Chen · European journal of epidemiology 2025 · prospective cohort study · n=296,715

Associations of alcohol drinking with incident dementia: a prospective study from the UK Biobank.

Cited 4 times in the scientific literature.

Level 3 - non-randomized controlled study

Prospective observational cohort study

PubMed 41014391 · doi:10.1007/s10654-025-01304-y · record verified 2026-08-29

What was done

Analyzed prospective cohort data from 296,715 UK Biobank participants (mean age 56.54 years) followed for a median of 13.7 years, excluding participants with baseline dementia, dementia onset within two years of follow-up, or infrequent drinking patterns. Alcohol intake was categorized as non-drinking, low-moderate, or heavy drinking based on weekly units, and evaluated alongside drinking behaviors (drinking with meals, beverage type). Incident all-cause dementia was tracked as the primary outcome (4,242 cases). Multivariable Cox proportional hazards models estimated hazard ratios, with subgroup stratifications by cardiovascular disease (CVD) risk, APOE4 carrier status, and sex.

What was found

Compared with non-drinking, low-moderate alcohol consumption was associated with a 35% reduction in incident dementia risk (HR 0.65; 95% CI, 0.59-0.73). Heavy drinking showed no statistically significant association with dementia risk (HR 0.88; 95% CI, 0.75-1.02). The inverse association for low-moderate drinking persisted across stratified subgroups: high CVD risk (HR 0.66; 95% CI, 0.59-0.74), low CVD risk (HR 0.43; 95% CI, 0.30-0.61), APOE4 carriers (HR 0.71; 95% CI, 0.61-0.83), APOE4 non-carriers (HR 0.61; 95% CI, 0.52-0.71), females (HR 0.67; 95% CI, 0.58-0.77), and males (HR 0.63; 95% CI, 0.53-0.76).

Why it matters

These findings contribute large-scale longitudinal data showing that low-to-moderate alcohol intake is inversely associated with dementia risk across diverse genetic and cardiovascular risk profiles.

Limits

Observational design precludes causal inference. Alcohol consumption was self-reported, exposing data to misclassification and recall bias. The non-drinking reference group was not separated into lifetime abstainers versus former drinkers, introducing potential sick-quitter bias, and the UK Biobank cohort exhibits a healthy volunteer selection bias that limits generalizability.

Cited by