Panel Suggests Methodology Flawed of Recent CDC Obesity Study

May 27, 2005

David C. Holzman

May 27, 2005 (Boston) — The recent federal study by the Centers for Disease Control and Prevention (CDC) suggesting that extra pounds could be risk free raised concerns by panelists at a meeting yesterday at the Harvard School of Public Health (HSPH) in Boston, Massachusetts. The study published in the April 20 issue of JAMA had inspired the popular press to celebrate the prospect of gastronomy without guilt.

The meeting panelists criticized the study's methodology and suggested that the conclusions were biologically implausible and contrary to numerous other analyses.

The study, which was led by Katherine M. Flegal, PhD, of the National Center for Health Statistics, CDC, had found an excess of nearly 112,000 deaths annually among the obese (body mass index [BMI], 30 kg/m 2). While significant, that was a huge decrease from a CDC study last year, which found an excess of 365,000 deaths in this group, suggesting that obesity was second only to smoking on the list of the top U.S. killers. The Flegal study also found no excess mortality among the overweight (BMI, 25 to <29 kg/m 2) as compared with people of normal weight (BMI, 18.5 to <25 kg/m 2; -86,094 deaths; 95% confidence interval [CI], -161,223 to -10,966). However, underweight (BMI, <18.5 kg/m 2) was associated with 33,746 excess deaths (95% CI, 15,726 - 51,766). The study used data from the National Health and Nutrition Examination Survey (NHANES) cohorts I, II, and III. The study authors concluded that improvements in public health and healthcare might have reduced obesity's impact on mortality.

JoAnn Manson, MD, MPH, DrPH, a panelist who is professor of medicine, Harvard Medical School Chief, Division of Preventive Medicine, Department of Medicine, Brigham and Women's Hospital told attendees at the meeting that the study's conclusions lacked plausibility given that obesity causes all the elements of metabolic syndrome. Further, she noted that losing weight intentionally mitigated these elements.

Dr. Manson charged that inasmuch as obesity leads to hypertension, insulin resistance, hyperglycemia, and diabetes, abnormal lipids, and inflammation and that modest weight loss was shown to reduce blood pressure, blood glucose and insulin levels, and diabetes and to increase high-density lipoprotein cholesterol levels, the results of the Flegal study appeared to be biologically implausible. Moreover, she cited a 2004 JAMA analysis encompassing 136 studies and 22,000 patients that showed that bariatric surgery mitigated or eliminated hypertension, diabetes, hyperlipidemia, and obstructive sleep apnea in a majority of patients.

Michael Thun, MD, MS, vice president of epidemiology and surveillance research at the American Cancer Society told attendees that the Cancer Prevention Study II had shown that both breast cancer mortality in women and colon cancer mortality in men climbed steadily with BMI, the latter rising from a relative risk of less than 1 at BMIs of less than 22 kg/m 2 to nearly 2 at BMIs of more than 32.5 kg/m 2. That study included 1.2 million people in 50 states.

But, the crux of the problem with the CDC study, according to several panelists, was failure to control for "reverse causality," a theme that was echoed by several panelists. Implicit in the study's methodology was the notion that leanness can lead to death, when in fact, leanness may merely be an adverse effect of the sickness that leads to death.

For example, chronic diseases such as cancer, cirrhosis, and chronic obstructive pulmonary disease can lead to wasting. Even conditions that may result from overweight, such as heart disease, can ultimately lead to weight loss. The overweight may precede the disease-induced weight loss by years or even decades, Meir Stampfer, MD, DrPH, comoderator of the meeting, and chair, the department of epidemiology, HSPH, told Medscape. And it is possible for the elderly — people in their 70s and 80s — to lose weight, often due to chronic conditions.

Despite the uncertainties about using BMI in the elderly, the CDC study erred in using it as a predictor of death, Dr. Manson told Medscape. That, Dr. Manson said, was one of the two biggest difficulties with the study.

The other was the short follow-up. NHANES I had a 19-year follow-up, but NHANES II and III had only 14 and 9 years, respectively. Extrapolating the death rate in NHANES I, the cohort with the longest (19 years) mean follow-up, to the entire U.S. population resulted in nearly 300,000 deaths annually attributable to obesity, Dr. Manson told attendees at the meeting, while extrapolating cohorts II and III, with follow-ups of 14 and 9 years, respectively, resulted in 29,917 and 43,650 deaths from obesity, respectively. The Flegal study was based on data from all three NHANES cohorts.

Panelist Graham Colditz, MD, DrPH, of HSPH told attendees that "The public health response has ignored the morbidity burden that is real. Dr. Colditz cited data showing that type 2 diabetes, cholelithiasis, hypertension, and coronary heart disease all rise steadily with BMI, and said that direct costs totaled $70 billion annually, 7% of the U.S. healthcare budget.

Donna Stroup, PhD, director of the coordinating center for health promotion at CDC, represented her agency on the panel. The authors of the Flegal study all declined invitations. Dr. Stroup told attendees the CDC study reflected a "remarkable" decline in risk factors, such as high blood pressure among the obese, for the last four years. But asked by Medscape about the panelists' criticisms of the study, Dr. Stroup told attendees that "I would not disagree with my colleagues on the [HSPH] program." Dr. Stroup added that she did not disagree with their numbers.

The 300,000 deaths extrapolated from NHANES I corresponded with an analysis of mortality from obesity that Dr. Manson and others published in JAMA in 1999, based on data from six prospective cohort studies including Nurses' Health Study, NHANES I, and the Framingham Heart study, Dr. Manson told attendees at the meeting. When restricting the data to never/nonsmokers, that analysis attributed 325,000 deaths/year to obesity.

Harvard Forum: "Weighing the Evidence: A Forum to Examine the Latest News About Overweight, Obesity and Mortality in America." Presented May 26, 2005.

Reviewed by Gary D. Vogin, MD

David C. Holzman is a freelance writer for Medscape.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.