Drinking any amount of alcohol likely increases the risk ofdementia, suggests the largest combined observational and genetic study to date, published online inBMJ Evidence Based Medicine. Even light drinking-generally thought to be protective, based on observational studies-is unlikely to lower the risk, which rises in tandem with the quantity ofalcoholconsumed, the research indicates. Current thinking suggests that there might be an ‘optimal dose’ of alcohol forbrain health, but most of these studies have focused on older people and/or didn’t differentiate between former and lifelong non-drinkers, complicating efforts to infer causality, note the researchers. To try and circumnavigate these issues and strengthen the evidence base, the researchers drew on observational data and genetic methods (Mendelian randomisation) from two large biological databanks for the entire ‘dose’ range of alcohol consumption. These were the US Million Veteran Program (MVP), which includes people of European, African, and Latin American ancestry, and the UK Biobank (UKB), which includes people of predominantly European ancestry. Participants who were aged 56–72 at baseline, were monitored from recruitment until their first dementia diagnosis, death, or the date of last follow-up (December 2019 for MVP and January 2022 for UKB), whichever came first. The average monitoring period was 4 years for the US group, and 12 for the UK group. Alcohol consumption was derived from questionnaire responses—over 90% of participants said they drank alcohol—and the Alcohol Use Disorders Identification Test (AUDIT-C) clinical screening tool. This screens for hazardous drinking patterns, including the frequency of binge drinking (6 or more drinks at a time). In all, 559,559 participants from both groups were included in observational analyses, 14,540 of whom developed dementia of any type during the monitoring period:10,564 in the US group; and 3976 in the UK group. And 48,034 died: 28,738 in the US group and 19,296 in the UK group. Observational analyses revealed U-shaped associations between alcohol and dementia risk: compared with light drinkers (fewer than 7 drinks a week) a 41% higher risk was observed among non-drinkers and heavy drinkers consuming 40 or more drinks a week, rising to a 51% higher risk among those who were alcohol dependent. Mendelian randomisation genetic analyses drew on key data from multiple large individual genome-wide association studies (GWAS) of dementia, involving a total of 2.4 million participants to ascertain lifetime (rather than current) genetically predicted risks. Mendelian randomisation leverages genetic data, minimising the impact of other potentially influential factors, to estimate causal effects: genomic risk for a trait (in this case, alcohol consumption) essentially stands in for the trait itself. Three genetic measures related to alcohol use were used as different exposures, to study the impact on dementia risk of alcohol quantity, as well as problematic and dependent drinking. These exposures were: self-reported weekly drinks (641 independent genetic variants); problematic ‘risky’ drinking (80 genetic variants); and alcohol dependency (66 genetic variants). Higher genetic risk for all 3 exposure levels was associated with an increased risk of dementia, with a linear increase in dementia risk the higher the alcohol consumption. For example, an extra 1-3 drinks a week was associated with a 15% higher risk. And a doubling in the genetic risk of alcohol dependency was associated with a 16% increase in dementia risk. But no U-shaped association was found between alcohol intake and dementia, and no protective effects of low levels of alcohol intake were observed. Instead, dementia risk steadily increased with more genetically predicted drinking. What’s more, those who went on to develop dementia typically drank less over time in the years preceding their diagnosis, suggesting that reverse causation-whereby early cognitive decline leads to reduced alcohol consumption-underlies the supposed protective effects of alcohol found in previous observational studies, say the researchers. They acknowledge that a principal limitation of their findings is that the strongest statistical associations were found in people of European ancestry, because of the numbers of participants of this ethnic heritage studied. Mendelian randomisation also relies on assumptions that can’t be verified, they add. Nevertheless, they suggest that their findings “challenge the notion that low levels of alcohol are neuroprotective.” And they conclude: “Our study findings support a detrimental effect of all types of alcohol consumption on dementia risk, with no evidence supporting the previously suggested protective effect of moderate drinking. “The pattern of reduced alcohol use before dementia diagnosis observed in our study underscores the complexity of inferring causality from observational data, especially in ageing populations. “Our findings highlight the importance of considering reverse causation and residual confounding in studies of alcohol and dementia, and they suggest that reducing alcohol consumption may be an important strategy for dementia prevention.” Topiwala, A., et al. (2025) Alcohol use and risk of dementia in diverse populations: evidence from cohort, case-control and Mendelian randomisation approaches. BMJ Evidence Based Medicine. DOI: 10.1136/bmjebm2025-113913