Alcohol consumption and risk of dementia

Alcohol occupies a unique place in our societies. A cultural, social, and sometimes identity-defining product, it is also one of the most widely consumed psychoactive substances in the world. For decades, its impact on cardiovascular, metabolic, and liver health has been widely documented. However, when it comes to brain health and cognitive aging, the scientific message has long been unclear, oscillating between health warnings and reassuring statements about the supposed benefits of moderate consumption.

For a long time, numerous observational studies suggested a “J” or “U” relationship between alcohol and dementia: abstainers and heavy drinkers were thought to be at higher risk, while light drinkers appeared to be protected. This idea had a profound impact on collective perceptions, to the point of fueling the idea that a daily drink could be beneficial for the brain. However, this research was often based on flawed methodologies, subject to major biases related to age, lifestyle, socioeconomic status, and changes in consumption patterns over time. In a context of rapid demographic aging and an expected explosion in dementia cases, clarifying the real effects of alcohol has become a major public health issue.

The study

This is precisely what an international team has attempted to do by combining traditional observational approaches with modern genetic methods in order to overcome the limitations of previous studies and better understand the relationship between alcohol and the risk of dementia throughout adulthood. To do this, the researchers analyzed data from two large cohorts: the Million Veteran Program in the United States and the UK Biobank in the United Kingdom. In total, more than 559,000 adults, aged 56 to 72 at the time of inclusion, were followed for several years, with an average follow-up of approximately 4 years in the US cohort and more than 12 years in the UK cohort.

Alcohol consumption was assessed using standardized questionnaires, expressed in number of drinks per week (one “drink” containing an average of 14g of pure ethanol), as well as clinical scores to identify patterns of problematic consumption or dependence. Cases of dementia were identified through medical records and diagnostic codes, carefully excluding cases already present at the time of inclusion in order to limit biases related to pre-existing disease.

However, the major originality of this work lies in the integration of Mendelian randomization analyses. This approach uses genetic variants associated with alcohol consumption or alcohol-related disorders as instruments to estimate the effects of exposure over a lifetime, regardless of behaviors reported at a given point in time. In practice, this largely eliminates the confounding and reverse causality biases that are common in traditional observational studies.

The authors combined prospective cohort analyses, case-control analyses, longitudinal consumption trajectories, and genetic approaches across populations of different ethnic origins. This methodological triangulation aimed to answer a central question: is there really a level of alcohol consumption that is safe, or even protective, for the brain?

Results & Analysis

The main results of this study show that conventional observational analyses initially reproduce what the literature has been reporting for years.

When only self-reported data are analyzed, a nonlinear relationship emerges: abstainers, heavy drinkers (more than 40 “drinks” per week), and people with alcohol-related disorders have a higher risk of dementia, while light drinkers (7 to 14 drinks per week) appear to be in a lower risk zone. This result could support the idea that moderate consumption has a protective effect.

However, when researchers examine behavioral changes over time more closely, a key phenomenon emerges. In many people who develop dementia, alcohol consumption gradually decreases in the years preceding diagnosis. In other words, some of the abstainers or light drinkers observed in the cohorts are not abstainers, but individuals who have reduced or even stopped drinking due to emerging health problems. This early decline in consumption, linked to early cognitive symptoms or increasing frailty, creates the statistical illusion of a high-risk group of abstainers, when in fact the causal link is reversed.

When researchers focus on genetic predispositions associated with alcohol consumption or problematic drinking behaviors, the relationship becomes linear. The higher the exposure to alcohol, the greater the risk of dementia, with no identifiable protective threshold. For example, an increase from 1 to 3 drinks per week was associated with a 15% increase in the risk of dementia. This relationship is observed for the amount of alcohol consumed, but also for patterns of problematic consumption and dependence.

In other words, once confounding and reverse causality biases have been largely neutralized, the hypothesis of a neuroprotective effect of alcohol becomes null. The results suggest that even relatively low levels of consumption could contribute, in the long term, to an increased risk of dementia, and that the deleterious effects of alcohol on the brain accumulate over a lifetime. These findings are consistent with independent data from brain imaging, showing associations between alcohol consumption and reduced brain volume, alterations in white matter, or iron accumulation in certain regions of the brain, even at levels of consumption considered moderate.

Practical applications

From a practical standpoint, these findings suggest that the argument that moderate alcohol consumption is potentially beneficial for the brain is based more on methodological artifacts than on a genuine protective biological effect. While some people tolerate alcohol better than others in metabolic or cardiovascular terms, current data suggest that the brain does not benefit from repeated exposure, even at low doses. The question is therefore no longer one of occasional consumption, but of cumulative exposure over several decades.

In terms of prevention, this research reinforces the idea that reducing alcohol consumption, particularly problematic consumption, could be an important lever for reducing the future incidence of dementia. Unlike non-modifiable factors such as age or genetics, alcohol consumption is an adjustable behavior that can be influenced by individual and collective interventions.

Finally, these results highlight the importance of interpreting data rigorously and patiently. Public health messages based on uncorrected observational associations can, when misinterpreted, unintentionally encourage harmful behaviors. In the case of alcohol, caution is warranted: in the absence of proven benefits for the brain, moderation or even abstinence appears to be the most consistent option based on current knowledge about cognitive aging.

Reference

Topiwala A, Levey DF, Zhou H, Deak JD, Adhikari K, Ebmeier KP, Bell S, Burgess S, Nichols TE, Gaziano M, Stein M & Gelernter J. Alcohol use and risk of dementia in diverse populations: evidence from cohort, case-control and Mendelian randomisation approaches. BMJ Evidence-Based Medicine 2025, Sep 23:bmjebm-2025-113913.