What's Wrong With Fat? (23 page)

However, none of the news reports drew on Blair’s research to criticize other articles in the 1999
JAMA
issue for focusing exclusively on body weight without rigorously statistically controlling for physical fitness, as Blair’s research suggests they should. This is not surprising given that science journalists, who tend to lack scientific training, do not typically independently evaluate the validity of scientific research. 50 But it was also largely driven by the authors themselves, who did not seem especially intent on directly challenging other research reports in the theme issue or in the larger field of obesity research. “I don’t mean to leave diet out completely,” Blair was quoted as saying. “Big portions and high-fat foods are a problem. But let’s not obsess so much about our weight and focus on getting exercise.” 51 The study’s author thus toned down the radical implications of his work, while also conflating diet and body weight.

In other words, in the cases I examined, the news media were not the instigators of scientific controversy and did not independently highlight scientific uncertainty. Instead, they tended to stick closely to the individual studies on which they were reporting at any one time. If a scientific controversy were to emerge in the news media, it seemed likely that it would have to be framed as such by scientists themselves. I began wondering about the conditions under which a scientific controversy over the health risks associated with “obesity” might capture the attention of the news media. I further wondered whether a study that seriously challenged the public health crisis narrative would lead the news media to frame fat differently. I wondered how, faced with contradictory claims from scientific authorities, the media would weigh these claims and assess credibility.

I kept an eye out for a scientific controversy that would allow me to answer these questions. I wanted to identify two research studies that arrived at very different conclusions about, say, the extent to which obesity represents a public health crisis. Ideally, teams of scientists affiliated with equally prestigious institutions would pen both studies and publish them in equally prestigious scientific journals, so as to be able to hold prestige of author and publication constant. If I were able to identify two such studies, I thought that I could then examine the extent to which the news media treat the two sets of findings and their authors in systematically different ways. Of course, I could not make such studies appear. I could only wait and hope.

As it so happened, two studies that perfectly correspond to these criteria were published in 2004 and 2005. Two different groups of CDC scientists authored both studies, and the
JAMA
published both of them, effectively holding constant the credibility and visibility of both researchers and journal. And yet, they came to strikingly different conclusions. The 2004 report, which I call the Eating-to-Death study, estimated that overweight and obesity contributed to hundreds of thousands of excess deaths in the year 2000. In contrast, the 2005 report, which I call the Fat-OK study, concluded that overweight is associated with
fewer
deaths than normal weight and that the number of excess deaths associated with obesity in 2000 were much fewer than previously thought. 52

The Eating-to-Death study was a replication of “Annual Deaths Attributable to Obesity in the United States,” one of the articles in the 1999 theme issue of
JAMA
discussed above, and drew on several large-sample studies, many of which were not representative of the U.S. population. The bulk of deaths claimed to be associated with “poor diet and inactivity” was actually estimated based on calculations of the number of excess deaths observed in the overweight or obese categories, compared to what would have been expected if all of these people had a BMI in the upper range of the “normal weight” category (BMI 23–25). In other words, the Eating-to-Death study assumed—rather than tested—that overweight and obesity result from poor diet and inactivity. They further assumed that, if people in the overweight and obese categories were dying at a higher rate than those at the upper end of the “normal weight” category, it was due to their poor diet and physical inactivity. Science and technology scholars use the term
coproduction
to refer to this sort of approach, in which the investigation of a phenomenon uses techniques that presume much about the very nature of the phenomenon. 53

Using this method, the authors of the Eating-to-Death study attributed 385,000 excess deaths in the year 2000 to overweight and obesity combined, which was higher than the earlier annual estimate. The researchers then estimated that poor diet and physical inactivity caused an additional 15,000 deaths in 2000 among those who are not overweight or obese but did not explain how they arrived at this figure.

