Overview

• The analysis, published in BMJ Evidence-Based Medicine, pooled cohort data from 559,559 adults in the UK Biobank and US Million Veteran Program with Mendelian randomization across genome studies totaling about 2.4 million people.
• Observational results showed a U-shaped pattern: compared with light drinkers (under seven drinks a week), non-drinkers and those consuming 40 or more drinks weekly had a 41% higher risk of dementia, rising to 51% among people with alcohol dependence.
• Genetic analyses indicated no protective effect at low intake and a steady increase in risk with greater exposure, with roughly a 15% higher risk linked to an extra 1–3 drinks per week and a 16% increase with a doubling of genetic risk for alcohol dependence.
• Repeated screening data suggested people who later developed dementia tended to reduce drinking before diagnosis, supporting reverse causation as a contributor to earlier findings that made light use appear protective.
• Researchers and outside experts noted caveats including self-reported intake, assumptions underlying Mendelian randomization, and stronger statistical signals in European-ancestry groups, while stating that population-wide reductions in drinking could help prevent dementia.