Sabia · BMJ (Clinical research ed.) 2018 · prospective cohort study · n=9087

Alcohol consumption and risk of dementia: 23 year follow-up of Whitehall II cohort study.

Cited 285 times in the scientific literature.

Level 3 - non-randomized controlled study

Prospective observational cohort study

PubMed 30068508 · doi:10.1136/bmj.k2927 · record verified 2026-08-29

What was done

Prospective cohort analysis of 9,087 London civil servants aged 35–55 at inception (1985/1988) in the Whitehall II study, tracked for a mean follow-up of 23 years (through 2017). Midlife alcohol intake was calculated from three assessments (1985/88 to 1991/93) and classified into abstinence, 1–14 units/week, and >14 units/week. Additional measures included 17-year intake trajectories (five assessments between 1985/88 and 2002/04), CAGE questionnaire screening in 1991/93, and alcohol-related hospital admissions. Incident dementia was ascertained via linkage to hospital, mental health services, and mortality registries.

What was found

Over follow-up, 397 dementia cases occurred. Compared to consuming 1–14 units/week, midlife abstinence was associated with a higher risk of dementia (hazard ratio 1.47, 95% CI 1.15 to 1.89); this excess risk was partly explained by cardiometabolic disease (HR 1.33, 95% CI 0.88 to 2.02 among abstainers without cardiometabolic disease). For consumption >14 units/week, each 7 unit/week increase was associated with a 17% increase in dementia risk (95% CI 4% to 32%). A CAGE score >2 (HR 2.19, 95% CI 1.29 to 3.71) and alcohol-related hospital admissions (HR 4.28, 95% CI 2.72 to 6.73) were also linked to increased dementia risk. Trajectory analyses showed higher risk for long-term abstinence (HR 1.74, 95% CI 1.31 to 2.30), decreasing consumption (HR 1.55, 95% CI 1.08 to 2.22), and long-term consumption >14 units/week (HR 1.40, 95% CI 1.02 to 1.93).

Why it matters

This study provides long-term longitudinal evidence that drinking above 14 units/week in midlife increases dementia risk, supporting the lowering of clinical and public health thresholds for harmful alcohol intake.

Limits

The observational design cannot establish causality or rule out residual confounding. Alcohol intake was self-reported and prone to reporting bias. Dementia cases were identified solely through administrative health and mortality records, likely under-ascertaining milder cases, and the cohort consisted of British civil servants, which may limit generalizability.

Cited by