Abstract
BACKGROUND: Body weight variability (BWV) refers to intraindividual weight loss and gain over a period. The association of long-term BWV with dementia remains unclear and whether this association is beyond body weight change is undetermined. METHODS: In the Health and Retirement Study, a total of 5 547 dementia-free participants (56.7% women; mean [SD] age, 71.1 [3.2] years) at baseline (2008) were followed up to 8 years (mean = 6.8 years) to detect incident dementia. Body weight was self-reported biennially from 1992 to 2008. BWV was measured as the coefficient of variation utilizing the body weight reported 9 times across 16 years before baseline. Cox-proportional hazard model was used to estimate the hazard ratio (HR) and 95% confidence interval (CI). RESULTS: Among the 5 547 participants, a total of 427 incident dementia cases were identified during follow-up. Greater long-term BWV was significantly associated with a higher risk of dementia (HR comparing extreme quartiles: 2.01, 95% CI: 1.48-2.72; HR of each SD increment: 1.21, 95% CI: 1.10-1.32; p-trend < .001) independent of mean body weight and body weight change. This significant association was even observed for BWV estimated approximately 15 years preceding dementia diagnosis (HR of each SD increment: 1.13, 95% CI: 1.03-1.23) and was more pronounced for that closer to diagnosis. CONCLUSION: Our prospective study suggested that greater BWV may be a novel risk factor for dementia.