TCS Daily

Checking the Obesity Math

By Duane D. Freese - December 8, 2005 12:00 AM

Sometimes when you arrive late to a party, you don't have a clue what's happening.

That was the case for Connecticut Congresswoman Rosa DeLauro who recently turned up 15 minutes late for a National Press Club Newsmaker session, "Food Fight: Childhood Obesity, Crisis or Hype?"

The three other speakers -- Michael Jacobson of the Center for Science in the Public Interest (SSPI), a liberal NGO; Dan Mindus of the free market Center for Consumer Freedom, a free market NGO; and a spokesman for the National Association of Grocery Manufacturers -- had already made their presentations.

Mindus had just announced his group would soon be releasing a book entitled An Epidemic of Obesity Myths, in which the authors take aim at a number of long-asserted claims, including one that says obesity costs the nation $117 billion in healthcare costs. Rep. DeLauro -- author of a bill to require chain restaurants with more than 50 outlets to nutritional information on menus and counter boards - used that figure in her opening remarks:

"And so in my view, with these problems getting worse, not better -- with diseases caused primarily by obesity having become the single largest drain on our nation's health care system at $117 billion -- it is becoming increasingly clear that Congress is abdicating a responsibility of its own."

And Rep. DeLauro isn't the only one to catch the $117 billion dollar bug. In a release made available at the Food Fight event, CSPI called for nutrition labeling at chain restaurants, claiming: "Obesity costs American families, businesses and governments about $117 billion a year in healthcare and related costs."

The Number's Trail

Where did this number come from?

The Government Accountability Office (GAO) of Congress has quoted Eileen Salinsky and Wakina Scott's "Obesity in America: A Growing Threat." Those researchers noted that $61 billion of the cost was due to medical expenses and $56 billion was due to indirect costs, "such as the loss of future earnings due to premature death."

And how did they arrive at those figures? They cite the "Surgeon General's Call to Action to Prevent and Decrease Overweight and Obesity 2001."

And where did then Surgeon General David Satcher get his information? Well, the footnotes say from a "personal communication from A. Wolf" of November 26, 2001. This communication updated the figures from Anne Wolf's and Graham Colditz's study published in a March 1998 edition of Obesity Research, "Current Estimates of the Economic Costs of Obesity in the United States," a study that Wolf amplified with another article in Obesity Research, "What is the Economic Case for Treating Obesity?". That article, also cited by the Surgeon General, emphasized that "most of the cost of obesity is due to Type 2 diabetes, coronary heart disease and hypertension."

Now that we know where the number came from, we can ask how reliable it is. As it turns out, the $117 billion figure just isn't very good at all.

Studies providing estimated costs associated with diseases are based on comparisons between different groups. In the case of obesity cost studies, the groups are usually divided by the Body Mass Index or "BMI" into the classifications set by the National Institutes of Health (NIH):

  • underweight (a BMI under 18.5),
  • normal weight (a BMI 18.5 to under 25),
  • overweight (a BMI 25 to under 30)
  • and obese (30 and above).

The studies are generally of two types - attributable risk/cost studies and cohort studies.

Attributable risk/cost studies -- such as the one from Colditz and Wolf -- derive their medical costs from national estimates for the aggregate cost of a variety of diseases -- diabetes, hypertension, coronary heart disease (CHD), stroke, congestive heart failure, pulmonary embolism, obstructive sleep apnea, hypercholesterolemia, colon cancer, endometrial cancer and gastroesophageal reflux disease, among others. Researchers then apply a formula based on assessed risks in the various categories to set costs.

But there's a problem with this. University of Pennsylvania economists and health researchers Henry Glick and Adam Tsai noted in an October presentation "Health Care Costs Associated with Elevated Body Mass Index and the Cost Effectiveness of Weight Loss" that in such studies "the costs of diabetes can include CHD, hypertension and hypercholesterolemia; the costs of hypertension can include CHD and stroke, and the cost of hypercholesterolemia can include CHD. Thus the cost of CHD might be counted four times in these studies."

