Junkfood Science: Fat and long life — The “obesity” crisis is crumbling

November 09, 2007

Fat and long life — The “obesity” crisis is crumbling

The latest study from CDC senior research scientists at the National Center for Health Statistics on the associations between various body weights and deaths has been in the news this week. The media has been giving us bits and pieces of the story and various angles, but not all of it accurate or complete. With no clear overview, it’s been virtually impossible for most of us to figure out what the study actually showed or to understand what it means for us. In fact, the study findings we haven’t heard are some of the most important parts.

Here's the cut-to-the-chase version: There is no credible evidence to support the belief that anyone dies 'of fat.'

This paper, led by Dr. Katherine Flegal, Ph.D., is a continuation of these researchers’ first report published in 2005 in the Journal of the American Medical Association. That famous paper showed the government’s claim that 400,000 Americans were dying from “overweight and obesity” — begun by the director of the CDC, Julie Gerberding, and Secretary of Health and Human Services, Tommy Thompson — to be false. Instead, ‘overweight and obesity’ together were associated with 25,815 excess deaths, with 89,094 fewer deaths associated with those in the ‘overweight’ category. But even ‘obesity’ wasn’t nearly the risks we’ve been led to believe. Nonsmokers with ‘class I obesity’ (BMI 30 to <35), for example, had 23% lower risks than ‘normal’ weight people. Since age is the biggest risk factor for death, looking at the elderly, even the most ‘morbidly obese’ with BMIs >35 were associated with a mere untenable 12-17% increased risk — still nowhere near those of thin people with 50-69% higher risks. And yes, the researchers had accounted for smoking, chronic diseases and preexisting health problems and cancer, involuntary weight loss and long-term obesity... and the results were the same.

Rather than 65% of Americans being accused of being “too fat” and risking early graves, it turns out that most fat people may actually outlive those of ‘normal’ weight. The “obesity crisis” isn’t nearly as dire as we were hearing.

This new study published in JAMA built upon that paper.


The researchers compiled mortality data through 2000 from the National Health and Nutrition Examination Surveys (NHANES) I (1971-1975), II (1976-1980), and III (1988-1994). The NHANES surveys interview and perform physical examinations on a nationally-representative cross section of the population. The researchers then used computer modeling to find associations between adult deaths reported from 2004 U.S. vital statistics, various “causes of death” recorded on ICD-9 and ICD-10 codes, and weight categories and relative risks from the NHANES surveys. They then calculated estimated excess deaths. Two critical points need to be emphasized before proceeding.

1. It bears repeating over and over again that these statistical findings are associations, not that weight — thin or fat — is the cause of any links with deaths.

2. Lumping these three NHANES reports together and using their relative risks for weight-associated mortality comes with an important caveat. With each NHANES survey, the associations between obesity and mortality, with similar lengths of follow-up, dramatically diminished. NHANES I was the survey which showed the strongest association between obesity and mortality, they noted, the likes of which haven’t been seen in later surveys. NHANES III has the fewest number of deaths. Their approach, however, “gives less emphasis to the most recent data,” they wrote.

In other words, this report will over-estimate any associated risks to obesity that would be seen using the most current data. [The risks from the latest NHANES 1999-2002 survey of adults were not included in this analysis, due to short follow-up time.]


Here are the results, from their balanced follow-up figures, which use the same lengths of follow-up for each report.

BMIs <18.5 (Hollywood thin), compared with ‘normal’ weight, were associated with notably higher deaths from:

other cardiovascular diseases

chronic respiratory diseases

acute respiratory diseases and infections

all non-cancer/non-CVD causes.

It was also associated with insignificantly higher deaths from:


coronary heart disease

lung cancer

all cancers (including those considered ‘obesity-related’)

Most of the excess deaths associated with underweight, no matter what that age, were for non-cancer/non-CVD causes — which accounted for about 39% of all adult deaths in 2004. This fact negates suggestions that the underweight deaths were because people were sick from cancer or heart disease. The researchers ruled out those who were thin because they may have been sick, and analyzed the data to ensure this, even excluding those who had ever smoked, had a history of CVD, died in the first 3 years of follow-up or were over age 70 and the estimates held. “Thus, there was no indication that our results were biased by illness-induced weight loss prior to baseline (“reverse causation”) or residual confounding due to smoking,” they wrote.

· BMIs 25 to <30 (“overweight”), compared with ‘normal’ weights, were associated with lower deaths from:

coronary heart disease

lung cancer

chronic respiratory diseases*

acute respiratory diseases and infections


all non-cancer/non-CVD causes*

And slightly higher deaths from diabetes/kidney disease.

