Effects of Weight History on the Association Between Directly Measured Adiposity and Mortality in Older Adults

Background: It is controversial whether an altered relationship between adiposity and mortality occurs with aging. We evaluated associations between adiposity and mortality in younger and older participants before and after considering historical weight loss. Methods: This study used whole-body dual-energy x-ray absorptiometry data from the National Health and Nutrition Examination Survey in adults at least 20 years of age. Fat mass index (FMI), determined by dual-energy x-ray absorptiometry, was converted to age-, sex-, and race-specific Z-Scores. Percent change in weight from the maximum reported weight was determined and categorized. Cox proportional hazards models assessed associations between quintile of FMI Z-Score and mortality. Sequential models adjusted for percent weight change since the maximum weight. Results: Participants with lower FMI were more likely to have lost weight from their maximum, particularly among older participants with lower FMI. Substantially greater risk of mortality was observed for the highest quintile of FMI Z-Score compared to the second quintile among younger individuals [HR 2.50 (1.69, 3.72) p < .001]. In contrast, a more modest association was observed among older individuals in the highest quintile [HR 1.23 (0.99, 1.52) p = .06] (p for interaction <.001). In both the younger and older participants, the risks of greater FMI Z-Score were magnified when adjusting for percent weight change since maximum reported weight. Conclusions: Older people with low fat mass report greater historical weight loss, potentially explaining substantially altered relationships between fat mass and mortality in older individuals. As a result, epidemiologic studies performed in older populations will likely underestimate the causal risks of excess adiposity.