The prevalence of dementia is increasing globally and is linked to obesity and unfavorable dietary habits. The present study analyses the association of alcohol intake from wine and non-wine alcoholic beverages (non-wine) in g/d, as well as coffee and tea in cups/d, with incident dementia. Over 4.2 million person-years, 4270 dementia cases occurred in 351,436 UK Biobank participants. Hazard ratios (HRs) for incident dementia were defined with Cox proportional hazard regression models in which beverage intake was fitted as penalized cubic splines. Wine intake showed a significant U-shaped association with the lowest risk for incident dementia (nadir) ranging from 21 to 23 g alcohol/d in all participants and in males. In contrast, non-wine consumption was significantly and dose-dependently associated with incident dementia, and the nadir was found at 0 g alcohol/d. Coffee consumption was not related to dementia risk, while moderate-to-high tea intake was negatively associated with incident dementia. Taken together, the current study shows on a population level that moderate consumption of wine and moderate-to-high tea intake is associated with a decreased risk of incident dementia. In contrast, non-wine is positively related to dementia risk in a linear fashion, and no clear association is found for coffee.
Keywords: alcohol; body weight; coffee; dementia; non-wine; obesity; prospective cohort study; tea; wine.