Abstract Background Previous studies have not typically separated body mass index (BMI) slope and variability as distinct constructs when examining dementia risk. This study assessed the association between the slope-adjusted visit-to-visit BMI variability and dementia risk. Subjects/methods We conducted a retrospective cohort study using Japanese national health insurance data (2015–2023) of individuals aged 50–74 years who underwent five annual health checkups. BMI variability was assessed using the slope-adjusted standard deviation (SD) to account for underlying temporal trends. The proxy outcome for dementia was antidementia drug initiation, analyzed using Fine-Gray competing risk models, accounting for death as a competing risk. Results During the mean 2.17 ± 1.19 years of follow-up among 303,042 participants (mean age: 66.6 years, men: 38.6%), antidementia drugs (predominantly donepezil: 67.4%) were initiated in 665 and 2394 died. After adjusting for covariates including BMI at baseline and annual BMI change, the highest tertile of slope-adjusted BMI-SD (≥0.50 kg/m²) was significantly associated with increased dementia risk compared with the lowest tertile (≤0.31 kg/m²). Annual BMI change showed a U-shaped association with dementia risk, with pronounced elevation in the first tertile (BMI decline ≤−0.31%, hazard ratio: 1.60, 95% confidence interval: 1.32–1.93). In the basic model including baseline covariates except BMI at baseline, there was no significant difference in the C-statistics improvements when BMI at baseline or adding slope-adjusted BMI-SD (+0.0147 vs +0.0146) were added, while the greatest C-statistics improvement was observed when BMI decline ≤−0.31% was added. The association between the highest slope-adjusted BMI-SD tertile and dementia risk was stronger in females than males ( P for interaction = 0.0039). Conclusions Slope-adjusted visit-to-visit BMI variability is independently associated with dementia risk, particularly among females, while BMI decline patterns are strong risk factors of dementia. Incorporating longitudinal monitoring of visit-to-visit BMI variability into routine dementia screening may be beneficial.
Satoh et al. (Fri,) studied this question.