on methods, but on the finding that overweight people might actually live a little longer. And being overweight is one step (or lack thereof) closer to being obese.

“…[A]t the population level, epidemic obesity is incontrovertibly established as a clear and all-but-omnipresent danger,” Dr. Katz writes, much in the same way a military official might write about terrorism.

It is this very construction of a disease or a condition that sheds light on the discomfort with scientific findings that are inconsistent with the current Zeitgeist in public health: we are at war with obesity.

Indeed, the lexicon of war is manifest in public health—the war on drugs, the war on tobacco. Public health, like war, involves campaigns against diseases, disorders, or conditions we deem our enemies. As in military wars, our objective in public health campaigns is victory.

In no war have the leaders of our armed forces embraced partial defeat. Similarly, no public health campaign has ever had as its goal a partial victory. For example, in the war on tobacco, the public health messages focus on elimination of the exposure: quit smoking, or don’t ever start. The message is not, “Smoking Marlboro Reds occasionally is fine.” Similarly, in the war on drugs, moderation does not feature into public health messages in the United States, in contrast to other countries, like Denmark, which support needle exchange programs to reduce the incidence of diseases transmitted through shared needles in intravenous drug use. The Just Say No campaign offers evidence of that.

In yet other areas, like the campaign to reduce salt intake, public health may even ignore scientific evidence in order to wage a war.

However, engaging in a public health campaign and dismissing evidence that challenges our course of action in population health is akin to waging a ground war with incomplete intelligence or no intelligence at all. In the case of the public health war on obesity, findings like those of Flegal and her colleagues are valuable because they are consistent with the obesity message while at the same time urging further research with regard to the impact of being overweight, the intermediate between normal weight and obesity, on mortality and its causes. Is it truly protective? If so, why? Are there methodological or biological reasons, or both, for this observation?

Unfortunately, resistance may hamper such research. If the overweight finding were simply that being overweight does not increase mortality risk, scientists and commentators might be somewhat less dismissive of the study. That the study found overweight individuals lived longer—a protective effect—poses a particular challenge in public health for two reasons.

First, we are traditionally more comfortable with findings that show an increase in the exposure (here, BMI) is related to an increase in the outcome. We would be most at ease with a study that showed that being overweight shortened life, and being obese shortened life even more.

Second, we are traditionally more comfortable with harmful exposures because public health is good at eliminating exposures or taking action that prevents the development of disease in response to exposure—think smallpox eradication.

Within this context, scientific, political, and social discomfort with the evolution of knowledge, particularly when that involves shifts in our understanding of risk known as crossovers—going from harm to protection or the reverse—is both palpable and potentially dangerous.

For example, the evidence addressing hormone replacement therapy (HRT) and health has evolved markedly, and indeed, has crossed over twice—HRT went broadly from good to bad, and now it’s back to good. A recent reevaluation of the evidence that shows that the benefits of HRT exceed the risks. These findings have marked implications for clinical practice, which has been shaped sharply by the shifting evidence base. Moreover, the evolution of HRT science has implications for whether and how we embrace—or reject—scientific findings that counter our current clinical practice and public health action.

It is for this reason that dismissing scientific findings that are contrary to current public health messages may be perilous. We risk not evaluating potentially important mechanisms, thereby missing potentially effective courses of clinical and public health action.

Our problems with obesity have particular contours, as do our consumption behaviors regarding food—and science in service of public health. The findings reported by Dr. Flegal and her colleagues in the recent JAMA meta-analysis, and the spectrum of reactions to them, demonstrates clearly that for evidence in the war on obesity, one size does not fit all.

Edited by Elaine Meyer. Additional research by Joshua Brooks.

Elevate the conversation

 
The views and opinions expressed on this website are solely those of the authors and do not represent those of the Department of Epidemiology, the Mailman School of Public Health, or Columbia University.