Late-night dinner, skipping breakfast, and the risk of dementia in Japan: the LIFE Study

Evidence on meal timing and dementia is limited. We examined late-night dinner and skipping breakfast in relation to incident dementia. We conducted a population-based cohort study using linked medical insurance claims, long-term care insurance claims, and government-subsidized health checkup data from 19 participating in the Longevity Improvement & Fair Evidence (LIFE) Study. Cohort entry was from April 2016 to March 2021. Late-night dinner (dinner within 2 h of bedtime ≥ 3 times/week) and skipping breakfast (≥ 3 times/week) were assessed at health checkup and analyzed as dichotomous exposures (yes/no). Incident all-cause dementia was identified in claims data using ICD-10 codes. Associations were estimated using cause-specific Cox proportional hazards models, treating death as censoring and reporting hazard ratios with 95% confidence intervals, with prespecified adjustment for baseline demographic, lifestyle, and comorbidity factors. Among 283,166 participants (mean [standard deviation] age, 72.8 [5.5] years; 58.5% women), late-night dinner was reported by 36,609 (12.9%) and skipping breakfast by 17,727 (6.3%). During follow-up, 18,056 incident dementia cases and 16,390 deaths occurred. Both late-night dinner (hazard ratio, 1.17; 95% confidence interval, 1.12-1.20) and skipping breakfast (1.13; 1.06-1.20) were associated with higher hazards of incident all-cause dementia. Late-night dinner and skipping breakfast were associated with a higher risk of incident all-cause dementia in a large cohort of older Japanese adults. Meal timing behaviors assessed during routine health checkups may be relevant for dementia prevention strategies, although causal inference requires further study.