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


Increased body mass index predicted mortality in women

ACP J Club. 1996 Mar-April;124:50. doi:10.7326/ACPJC-1996-124-2-050

Source Citation

Manson JE, Willett WC, Stampfer MJ, et al. Body weight and mortality among women. N Engl J Med. 1995 Sept 14;333:677-85.



To determine the association in women between body mass index (BMI) and mortality.


16-year cohort analytic study of women in the Nurses' Health Study who responded to a mailed questionnaire.


Community-based study in the United States.


115 195 registered nurses aged 30 to 55 years who were free of cardiovascular disease (CVD) and cancer in 1976 and provided data on height and weight.

Assessment of Risk Factors

Data collected in 1976 included age, current height and weight, current and past cigarette smoking, other risk factors, and medical history. Follow-up questionnaires mailed every 2 years updated this information and recorded the occurrence of major illness. The 1980 questionnaire included a question on the participants' weight at age 18 years and requested information about food frequency and physical activity. Participants were grouped into 7 categories of BMI; the participants at the lowest range (< 19.0) were the reference group.

Main Outcome Measures

Relative risk (RR) for all-cause mortality and death from CVD and cancer.

Main Results

During 16 years of follow-up, 4726 women died (881 of CVD, 2586 of cancer, and 1259 of other causes). The age-adjusted RR for all deaths in all women was 1.0 in women with a BMI < 19.0, < 1.0 in women with a BMI ranging from 19.0 to 28.9, and 1.3 in women with a BMI ≥ 32.0 (P for trend = 0.001). Multivariate adjustment for smoking and other risk factors strengthened the association (P < 0.001) between obesity and mortality. When the analysis was restricted to women who had never smoked (1499 deaths), no J-shaped relation was found; rather, the RR increased consistently from 1.0 for a BMI < 22.0 to an RR of 1.9 (95% CI 1.5 to 2.5) for a BMI ≥ 32.0 (P for trend < 0.001). 53% of the deaths among women with a BMI ≥ 29.0 were attributable to their obesity. Compared with the reference group, women with no smoking history and a BMI ≥ 32.0 had an RR for death from CVD of 4.1 (CI 2.1 to 7.7) and an RR for death from cancer of 2.1 (CI 1.4 to 3.2). Both weight gain ≥ 10 kg after age 18 years and a BMI of ≥ 22.0 at age 18 years were predictors of overall mortality and death from CVD in middle adulthood.


After controlling for the confounding effects of smoking and disease, a direct association was seen between body mass index and both all-cause mortality and death from specific causes. The lowest mortality was found in the leanest women who had never smoked and whose weight had remained stable since age 18 years.

Source of funding: National Institutes of Health.

For article reprint: Dr. J.E. Manson, 180 Longwood Avenue, Boston, MA 02115, USA. FAX 617-432-0335.


Weight loss and weight fluctuation were associated with death in men with pre-existing disease and men who smoked

The prevalence of obesity is increasing in industrialized Western societies. According to 1988 to 1991 national health statistics (1), a third of all adults in the United States were at least 20% above their ideal weight. This is a striking increase from the stable, 25% prevalence of overweight documented between 1960 and 1981. Although the health hazards of excess weight have been documented (2), the relation between obesity and morbidity is not as well understood as many other risk factor-morbidity relations. These reports from the Honolulu Heart Program and Nurses' Health Study provide no startling new information, but they do advance our understanding of obesity and its risks.

One source of confusion in the body mass-mortality relation is the J-shaped curve issue; that is, the observation in some studies that increased mortality is found at both the low and high ends of the body weight spectrum. The reports from both the Nurses' Health Study and the Honolulu Heart Program Study showed the importance of smoking and preexisting illness as confounders in the body mass-mortality curve. When smoking and preexisting diseases were considered, minimal or no evidence of increased mortality in underweight persons was observed.

With regard to the strength of the body mass-mortality relation, the population samples from both studies showed a modest but significant increased risk for overall mortality when the most overweight subgroups were compared with the lowest-risk subgroups. In 1990, the Nurses' Health Study published its 8-year follow-up data on overweight as a major health risk for coronary artery disease in women (3). The current 16-year mortality data are even more striking because nonsmoking women with a BMI ≥ 29.0 (> 30% above ideal weight) showed a 5-fold increase in mortality from coronary heart disease compared with women with a BMI < 22.0. Although this increase was partly related to the high prevalence of hypertension, dyslipidemia, and diabetes mellitus among overweight persons, significant excess risk persisted even after controlling for these variables. The cause of this "specific" obesity-related cardiovascular risk is not clearly understood.

The current Nurses' Health Study report also analyzed cardiac mortality on the basis of self-measured waist-to-hip circumferences. The importance of body fat distribution as a predictor for coronary death was shown by the RR of almost 9.0 for nonsmoking women in the highest quintile for waist-to-hip ratio. This association was somewhat stronger than that between BMI and cardiovascular deaths.

Byers (4) previously noted the difficulty of interpreting the mortality findings in these studies as they relate to weight change and weight fluctuation. Data on changes in weight were collected and analyzed differently, and the sex, ethnic background, and weight distributions of the participants were very different in the 2 studies. Perhaps the most consistent finding came from the large population of the Nurses' Health Study, which showed an increase in all causes of mortality with a >10-kg weight gain after age 18 years.

The limitations of these well-designed, long-term cohort studies were well described by the authors, and the conclusions were reasonable extensions of the data. The findings should allay concern about any adverse effects of low body weight in otherwise healthy persons and should reinforce concern about increased mortality, particularly cardiovascular, in overweight persons.

Paul D. Levinson, MD
Brown University School of MedicineProvidence, Rhode Island, USA


1. Kuczmarski RJ, Flegal KM, Campbell SM, Johnson CL. Increasing prevalence of overweight among United States adults: the National Health and Nutrition Examination Surveys, 1960-1991. JAMA. 1994;272: 205-11.

2. Pi-Sunyer FX. Medical hazards of obesity. Ann Intern Med. 1993;119:655-60.

3. Manson JE, Colditz GA, Stampfer MJ, et al. A prospective study of obesity and risk of coronary heart disease in women. N Engl J Med. 1990;322:882-9.

4. Byers T. Body weight and mortality. N Engl J Med. 1995;333:723-4.