UCL Queen Square Institute of Neurology


No proof that moderate drinking prevents dementia

2 August 2018

This article has been reproduced from NHS Choices.

"Middle aged drinking may reduce dementia risk, new study finds," is the misleading and irresponsible headline in The Daily Telegraph.

The study found that people who do not drink alcohol in middle age are 45% more likely to develop dementia than those who drink within UK recommended limits (no more than 14 units a week). But the same study found that dementia risk also increased for people who drank over the recommended limits.

Researchers used data from an ongoing project that has been following more than 9,000 London civil servants since 1985 when they were aged 35 to 55. Over the years, the researchers have documented their drinking habits, lifestyle and health. A total of 397 have so far developed dementia. Dementia was more likely if people smoked, were obese, had cardiovascular disease or had diabetes.

While it's true that people who did not drink, or who had the occasional glass, were also found to be more likely to develop dementia, we cannot say that alcohol protects against dementia. We do not know how much they drank when they were younger.

These higher-risk people may have stopped drinking because of health worries, or possibly because some had concerns about their alcohol use when they were younger.

Also, it's worth noting that those who did not drink alcohol and did not have cardiovascular disease or diabetes were not at increased risk of dementia.

The known ways you can help reduce your dementia risk include doing regular exercise, eating a healthy diet and quitting smoking if you smoke.

Where did the story come from?

The study was carried out by researchers from the Université Paris-Saclay and Université Paris Diderot, and University College London. It was funded by the US National Institute on Aging, the UK Medical Research Council and the British Heart Foundation.

The study was published in the peer-reviewed British Medical Journal on an open-access basis, so it is free to read online.

In general, the UK media reported the study accurately. The Mail Online included a responsible quote from one of the researchers, Severine Sabia, who said: "This should not motivate people who do not drink to start drinking due to adverse effects of alcohol on mortality, cirrhosis of the liver and cancer." However, the Mail Online has overstated the case that "low" alcohol intake protects against cardiovascular disease and strokes as we don't know how much alcohol people drank in earlier adulthood. They may have stopped drinking because of developing cardiovascular disease or diabetes.

And, as so often is the case, some headline writers oversimplified the issue. As well as the Telegraph's misleading headline, The Sun claim's that "Drinking six pints of beer or glasses of wine a week could save you from deadly dementia," is unsupported.

What kind of research was this?

This was a cohort study where a group of people are observed over a period of time to see if outcomes are different for those exposed to something – in this case, varying amounts of alcohol – compared with those who are not exposed to it. This type of study is the best available when a randomised controlled trial (RCT) is not possible.

Unfortunately though, the groups cannot be matched in terms of age, demographics and other health and lifestyle factors, as they can with an RCT. So a cohort study is not able to prove cause and effect.

What did the research involve?

The researchers used data on 9,087 adults from the large ongoing Whitehall cohort study. This study started in 1985 and has been following up adults employed by the British civil service in London. The men and women were aged between 35 and 55 at the start of the study. Every 5 years since then, they have had a clinical assessment by a nurse and completed questionnaires about their lifestyle, including alcohol consumption.

This information was backed up by national Hospital Episode Statistics data and the Mental Health Services Data Set, to identify people with a diagnosis of dementia and any alcohol-related conditions.

The data was then analysed to see if different levels of alcohol consumption were linked to the likelihood of developing dementia. For this, they put people into 3 main groups:

  • abstainers (people who did not currently drink, including people who used to drink and those who have a very occasional drink)
  • those who regularly drank between 1 and 14 units per week
  • those who drank more than 14 units per week (above the recommended UK guidelines)

The results were adjusted to take the following potential confounding factors into account:

  • age
  • sex
  • ethnicity
  • education
  • occupational position
  • marital status
  • physical activity
  • smoking status
  • fruit and vegetable consumption
  • systolic blood pressure
  • total cholesterol
  • diabetes
  • body mass index
  • general health questionnaire score
  • cardiovascular disease
  • cardiovascular disease drugs

What were the basic results?

In total, 397 people developed dementia and this was more likely if they were smokers, were obese, had cardiovascular disease or had diabetes.

Compared with people who usually drank between 1 and 14 units per week:

  • people who did not drink alcohol were 45% more likely to develop dementia (adjusted hazard ratio [aHR] 1.45, 95% confidence interval[CI] 1.12 to 1.86)
  • for people who drank more than 14 units per week, every 7 excess units increased the risk of dementia by 18% (aHR 1.18, 95% CI 1.04 to 1.34)
  • people who did not drink alcohol and did not have cardiovascular disease or diabetes were not at increased risk of dementia (aHR 1.33, 95% CI 0.88 to 2.02)

How did the researchers interpret the results?

The researchers conclude that: "The risk of dementia was increased in people who abstained from alcohol in midlife or consumed >14 units/week. In several countries, guidelines define thresholds for harmful alcohol consumption much higher than 14 units/week. The present findings encourage the downward revision of such guidelines to promote cognitive health at older ages."


This well-conducted cohort study found that people who drink in moderation in middle age are less likely to develop dementia than those who abstain or drink excessively.

It is not entirely clear why those classed as abstainers were more likely to develop dementia. This group included people who used to drink, and theoretically this could have been to excess as younger adults.

It is interesting that the increased risk was no longer there when people with cardiovascular disease or diabetes were taken out of the group. This perhaps indicates that these conditions are higher risk factors for dementia and are possibly the reason for abstinence in middle age.

This is a strong piece of research and the results are likely to be reliable. Strengths include the long follow-up time and the repeated questionnaires. This provides a more accurate picture of lifestyle and alcohol consumption over time than many cohort studies that just provide baseline measurements.

The researchers tried to minimise the potential bias of people reporting lower alcohol use than they actually consumed by using hospital data on admissions related to excess alcohol use.

However, as with any cohort study there are some limitations.

While the researchers accounted for many potential confounding factors, there could have been other unmeasured factors, such as medication for other conditions, that could have affected the results.

The participants were all office-based workers in London, so the results may not be the same for the general population.

Some milder cases of dementia may have been missed, or people could have died from other conditions before dementia became apparent.

Cohort studies are observational, so they can only show an association between 2 factors – they cannot prove that one causes the other.

Nevertheless, the findings support current guidelines to drink alcohol only in moderation.