Alcohol consumption from midlife and risk of disabling dementia in a large population-based cohort study in Japan.

Objectives: The association between alcohol consumption and dementia in Japanese is poorly understood, and use of single-point alcohol assessment may cause measurement error. We explored this association in Japanese using repeated alcohol assessments.

Methods: Participants in the Japan Public Health Center-based Prospective Study (JPHC Study) since 1990 and who were alive in 2006 were followed from 2006 until 2016 for dementia ascertainment. Disabling dementia was identified through long-term care insurance records. Alcohol consumption was assessed at the 5-year questionnaire survey (1995-1999) and drinking patterns were assessed on repeated follow-up (2000-2003). We performed Cox proportional hazards models with age as the time-scale with adjustment for various lifestyle factors and medical history using light consumption (<75 g ethanol/week, hereinafter "g") as reference. Analysis considering death as a competing risk was also conducted.

Results: Among 42,870 participants aged 54-84 years, 4802 cases of disabling dementia were newly diagnosed. Average years from alcohol assessment until dementia incidence was 14.9 years. Non-drinkers and regular drinkers with ≥450 g at 5 years had adjusted HRs (95% CI) of 1.29 (1.12-1.47) and 1.34 (1.12-1.60). Patterns of long-term abstinence, former drinking, and regular heavy weekly consumption of ≥450 g showed increased adjusted HRs of 1.61 (1.28-2.03), 2.54 (1.93-3.35), and 1.96 (1.49-2.59), respectively. Competing risk analysis yielded similar results.

Conclusions: In Japanese, non-drinking and regular weekly consumption of ≥450 g from midlife were associated with high risk of disabling dementia compared with light drinking.