While the authors of the Eating-to-Death study spoke of deaths attributed to “poor diet and inactivity,” albeit based almost exclusively on data about mortality rates by BMI category, the news media typically pinned all of the 400,000 estimated excess deaths on
overweight and obesity
.
The research study’s authors’ statement that “if the increasing trend of overweight is not reversed over the next few years, poor diet and physical inactivity will likely overtake tobacco as the leading preventable cause of mortality” was widely paraphrased as a statement of how
overweight and obesity
would overtake tobacco as the leading cause of preventable mortality.
This provided the justification for a U.S. government–sponsored obesity education campaign, strategically launched shortly after publication of the study. 54 These findings were also cited in Secretary of the Department of Health and Human Services Tommy G. Thompson’s testimony before the U.S. Senate Committee on Appropriations, as justification for funding for NIH-sponsored obesity research and CDC-sponsored obesity prevention programs. 55
After a few computational errors were brought to their attention, in the following year, the authors revised their final estimate down slightly to 365,000. 56

The Eating-to-Death study represented precisely the type of dramatic finding—400,000 people eating themselves to death each year—that we have seen makes for compelling news stories. It is therefore not surprising that this study generated considerable news media attention. The Fat-OK study might have gone unnoticed, in that its findings, that there were less than 26,000 excess deaths associated with obesity and overweight combined in the year 2000, were not especially alarming. Yet, because it was presented as directly challenging the earlier study, the Fat-OK study was extensively covered as indicating a scientific controversy, as in the following news report: “Published in the
Journal of the American Medical Association
last week, the study reports that excess weight accounts for a mere 25,814 deaths in the United States each year. In 2004, the CDC placed that estimate at 400,000 deaths.” 57 Or as an article published two months later put it:
“The true cost of obesity has been up for debate recently when the Centers for Disease Control and Prevention slashed in April its estimates for deaths per year in the United States due to obesity [and overweight combined] from 365,000 to just 25,814.” 58

The Fat-OK study relied solely on the nationally representative National Health and Nutrition Examination Survey (NHANES), including a newly released wave of data. The 2005 team also correctly adjusted for gender, smoking, and age, which had not been done in the first study. Adjusting correctly for age was especially important since the relative risk of death associated with overweight and obesity (or, for that matter, cigarette smoking) declines with increasing age. Thus, if one does not properly adjust for age, one ends up with inflated estimates of mortality. 59 The Fat-OK study also used the full “normal weight” category as the reference category, rather than only the upper range of the “normal weight” category.

Properly adjusting for age and other factors, using the NHANES data and the full “normal weight” category as the reference group, the Fat-OK study found that being overweight-but-not-obese (BMI greater than 25 but less than 30) was associated with 86,094
fewer
deaths. It further found that having a BMI over 30 was associated with 111,909 excess deaths in 2000 and that these deaths were concentrated in the category of BMI greater than 35, a category into which about 15 percent of the U.S. adult population fell in 1999–2000. The annual deaths associated with overweight and obesity combined were 25,815, a small fraction of the earlier estimate. This was also less than the excess deaths associated with underweight (33,756), even though underweight accounts for roughly 2 percent of the U.S. population in 1999–2000, compared to about two-thirds that are overweight or obese. 60 The researchers repeated their analyses in a sample that included only people who had never smoked and found that the increased risk of underweight persisted.

Despite the fact that the data presented in both studies are purely associational, the actual title of the Eating-to-Death study was “Actual
Causes
of Death in the United States, 2000.” 61 In contrast, the authors of the Fat-OK study were careful to specify that “associations are not necessarily causal” and that “other factors associated with body weight, such as physical activity, body composition, visceral adiposity, physical fitness, or dietary intake, may be responsible for some or all of the apparent associations of weight with mortality.” 62

These two studies suggested dramatically different policy implications. Based on the Eating-to-Death study, overweight and obesity was quickly becoming the leading “cause of preventable death,” supporting the argument that the U.S. federal and state governments should invest more resources in studying and combating the growing “obesity epidemic.”
However, the Fat-OK study suggested that fewer deaths can be attributed to overweight and obesity combined than to microbial agents (75,000), toxic agents (55,000), motor vehicle crashes (43,000), or incidents involving firearms (29,000), throwing doubt on the premise that the war on obesity should be a public health priority. 63

