the Natural Condition

Being overweight is bad for your health, right? With all the talk of the American obesity epidemic and the consequent rise in diabetes and associated ailments, combined with the fact that cardiovascular disease is known to be the leading killer in America and is exacerbated by lack of exercise, most people assume that excess poundage will lead to a shortened life span. While the information I am about to divulge is not new, I would like to summarize the findings of a 2005 study that, very shockingly, discovered a decrease in the death rate for “overweight” persons.

In 2005, JAMA published an article by Flagel et al. of the National Center for Health Statistics (a division of the CDC) and the National Cancer Institute entitled “Excess Deaths Associated With Underweight, Overweight, and Obesity.” The authors sought to answer broad questions about the impact of weight (BMI) on mortality. They used 3 of the National Health and Nutrition Examination Surveys (NHANES), which comprise data on over 35,000 people dating as far back as 1971. They grouped patients into 4 categories of interest, with one control, all based on BMI. They used the national standards for BMI classification:

1. <18.5 : Underweight
2. 18.5 - <25 : Normal (control)
3. 25 - <30 : Overweight
4. 30 - <35 : Obese
5. ≥35 : Morbidly Obese

When performing their analyses, each group was controlled for known confounding factors to mortality, such as smoking. The table below outlines their findings for relative risk of death by age group and BMI. A RR<1 means that a population is at less of a risk for dying than the Normal-weight group, RR>1 means they do worse.

^{Relative risk of death by age group and BMI. Image courtesy of JAMA.}

What we see in this table is that both the Obese and Underweight cohorts have substantially higher relative risks than Normal-weight individuals, and more markedly so in younger age groups. The big, and unanticipated, anomaly is the Overweight cohort, which exhibited a relative risk <1 in every case. Flagel et al. then used these relative risks, in conjunction with US prevalence data on BMI, to yield estimations of excess deaths due to weight … or, for the Overweight category, lives saved.

^{Annual excess death estimates by cohort, paneled by BMI category. Image courtesy of JAMA.}

If that last line about ‘lives saved’ for being overweight chaffed you: it chaffed me, too, so let’s dig deeper. While I am not a statistician, I did read their Methods thoroughly, but still do not quite understand how they reached their estimates for excess death, which incorporate the relative risks. One seemingly important point about the relative risks for the Overweight group (above table) is that, while each estimate fell below 1, the 95% confidence intervals for every Overweight RR overlapped 1. Because a relative risk of 1 never fell outside of their confidence band for an alpha of .05 for any age group, how could they reach the conclusion that there is always a negative number of excess deaths (positive number of lives saved), “–86 094 deaths; 95% CI, –161 223 to –10 966″¹? They say that the uncertainty of the relative risk is propagated in the excess death estimate, but as I just outlined, the result does not seem coherent.

Their findings are, nonetheless, suggestive that having an ‘Overweight’ BMI may confer additional years of life, although “bias may also result from failure to control for unknown confounders that are associated with body weight and mortality,”¹ as the authors indicate. I should mention that nowhere in the paper do the authors esteem their finding of a decreased relative risk for ‘overweight’ BMIs. In a way, they elegantly dismiss that peculiar result, stating simply that “Overweight was not associated with excess mortality.”¹

In any case, the more important fact to glean from this discussion is that any large deviation from a normal BMI is associated with increased mortality. In response to the Flagel study, Thomas Hoerher, PhD wrote, “although mortality is an important measure of the burden of a disease, it is not the only one. Obesity also has significant impacts on morbidity, health care costs, and quality of life.”² Obesity is not healthy, and 112,000 excess death per year is a serious matter.

Footnotes/Further Reading:
(1) Excess Deaths Associated With Underweight, Overweight, and Obesity. JAMA. 2005.
(2) Controversies in Obesity Mortality: A Tale of Two Studies. RTI-UNC Center of Excellence in Health Promotion Economics. 2006.
(3) CDC Downscales Mortality Risk From Obesity, USA. MedicalNewsToday.com. 2005.