It was also associated with insignificantly lower deaths from:

cancers considered ‘obesity-related’

other cardiovascular disease

No way around it, being overweight was associated with a total of 138,281 fewer deaths from all causes. That’s a generous 72 pound weight range for a 5’-4” woman that’s linked with lower or the same risks as someone of a government-recommended ‘normal’ weight. But that’s not all...

· BMIs >30 (‘obese”), compared with ‘normal’ weights, were associated with lower deaths from:

lung cancer

chronic respiratory diseases


all non-cancer/non-CVD causes

acute respiratory diseases and infections

Obesity was associated with higher deaths from coronary heart disease and cardiovascular diseases — but CVD-related risks only reached a barely tenable relative risk (2.0) among younger cohorts with the highest BMIs >35. While cardiovascular disease accounted for 37% of all adult deaths in the United States in 2004, said the researchers, the balanced follow-up data showed that only an estimated 9% of those heart-related deaths were associated with obesity. But when they used the lower relative risks seen in the more recent NHANES III, obesity’s relationship to heart disease deaths dropped to 5%.

Compared to ‘normal’ weight, for all non-cancer/non-CVD deaths, there were 560 fewer deaths associated with obesity (the entire BMI range). Yes, obesity was beneficial for all of these causes of death. To flesh out the details, the researchers divided all of these causes into 5 subgroups, with diabetes and kidney disease grouped together because they said there were too few diabetes deaths in the younger ages to make reliable estimates. They found that for all of the other 4 subgroups, obesity was associated with reduced mortality. So, while obesity was associated with higher risks for diabetes/kidney disease-related mortality, it was associated with lower deaths from all of the non-cancer/non-CVD causes.

No difference was seen for cancers: While associations with cancers considered ‘obesity-related’ were slightly higher, they were lower for others, leaving no higher deaths from cancers overall.

The authors spent considerable time discussing cancer-related mortality because it accounted or 24% of total deaths among U.S. adults in 2004, they wrote. “Our results showed little or no association of excess all-cancer mortality with any of the BMI categories. None of the estimates of excess deaths was statistically significantly different [from null].”

Even the graphics are another of those splitting hair things, all hugging a null relative risk.

Preserving the ‘obesity crisis’

We’ll look at the media coverage more in a moment, but generally the reporting of this study has told us that it found that being a few pounds overweight or being “pleasantly plump” is beneficial, but cross that line into ‘obesity’ and you’re a goner.

While that may help to preserve perceptions of an ‘obesity crisis,” a BMI of 30 is not a magic threshold of risk, either.

It’s easy to see why reporters might have come away with that impression, though. Unlike their first paper, for the death estimates in this paper, Dr. Flegal and colleagues lumped together all BMIs >30, leaving the erroneous impression with most readers that all levels of obesity carry the same risks and that most people in the ‘obese’ range are at higher risks of dying. In fact, most aren’t.

While the CDC authors didn’t breakdown their estimated excess death rates in the ‘obese’ range for this paper, we can turn to their first paper, which stated that, although the prevalence of BMI 35 and greater is low, most of their estimated excess deaths associated with obesity were at that uppermost category. Even at the highest BMIs, though, the relative risks were nominal, 1.17-1.83. Meaning, weight itself is unlikely to be a causative factor in mortality. The bulk of the evidence to date has demonstrated that overall mortality risks associated with obesity don’t even reach tangible levels until the topmost BMIs — a small percentage of the population — and even the most “morbidly obese” never reach the same risks as those who are underweight.

BMI markers

Just like other health risk factors, BMI is not equivalent to actual risks and it is not the same thing as a disease or the cause for a disease. The fact that the very uppermost and lowermost BMIs are most associated with higher rates of premature deaths does not mean their weights are to blame.

Instead, it may say something far more important about our culture and healthcare priorities. Blame has no role in BMI. No amount of incentivizing or self control will transform someone’s inherent body type and size into a completely different one. If health is really our primary concern, public health would be addressing what these BMIs might be markers for that likely do impact death rates where we can be of help — such as socioeconomic status and access to quality healthcare and ADL assistance.

· For example, underweight people may be intentionally (controlling their food intakes due to food or weight fears) or unintentionally (hunger, malabsorption, dental health, poor access to food or cooking assistance) suffering from undernutrition.

· In contrast, the highest BMIs >40 (4.7% of the population) have been well-documented to be among the lowest social economic classes, to be suffering from highest rates of discrimination and stress, to have the highest rates of dieting and be prescribed risky weight loss medications; be underinsured and to be receiving poorer healthcare. Women at BMIs >40, for example, are less likely to receivepreventive gynecological care, pap tests and breast cancer screenings.

