Weight loss and mortality: a gender-specific analysis of the Tromsø study

Weight loss has been associated with increased mortality, but findings have been inconsistent.

The aim of this study was to examine the association between weight loss and mortality, with a focus on gender differences.

This was a population-based cohort study in northern Norway of adults, aged 20 to 54 years in 1979, who participated in 2 or 3 consecutive health surveys in 1979-80, 1986-87, and 1994-95. Weight and height were measured at each survey. The Cox proportional hazards regression model was used to estimate hazard ratios for mortality between levels of body mass index (BMI) change during 11 years of follow-up. Participants with prior cardiovascular disease or cancer, or incident cancer within the first 2 years of follow-up, were excluded, as were participants who were pregnant, had missing data, or did not give written consent.

A total of 4881 men and 5051 women participated in the present study. The mean age at start of follow-up was 50.8 years (range, 35-70 years) in men and 49.2 years (range, 35-65 years) in women. In men, weight loss was associated with increased all-cause, cardiovascular, and noncardiovascular mortality. The hazard ratio for men for all-cause mortality with a 10-year BMI decrease of 2 kg/m(2) versus a BMI increase of 1 kg/m(2) was 2.09 (95% CI, 1.56-2.81). The association was not significantly modified by initial BMI, age, smoking status, or self-reported attempts of weight loss, or by exclusion of subjects with self-reported poor health, diabetes mellitus, high blood pressure, or high alcohol intake. In women, no association between BMI change and mortality was observed. However, in the subgroup of women who reported no weight-loss attempts, BMI change was significantly associated with mortality risk (P = 0.022).

In this study of a Norwegian population, weight loss was associated with excess mortality in men in all subgroups of weight-loss attempts, daily smoking, and overweight. In women, the only significant effect of BMI change on mortality was observed in those who reported no weight-loss attempts. The observed findings could not be explained by preexisting disease.