Indeed, Mindus pointed out that Colditz and Wolf had themselves admitted their study included double counting of medical costs. If you double count the costs, your figures simply can't be accurate.

So what about cohort studies? These studies use data from studies of population samples, the best being nationally representative samples such as the Medical Expenditure Panel Survey, National Health Interview Survey, Healthcare for Communities telephone survey and the National Health and Nutrition Examination Survey. Researchers put people in the appropriate categories and then estimate the health costs of participants and attempt to predict the cost difference.

The most widely quoted of these studies, particularly by the CDC, is one by Eric A. Finkelstein, Ian C. Fiebelkorn, and Guijing Wang, "National Medical Spending Attributable
To Overweight And Obesity: How Much, And Who's Paying?"
found in the May 2003 edition of Health Affairs. It placed the medical cost of overweight and obesity at representative samples spreading costs of either cohorts at $72.5 billion in 1998, adjusted for inflation by CSPI in another pamphlet at the Food Fight event to $95 billion.

But again, there are problems with the numbers. Like the Wolf-Colditz study, Finkelstein et al are at the high end of the estimated cost range, with their estimated per person excess annual cost for obesity at $762 being almost three times higher than the $262 low cost estimate. Their excess cost for being overweight at $247 was $351 higher than the -$104 (yes, negative $104) estimate of a lower cost cohort study.

Furthermore, as Glick and Tsai noted, cost data in cohort studies are "highly skewed" by "observations with very high costs." Those include morbidly obese individuals with multiple maladies, some of which, including their obesity, are likely to be genetic in origin. In short, they are individuals for whom nutritional information at restaurants will do absolutely nothing.

The obesity health cost studies problems don't end there, though.

For example, Wolf's and Colditz's indirect costs for premature death of obese people rely on another funny number.

A study in the Journal of the American Medical Association in 1999 and referenced in the Surgeon General's Call to Action put obesity related premature deaths at 300,000. In a 2004 JAMA article, by CDC Director Julie Geberding and others, the figure was bumped up to 400,000.

But at the end of 2004, the CDC had to admit the 400,000 figure was wrong. Citing statistical errors, it put the number at 365,000.

But even that was way too high. In the Spring of 2005, a new study by the CDC's Katherine M. Flegal, Barry I. Graubard, David F. Williamson and Mitchell H. Gail using new data and more refined statistical techniques put premature annual deaths due to obesity at just under 112,000. In addition, the study found that people classed as overweight suffered 86,000 fewer premature deaths than people of so-called "normal" weight.

If you have two-thirds fewer deaths, obesity's burden for premature death drops by almost $15 billion.

And what does the lower death rate for overweight than normal weight suggest? It also suggests that the health costs associated with being overweight may well fall to that negative $104 per person, leading to a modest $25 billion reduction in the health costs for this manufactured epidemic.

Cause and Effect

Still, the overestimates aren't the biggest problem with the studies. There big problem is mixing up cause and effect. As Glick and Tsai noted, "BMI might not be the cause of the disease/cost, but might be another expression of a common cause of both BMI and disease/cost."

This is something highlighted by Paul Campos in his book The Obesity Myth. You would think, in light of the epidemic increase of more than 25 percent in obesity in the 1990s, death rates of associated diseases would rise, too. But they haven't. Rates for diabetes increased only 5 percent, from an 8.2 percent of adults to 8.6 percent. As for cardiovascular disease, death rates have declined by 60 percent since the 1950s while obesity has tripled. And death rates from most cancers associated with obesity are likewise declining.

Medical experts will say that these health gains were a result of a decline in smoking and improved drugs and cancer treatments.

But this merely emphasizes the questions that Glick and Tsai raise about the health and other costs researchers say obesity is imposing on society: "Do we know the independent effects of weight, physical activity and fitness?" Their answer: No. "Could our aesthetic judgments about overweight/obesity be affecting our scientific judgments?" Possibly.

Their conclusion is clear, "evidence is lacking" to support statements such as "overweight/obesity is a major public health problem that costs society billions of dollars, and we should be doing everything we can to combat it."


TCS Daily Archives