An “obesity paradox,” in which overweight and obese individuals with established cardiovascular disease have a better prognosis than normal weight subjects, has been reported in a number of clinical cohorts, but little is known about the effects of weight loss on the obesity paradox and its association with health outcomes.

Weight was determined in 3834 men at the time of a clinically referred exercise test and again during a clinical evaluation a mean of 7 years later. The associations among weight changes, baseline fitness, and other risk markers with cardiovascular and all-cause mortality were determined by Cox proportional hazards analysis.

During the follow-up period, 314 subjects died (72 of cardiovascular causes). In a multivariate analysis (including baseline weight, weight change, exercise capacity, and cardiovascular disease), weight gain was associated with lower mortality and weight loss was associated with higher mortality (4% higher per pound lost per year, P<.001) compared with stable weight. For all-cause mortality, the relative risks for the no change, weight gain, and weight loss groups were 1.0 (referent), 0.64 (95% confidence interval, 0.50-0.83), and 1.49 (95% confidence interval, 1.17-1.89), respectively (P<.001). Those who died and exhibited weight loss had a significantly higher prevalence of deaths due to cancer and cardiovascular causes.

Weight loss was related to higher mortality and weight gain was related to lower mortality when compared with stable weight. The obesity paradox in our sample is explained in part by a combination of non-volitional weight loss related to occult disease and a protective effect of weight gain.