Ultimately, the CDC released a statement that they would use the Fat-OK study’s lower estimates of the number of deaths associated with obesity, rather than the higher Eating-to-Death estimates. 64 Specifically, the CDC stated: “The latest study based on a nationally representative sample of U.S. adults estimates that about 112,000 deaths are associated with obesity each year in the United States.” 65 Note how the CDC fails to mention that this same study had found the overweight category to be associated with
fewer
deaths and even states that “being overweight carries health risks.” 66
In contrast, when overweight had been thought to be associated with increased mortality, the CDC cited the number of excess deaths associated with overweight and obesity
combined
.
In other words, it included the mortality statistics for overweight when these produced a higher estimate and ignored them when they lowered the estimate. This demonstrates how scientific results can and are often manipulated to serve political ends.

These two studies and news coverage of these studies offered the opportunity for which I had been waiting, to compare how the news media report differently on research that challenges—versus research that reinforces—the public health crisis frame. Using a sampling strategy described in the methodological appendix, I identified a paired sample of 35 news reports on the Eating-to-Death study and 61 news articles on the Fat-OK study. 67

CRACKS IN THE PUBLIC CRISIS NARRATIVE

These contrasting scientific findings lead to strikingly different news reports. While most (69 percent) of news reports on the Eating-to-Death study evoke a public health crisis, or “public health emergency,” 68 only 3 percent of news reporting on the Fat-OK study do. 69 A typical news report on the Eating-to-Death study quotes obesity researcher Judith Stern saying, “Unless we view this as an emergency the way we viewed HIV (the virus that causes AIDS), we are not going to get to first base, let alone a home run.” 70 Use of war metaphors are also more common in reports on the Eating-to-Death, compared to Fat-OK, study (40 versus 16 percent): “[Health and Human Services Secretary Tommy Thompson] urged Americans to fight the battle of the bulge by changing diets that depend on fast food and prepared foods, and neglect fruits and vegetables.” 71 Indeed, news reports on the Eating-to-Death study frequently quote Thompson saying things such as: “Americans need to understand that overweight and obesity are literally killing us. We need to tackle America’s weight issues as aggressively as we are addressing smoking and tobacco.” 72

News coverage of the Eating-to-Death study is also significantly more likely, than news reporting on the Fat-OK study, to link overweight and/or obesity to specific diseases, including type 2 diabetes (77 versus 54 percent), heart disease (77 versus 49 percent), stroke (60 versus 8 percent), and cancer (69 versus 26 percent). Listing health risks, as in the following quotation, is exceedingly common in reporting on the Eating-to-Death, but not the Fat-OK, study: “We need to stop this growing trend and provide people with knowledge and tools to help them take control of their weight and reduce their risk for developing diabetes, heart disease, some forms of cancer, and other disabling medical conditions.” 73 In reporting on the Eating-to-Death study, such medical risks are deemed important not only because of their health implications but also for their economic repercussions: “Health officials say [overweight and obesity] increases the risk of heart disease, diabetes, and some forms of cancer, leading to medical costs and lost productivity that the US Surgeon General has estimated at $117 billion in 2000.” 74

Most news reports of
both
studies discuss the health risks of being obese, although the likelihood of this is significantly higher in coverage of the Eating-to-Death study, compared to the Fat-OK study (82 versus 97 percent). Roughly one-third of each news sample also discuss health risks associated with being “overweight,” and these rates are not significantly different. In reporting on the Fat-OK study, the typical line is that, while people in the overweight category had lower rates of mortality, they still had higher rates of certain diseases. Interestingly, news reports on the Fat-OK study are
more
likely than news reports in 2004 to discuss overweight and obesity as increasing the likelihood of having various risk factors, including hypertension (31 versus 0 percent) and high levels of bad cholesterol (23 versus 3 percent). That is, while less likely to argue or even contest that overweight and/or obesity are causing various
illnesses
,
many of these articles note that they might nonetheless be associated with other
risk factors
.
This speaks to how news reports on the Fat-OK study typically emphasize that, while the risks associated with obesity and especially overweight seem to have been greatly exaggerated, there are nonetheless some health risks associated with higher body mass.