Population numbers versus individual risks

When we read that this study reported that an estimated total of 95,442 deaths were associated with those in the obese category and 46,198 among the underweight, at first glance it might appear that obesity is associated with the greatest risks. But to make sense of those numbers, we have to put them in the context of how many people are fall into each of those categories: only 2.2% of the population is underweight compared to 23.3% obese. In other words, overall, an obese person has one-fifth the risks for premature death compared to someone who is Hollywood thin.

Don’t overlook the evidence

This study is not the first that researchers have looked at linking causes of deaths with weights and found ‘overweight’ and ‘obesity’ not nearly the risks popularly believed. In fact, it confirms decades of mortality data, much of which has been reviewed here.

Nor is this the first time that objections to this evidence been raised, mostly by those with interests in obesity. It pays to take a critical eye to the news.

Probably the most widely distributed story was the Reuters’ article titled, “Put down that fork: being fat is still unhealthy.” Quoted was Dr. Robert Kushner, who was described as “a professor of medicine at Northwestern University who specializes in nutrition and diet.” What wasn’t revealed to readers was that he is the Director of the Wellness Institute at Northwestern Memorial Hospital and authored the Personality Type Diet and co-authored Fitness Unleashed, A Dog and Owner's Guide to Losing Weight and Gaining Health Together. He is also Clinical Committee Chair for the Obesity Society (formerly NAASO) and authored the AMA’s Obesity Treatment Guide for Physicians, funded by Robert Wood Johnson Foundation. Also quoted in the Reuters’ article was Dr. Louis Aronne, who was described as “an obesity expert at New York-Presbyterian Hospital/Weill Cornell Medical Center.” JFS readers will remember him as co-chair of the Reality Council/Reality Coalition, spokesperson for Accomplia, and President of the Obesity Society (formerly NAASO). He has also authored Weigh Less, Live Longer and was the clinical investigator for the diet drugs sibutramine and orlistat.

Like the media flurry that surfaced after the first Flegal paper in 2005, those with the American Heart Association, American Cancer Society and Harvard have been some of the most outspoken critics. As the Independent reported:

“It's just rubbish," fumed Walter Willett, the professor of epidemiology and nutrition at the Harvard School of Public Health. “It's just ludicrous to say there is no increased risk of mortality from being overweight."... “This is a very puzzling disconnect," said Dr JoAnn Manson, the chief of preventive medicine at Harvard's Brigham and Women's Hospital.

No reporter has pulled up the studies co-authored by Manson and Willett themselves, where their own data showed the lowest mortality risks associated with those who are ‘overweight,’ and that even at the highest BMIs, the risks didn’t reach tenable odds (beyond chance or statistical error). For example, Dr. Manson led a study on premenopausal women, published in the New England Journal of Medicine in 1995, which found that age-adjusted relative risks were lowest among the ‘overweight’ and didn’t even begin to creep upward (from RR=1.3) until BMIs surpassed 31.9. Dr. Willett also co-authored a study on men published in the American Journal of Epidemiology in 2000, which concluded that among men 65 years and older, when most deaths occurred, “there were no significant relations between BMI and overall, cardiovascular disease, or cancer mortality risk.”

The New York Times quoted Dr. Manson on Wednesday saying that “other studies, including ones at Harvard, found that being obese or overweight increased a person’s risk for any of a number of diseases, including diabetes, heart disease and several forms of cancer. And, she added, excess weight makes it more difficult to move about and impairs the quality of life.”

An article in the same issue of JAMA led by Dawn E. Alley, Ph.D., of the Robert Wood Johnson Foundation Health and Society Scholars Program, is being exampled by some as evidence of how obesity is disabling and limits the ability to function. A closer look, however, finds the risks associated with disability among the elderly in that paper didn’t become tenable until the very highest extremes of BMI (class III+), which represent a very small fraction of the population who need our help and compassion. These are not sincere depictions of the vast majority of the ‘obese.’

The spunkiest responses to these claims were found at KateHarding.net, where ‘obese’ commenters wrote:

Manson is talking about excess weight like Shauna’s excess weight and Chiara’s excess weight, not just, say, my excess weight, which clearly renders me practically immobile. If only Shauna, Chiara, and I could spend our time weighing and measuring lettuce leaves instead of lugging our enormous bulks into our early graves. Our quality of life would be so much improved!

Just look at the ‘obese’ men and women depicted on this JFS article, as well as those here, here, here, and here. Do these honestly look like people who are having any difficulty moving about? Of course not. It shows how ready some are to believe the media portrayals and that those extreme examples on television are the true picture of obesity in America.

They are not.

© 2007 Sandy Szwarc. All rights reserved.

Bookmark and Share