Five-decade trajectories in body mass index in relation to dementia death: follow-up of 33,083 male Harvard University alumni.

ABSTRACT

BACKGROUND: In prospective cohort studies, obesity has been linked with a lower risk of subsequent dementia. Reverse causality, whereby neurodegeneration preceding overt dementia symptoms may lower weight, is a possible explanation of these findings. To explore further the weight-dementia association we followed people from early adulthood, an age at which neurodegeneration has typically yet to begin. METHODS: In all, 33,083 male participants in the Harvard Alumni Health Study underwent a medical examination as undergraduates (typically aged 18 years) during which height, weight, resting pulse rate, blood pressure, physical activity, and smoking status were assessed. Subsamples provided height and weight in 1962/6 (mean age 50.7 years), 1977 (58.6), 1988 (67.5), and 1993 (71.1). Dementia deaths were extracted from death certificates (mean follow-up 53.1 years). We used latent class mixed models to create body mass index (BMI) trajectories; for comparison, we also constructed models with cardiovascular disease (CVD) death. RESULTS: We found no association between early life BMI and subsequent dementia (age-adjusted HR 0.94, 95% CI 0.85, 1.04). We identified two latent class groups based on different BMI trajectories-"early decliners" whose BMI began to decline around age 50 years and "late decliners" whose BMI declined about two decades later. The former experienced a raised risk of dementia-related death compared to the latter (multivariable-adjusted HR 1.57, 95% CI 1.14, 2.17). Expected associations were identified between CVD risk factors and CVD death. CONCLUSIONS: In a population likely to be free of dementia neuropathology at BMI measurement, we found no association between BMI at baseline and subsequent dementia-related death. Earlier decline in BMI was, however, associated with dementia, which suggests that findings associating BMI with dementia risk may be influenced by reverse causality.