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


Weight change and mortality in high-risk, middle-aged men

ACP J Club. 1994 May-June;120:82. doi:10.7326/ACPJC-1994-120-3-082

Source Citation

Blair SN, Shaten J, Brownell K, Collins G, Lissner L. Body weight change, all-cause mortality, and cause-specific mortality in the Multiple Risk Factor Intervention Trial. Ann Intern Med. 1993 Oct 1;119(7 pt 2):749-57.



To study the association between weight variability and mortality in middle-aged men at high risk for coronary heart disease (CHD).


Cohort study of men from the Multiple Risk Factor Intervention Trial (MRFIT) followed for a mean of 3.8 years.


22 clinical centers in the United States.


Men from the MRFIT (upper 10% to 15% of CHD risk, weight ≤ 1.5 times ideal, and without clinical CHD at baseline) were enrolled if they survived for 7 years after randomization. Exclusions were suspected error in or missing weight data or presence or suspicion of cancer during the MRFIT. 10 529 men (aged 35 to 57 y at entry into MRFIT) were included; 5350 men from the "special intervention" group and 5179 who received "usual care."

Assessment of risk factors

Weight was recorded at each clinic visit. Intrapersonal standard deviation (ISD) of weight change was calculated (continuous measure of general weight variability) and was categorized into quartiles. Type of weight change was also calculated for 5 categories: no change (< 5% change from baseline), steady loss (≥ 5% loss), steady gain (≥ 5% gain), and cycling with last change (either a loss or gain).

Main outcome measures

Mortality was ascertained from national databases and the causes coded from death certificates.

Main results

During follow-up, 380 deaths (228 deaths from cardiovascular disease [CVD]) occurred. All-cause mortality rates per 1000 person-years for the ISD quartiles were 8.28, 8.25, 10.57, and 11.07. After adjusting for confounders, the relative risk (RR) for mortality comparing the highest with lowest quartile was 1.64 (95% CI 1.21 to 2.23). CVD mortality showed similar patterns with RR for highest compared with the lowest quartile of 1.85 (CI 1.25 to 2.75). Weight loss was associated with increased all-cause mortality (RR 1.62, CI 1.18 to 2.23) and CVD mortality (RR 1.61, CI 1.05 to 2.45). Weight cycling was also associated with increased all-cause mortality (RR for cycling ending with loss, 1.76 [CI 1.23 to 2.50], RR for cycling ending with gain, 1.53 [CI 1.13 to 2.07]) and increased CVD mortality (RR for cycling ending with loss, 1.73 [CI 1.08 to 2.79], RR for cycling ending with gain, 1.89 [CI 1.29 to 2.78]). The association between weight change and mortality was not observed in the heaviest men.


Greater weight variability was associated with increased all-cause and cardiovascular mortality in some groups of men at high risk for cardiovascular disease.

Source of funding: Not stated.

For article reprint: Dr. S.N. Blair, Cooper Institute for Aerobics Research, 12330 Preston Road, Dallas, TX 75230, USA. FAX 972-341-3224.


Body weight changes and all-cause mortality: a reviewandWeight loss and mortality in adults

Treating a patient who is obese is not a trivial problem; it may occur rather frequently in the practice of internal medicine. Obesity is a risk factor for multiple disorders, from cholecystitis to cardiovascular disease and certain carcinomas. Yet, does a reduction in disease occur with a reduction in weight? These 3 articles, in a weight-loss supplement to Annals of Internal Medicine, provide some information about this topic.

The article by Andres and colleagues reviews the literature about the association between weight change and mortality. The authors specifically excluded studies that purported to show a relation between weight loss and decreased mortality, which were reviewed in an article in the same supplement (1). Some of the articles reviewed by Williamson and Pamuk (2) appear to support a reduction in mortality with a reduction in weight. Taken together, the medical literature does not convincingly show that weight loss is associated with increased longevity. The lack of evidence for decreased mortality with a decrease in weight should be of concern, however, because weight reduction might be expected to decrease the various diseases associated with obesity that are risk factors for cardiovascular mortality, such as hypertension and diabetes mellitus.

The second article, by Pamuk and colleagues, analyzes a cohort drawn from the NHANES I study. The principal purpose was to analyze the effect of cigarette smoking and preexisting illness on the relation between weight loss and increased mortality in this data set. The authors found that preexisting illness may account for some of the effect in the noncardiovascular deaths, but an adverse effect of weight loss on mortality still existed.

The third article, by Blair and colleagues, analyzed the effect of body-weight change in the MRFIT data set. This provided further evidence that weight variability is associated with an increased mortality rate. Those with either weight loss only or weight cycling tended to have the highest relative risks for mortality, whereas the lowest risk tended to occur in the stable-weight group. This adds further information that weight cycling, the repeated gain and loss of weight, may be detrimental to survival, as reported elsewhere (3).

Noteworthy limitations exist in these reviews. The studies reviewed were heterogeneous and included persons of various age groups. The authors of the reviews chose substantially different criteria for inclusion and exclusion. The length of follow-up observation was variable. The weight of the participants may have been biased because it was often taken from a history rather than by objective measurement. The issue of weight caused by adipose tissue, rather than lean muscle mass or edema, was not addressed. Generally, only 2 "measures" of weight were used, and these may not be accurate reflections of a person's lifetime weight history. Confounders, such as smoking status, were handled in various ways. They may have been excluded or adjusted for, or the status may not have been assessed.

Most evidence suggests that the main emphasis of health care efforts should be on the primary prevention of obesity through life-long maintenance of a normal weight. Of course, this segment of the population tends not to be seen in the practice of internal medicine. Sports or pre-employment physical examinations, however, may provide an opportunity for this type of intervention.

Management of the obese adult is a separate issue. Specifically, should he or she be advised to lose weight? The highest mortality risk group appears to be those with weight cycling. This type of behavior should be discouraged based on most of the evidence. For those who are at or slightly above normal weight, maintenance of that weight should be the goal. The picture is less clear, however, for patients who are morbidly obese. Convincing evidence of mortality reduction with weight reduction is not available. Yet, obesity is a risk factor for multiple morbid and mortal conditions. Consequently, gradual and sustained weight reduction appears most warranted at present.

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