A Population-Based Study From the Swedish Twin Registry

Background: Midlife alcohol consumption (beer, wine, and spirits) was examined in relation to dementia incidence over 43 years. Methods: Participants were 12,326 members of the population-based Swedish Twin Registry born during 1907–1925 who responded to items about alcohol consumption in 1967/1970, subsequently classified as nondrinking (0 grams of ethanol per day), light (1–5 g/d), moderate (5–12 g/d), heavy (12–24 g/d), and very heavy (>24 g/d) drinking. Dementia was identified from the National Patient and Cause of Death Registries. Cox proportional hazard models adjusted for cluster-correlated data were used in cohort analyses. Conditional logistic regression (dementia-discordant pairs) and mixed effects models (dementia-concordant pairs) were used in twin analyses. Results: Overall, nondrinkers did not differ from light drinkers in dementia risk. Heavy drinking (hazard ratio = 1.10, p = .028) and very heavy drinking (hazard ratio = 1.18, p = .033) were associated with increased dementia risk controlling for sociodemographic, lifestyle, and cardiovascular factors. More alcohol from spirits was related to increased risk of dementia, whereas more alcohol from wine with decreased risk, although the association for wine reversed direction at high amounts