Abstract
BACKGROUND AND OBJECTIVES: Midlife obesity is a risk factor of dementia, but late-life obesity has been associated with lower dementia risk. We investigated this paradox by exploring the relationship between late-life body mass index (BMI) category and dementia, with and without considering midlife to late-life BMI change. METHODS: This observational cohort study included participants of the community-based Atherosclerosis Risk in Communities (ARIC) study who were dementia-free at visit 5 (2011-2013). Dementia was ascertained by expert-adjudicated, algorithmic classification from an in-person neuropsychological battery, as well as telephone interviews and International Classification of Diseases codes from medical records. We first assessed the association of incident dementia with visit 5 BMI categories (normal weight, overweight, obese). Next, we used a cross-classification of visit 5 BMI categories with visit 4-visit 5 BMI change (decrease [loss of ≥2 kg/m(2)], increase [gain of ≥2 kg/m(2)], or stable [loss or gain of <2 kg/m(2)]) occurring during the 15 years before baseline. Cox regression was used. RESULTS: A total of 5,129 participants were included in the study (59% female; 22% identified as Black; mean (standard deviation) age at visit 5 of 75 (5) years). Over 8 years of follow-up, 20% of the sample developed dementia (n = 1,026). After covariate adjustment, participants with high late-life BMI had a lower risk of dementia; the hazard ratio (95% CI) was 0.86 (0.73-1.00) for overweight and 0.81 (0.68-0.96) for obesity. In stratified models, elevated dementia risk was observed only for participants of each late-life BMI category whose BMI had decreased from midlife to late life. Compared with normal-weight individuals who had maintained BMI from midlife to late life, the hazard ratio (95% CI) for those with BMI loss was 2.08 (1.62-2.67) for normal-weight individuals, 1.62 (1.25-2.10) for those with overweight, and 1.36 (1.00-1.85) for those with obesity. DISCUSSION: Our results provide insight into the dementia obesity paradox at older ages, tempering a causal interpretation of high late-life BMI as protective against dementia. Instead, they highlight the importance of considering weight loss from midlife to late life in conjunction with late-life BMI in dementia risk stratification.