Current issues of ACP Journal Club are published in Annals of Internal Medicine


Body-weight change and mortality in men

ACP J Club. 1993 Mar-April;118:56. doi:10.7326/ACPJC-1993-118-2-056

Source Citation

Lee IM, Paffenbarger RS Jr. Change in body weight and longevity. JAMA. 1992 Oct 21;268:2045-9.



To determine if body-weight change is an independent risk factor for all-cause, coronary heart disease and for cancer mortality in middle-aged men.


Cohort analytic study with follow-up from 1977 to 1988.


Harvard University.


11 703 male alumni (mean age 58 y, mean weight, 78.5 kg in 1977) participating in the Harvard Alumni Health Study who were free of self-reported cardiovascular disease and cancer in 1977 and who had completed questionnaires in 1962 or 1966 and in 1977 (76% of survivors who had responded in 1962 or 1966). Fewer than 1% were lost to follow-up.

Assessment of risk factors

Weight change (from 1962 or 1966 to 1977) was assessed using self-reported weights on the 1962 or 1966 and 1977 questionnaires. Height was ascertained from the 1962 or 1966 questionnaire, and cigarette smoking and physical activity were determined from the 1977 questionnaire.

Main outcome measures

Mortality from all causes, from coronary heart disease, and from cancer, determined from death certificates.

Main results

1441 men died during follow-up. After adjusting for differences in age, height, smoking habits, and amount of physical activity, the lowest risk for all-cause mortality was observed among alumni whose weight remained unchanged between 1962 or 1966 and 1977. With this group as the referent (relative risk [RR], 1.00), the RRs of death from all causes associated with losing > 5 kg, losing between 1 and 5 kg, gaining between 1 and 5 kg, and gaining > 5 kg were 1.57 (95% CI 1.34 to 1.84), 1.26 (CI 1.10 to 1.46), 1.06 (CI 0.90 to 1.24), and 1.36 (CI 1.11 to 1.66), respectively. Of the 1441 deaths, 345 (24%) were from coronary heart disease and 459 (32%), from cancer. The RRs for coronary heart disease, for the same weight-change categories were 1.75 (CI 1.26 to 2.43), 1.43 (CI 1.05 to 1.93), 1.28 (CI 0.91 to 1.80), and 2.01 (CI 1.36 to 2.97), respectively. Weight change did not predict mortality from cancer. These findings were not explained by smoking habit, physical activity, body-mass index, time of death (between 1977 and 1982 or between 1983 and 1988), or by including patients with physician-diagnosed hypertension or diabetes mellitus on the 1977 questionnaire. Those losing or gaining more weight also reported greater total lifetime weight loss, which may indicate weight cycling.


Body-weight loss or gain might be an independent risk factor for mortality from all causes and from coronary heart disease but not for mortality from cancer in male Harvard alumni.

Sources of funding: National Heart, Lung, and Blood Institute and the National Cancer Institute.

For article reprint: Dr. I. Lee, Department of Epidemiology, Harvard University School of Public Health, 677 Huntington Avenue, Boston, MA 02115, USA. FAX 617-566-7805.


Repeated weight gain and loss (weight cycling) is a subject of current epidemiologic scrutiny. Some studies have found this to be an independent risk factor for arteriosclerotic cardiovascular disease, whereas other studies have not. Major difficulties exist in investigating this problem and are discussed elsewhere (1).

The limitations in the study by Lee and Paffenbarger include probable biases from a reliance on self-reported measures, lack of control for all major arteriosclerotic cardiovascular disease risk factors, limited socioeconomic class, men only, use of only 2 points in time for measurement (1962 or 1966 and 1977), a largely nonsmoking population (83%), and a 24% nonparticipation rate for questionnaires. The study's strengths are a substantial sample size (n = 11 703) and a 99% follow-up of death certificates of all alumni.

Many patients have great difficulty quitting smoking, adjusting their dietary lipids, and increasing their activity level, all factors of demonstrated importance in the prevention of arteriosclerotic cardiovascular disease. The magnitude of the effect of weight cycling appears to be smaller than that for the major arteriosclerotic cardiovascular disease risk factors. Thus, the clinical use for the information on weight cycling may be limited. This information could be used for select patients to persuade them to maintain weight loss because of a possible added risk for arteriosclerotic cardiovascular disease from weight cycling.

Kurt T. Hegmann, MD, MPH
Medical College of Wisconsin Milwaukee, Wisconsin, USA


1. Bouchard C. Is weight fluctuation a risk factor? [Editorial] N Engl J Med. 1991;324:1887-9.