Reports that frequent drinking prevents diabetes are inaccurate

"Drinking a moderate amount of certain drinks such as wine three to four times a week reduced diabetes risk by about 30%," The Guardian reports. That was the main reported finding of a Danish study looking at the impact of alcohol on diabetes risk.

Researchers looked at a group of more than 70,000 people who had completed a survey about their health and lifestyle in 2007-2008, which included questions about their drinking habits. They then checked whether any of the participants had been diagnosed with diabetes (either type 1 or 2) about four years after completing the survey, and looked at survey data for these people.

The researchers noticed a pattern that suggested people who developed diabetes were less likely to have drunk alcohol moderately and frequently compared with those who did not develop diabetes. The researchers reported that the lower risk for diabetes was associated with 14 units per week for men and seven units for women (current recommendations are that men and women shouldn't regularly drink more than 14 units per week).

However, the study had various weaknesses, which means it cannot conclusively show that drinking frequently and moderately protects against diabetes. For example, people were only asked about their drinking habits and other lifestyle choices at a single time point. Also, the study doesn't tell us whether those habits changed over the period in which people were monitored for diabetes.

Even if an association does exist, there are far healthier ways to reduce your diabetes risk, such as achieving or maintaining a healthy weight.

Where did the story come from?

This study was an analysis of data from the general Danish population that had been recorded in a previous cohort study. This particular piece of research was carried out with no specific funding, but the survey data had been funded by the Ministry of the Interior and Health, and the Tryg Foundation. It was published in the peer-reviewed journal Diabetologia.

The suggestion that regularly drinking alcohol may be good for you was met with glee by the UK media. The limitations of the study, or the lack of a definitive cause and effect, were not reported fully.

However, some sources carried sensible advice from independent experts, such as Dr Emily Burns, the head of research communications at Diabetes UK, who was quoted in The Guardian as saying: "While these findings are interesting, we wouldn't recommend people see them as a green light to drink in excess of the existing NHS guidelines, especially as the impact of regular alcohol consumption on the risk of type 2 will be different from one person to the next."

There were several reports that wine was particularly beneficial because it has "a role in helping to manage blood sugar", but this was based only on the authors' comments rather than on the results of the research.

What kind of research was this?

This cohort study assessed people for diabetes in 2012 about four years after their lifestyles had been assessed in 2007-2008. The researchers aimed to examine whether there was any association between alcohol drinking patterns and the risk of developing diabetes in people who did not already have the condition. They looked at the amount that people drank, how often they drank, and what types of alcohol were consumed.

The study benefitted from involving a large number of people in the Danish population, which meant a range of drinking patterns were found, and there were sufficient numbers of cases of diabetes to look for associations.

However, a major weakness of the study was that it only looked at alcohol drinking patterns at a single point in time. And people's drinking habits are known to change over time according to their circumstances, preferences and other health issues.

The researchers did attempt to take into account other confounding factors (such as diet and exercise) that may have influenced the results, but these factors may not have been recorded in enough detail to be useful, and other factors might not have been recorded at all.

What did the research involve?

The researchers identified 70,551 people from the Danish Health Examination Survey (an ongoing nationwide study) who were eligible to participate. These people had already completed a questionnaire in 2007-2008 about their lifestyle and health. People had to meet the following criteria to be selected for participation:

  • no existing diagnosis of diabetes at the start of the study
  • not pregnant and haven't recently given birth (within the last six months)
  • have provided at least some information about their drinking habits in the questionnaire

Information about drinking patterns was collected from questionnaires that people completed by themselves on how often they drank, whether they ever binged and how often this happened, and how much they drank different types of drinks (beers, wines or spirits).

The researchers also looked at information that had been collected at the start of the study on the following confounding factors:

  • age
  • sex
  • body mass index
  • education
  • smoking status
  • diet
  • leisure-time physical activity
  • high blood pressure (current or previous)
  • family history of diabetes

A diagnosis of diabetes was recorded using the Danish National Diabetes Register, which uses five different sources to detect diabetes cases, but does not distinguish between type 1 and type 2. During the course of the study the researchers carried out a "sensitivity analysis", where they excluded two of the diabetes cases because of concerns the data was unreliable.

The participants were followed up in the study until it ended in December 2012, unless they emigrated, died or developed diabetes before then. The researchers carried out an analysis that looked at the risk of developing diabetes over time, taking into account different risk factors. They used appropriate statistical methods for dealing with missing data.

What were the basic results?

During the course of the study, 859 men and 887 women developed diabetes. When looking at the average amount that people drank over the course of a week, they found that the lowest risk of diabetes was observed in:

Frequency of drinking

After adjusting for other factors, the researchers reported the consumption of alcohol on three to four days a week was associated with a lower risk of developing diabetes for men: HR 0.73 (95% CI 0.59 to 0.94) and for women: HR 0.68 (95% CI 0.53 to 0.88).

The researchers also looked at binge drinking and found no clear link between binge drinking and risk of diabetes.

Type of alcohol

Researchers noticed a number of patterns in terms of what types of alcohol people drank.

Men who drank 1-6 glasses of beer a week were found to have lower diabetes risk than those who did not.

In contrast, women who regularly drank spirits seven or more times a week had an increased risk of diabetes compared with those who drank spirits once a week or less. However, the researchers failed to take into account that some people drink a mixture of different types of alcohol either on a single occasion or over a week.

How did the researchers interpret the results?

The authors concluded that "light to moderate" alcohol consumption was associated with a lower risk of diabetes when compared to no alcohol consumption at all. They also noted that frequent consumption was associated with the lowest risk, even after taking into account the amount people drank on average during a week.

They noted that the strengths of their study included its size, the fact that they distinguished between people who currently didn't drink from those who had never drunk at all, and that their results were consistent even when they adjusted various conditions.  


Although this study found an interesting association between alcohol drinking habits and risk of developing diabetes, this study does not present strong enough evidence to recommend adopting a particular drinking pattern to reduce diabetes risk.

This study had a number of limitations that weaken confidence in the results:

  • People were only asked about their drinking habits and other risk factors at a single time point. The study doesn't tell us whether those habits changed over the period in which people were monitored for diabetes. Most studies related to alcohol consumption also run the risk that people are not always completely accurate when describing what and how much they drink.
  • The way diabetes cases were recorded for the study did not distinguish between type 1 and type 2 diabetes, even though these conditions have different causes and treatments.
  • The study only followed people up for an average of just under five years, whereas a condition like diabetes may develop due to risk factors experienced over a longer period.
  • The information collected on diet may have been too simplistic to properly allow an understanding of how nutrition may also affect the diabetes risk of the people in the study.
  • Although the researchers excluded people from the study if they already had a diagnosis of diabetes at baseline, they didn't exclude people if they had other chronic health conditions, some of which may contribute to diabetes risk. The only other condition that was considered in the analysis was high blood pressure.

Overall, it is unclear whether the link between moderate alcohol drinking and diabetes is real. It is not proof that starting to drink more, especially for those who do not currently drink, is useful in preventing diabetes. There are other risks, such as liver damage, to consider when drinking frequent or large volumes of alcohol above recommended limits.

If you are concerned that you might be at risk of developing diabetes, speak to your GP about the ways that lifestyle change can reduce your risk.

NHS Attribution