Wine and chocolate may not 'beat diabetes'

“Chocolate and red wine 'can beat diabetes'," is the misleading and potentially harmful headline on the Sky News website. The study it reports on was actually looking at specific compounds found in wine and chocolate, called flavonoids.

The study found that women with a flavonoid-rich diet appeared to have less biological signs they were heading for type 2 diabetes – specifically lower insulin resistance and lower insulin levels – compared to women consuming lower levels of flavonoids.

However, flavonoids are not just found in wine and chocolate, but are also found in plants, herbs, berries and tea.

The study was a cross sectional design meaning it cannot prove flavonoids reduce the risk of developing diabetes. It could be the case that women with a flavonoid rich-diet tended adopt healthier lifestyle choices, such as exercising regularly, and it was this that was contributing to the lower insulin resistance. Only a well-conducted, double-blinded randomised control trial could prove direct cause and effect.

Also, the study relied on signs of insulin resistance, rather than a diagnosis of diabetes itself. As not all women with these signs would actually develop diabetes in their lifetime, this weakens the reliability of the results.

Teasing apart the effect of one type of chemical on disease risk, when the disease risk can be influenced by a large range of other dietary and non-dietary factors is difficult.

This study does not give the green light to drink red wine above the recommended levels or to consume chocolate often – any potential benefits of diabetes prevention are likely to be overshadowed by the already known risks of excessive sugar, fat and alcohol consumption, including liver disease, cardiovascular diseasestroke and cancer.

Where did the story come from?

The study was carried out by researchers from University of East Anglia and King's College London and was funded by the Department of Nutrition, Norwich Medical School, University of East Anglia, and the Biotechnology and Biological Sciences Research Council.

The study was published in the peer-reviewed medical journal, the Journal of Nutrition.

Generally, most of the UK media’s reporting of the study fell into a similar pattern. The headlines overegged the implications of the findings and failed to report the significant limitations of the research, but the actual body of the reporting was accurate.

What kind of research was this?

This was a cross sectional study looking to see whether chemicals called flavonoids influenced signs of type 2 diabetes in a large group of women including insulin resistance and related inflammatory biomarkers.

The authors said that information from laboratory experiments suggests that several flavonoid subclasses are involved in glucose metabolism – a key part of diabetes. However, there was very little information from studies done on people.

As this was a cross sectional study it cannot prove causation, that is, that flavonoids prevent diabetes.

randomised control trial would be required for this. 

What did the research involve?

Intake of flavonoids (and a range of flavonoid sub-classes) from food and drink were calculated from food frequency questionnaires filled in by a group of 1,997 women aged 18 to 76 who were taking part in the Twins UK registry.

This a national registry of adult twin volunteers recruited from the general population (the benefit of using twins in research is that you can be pretty sure that genetic factors are the same in both, meaning you have one less set of confounders to worry about). 

A number of markers of type 2 diabetes were then measured during a clinical assessment between 1996 and 2000 including: fasting blood glucose, insulin, high sensitivity C-reactive protein, plasminogen activator inhibitor and adiponectin. The main analysis looked for links between the flavonoid levels and the markers associated with type 2 diabetes.

The results were balanced for a range of potentially influential factors, including:

  • age (years)
  • current smoking (yes or no)
  • physical activity (inactive, moderately active, or active)
  • body mass index (BMI)
  • menopausal status (premenopausal or postmenopausal)
  • use of hormone replacement therapy (yes or no)
  • use of diabetes or cholesterol-lowering drugs (yes or no)
  • use of vitamin supplements (yes or no)

Energy intake (kilocalories per day in quintiles) was also assessed, and this was further broken down into:

  • carbohydrate intake (percentage energy in quintiles)
  • wholegrains intake (grams per day in quintiles)
  • intake in terms of unsaturated/saturated fat ratio (quintiles)
  • alcohol intake (grams per day)

The researchers used old data from an existing study. The participants included in the analyses were a small sample of the original total population group of 5,119 women on the registry. A total of 36% (n = 1,857) were excluded for having an incomplete food frequency questionnaire or implausible energy intake, 24% (n = 1,211) did not attend a clinical session for insulin resistance assessment, and 1% (n = 54) had insulin values outside the inclusion criteria for the current analyses. The women analysed included 960 pairs of twins, and 77 isolated single twins.

The data analysis was appropriate.

What were the basic results?

Flavonoid and flavonoid subclass intake

Overall, tea was the main source of total flavonoid (81%), flavan-3-ol (91%), flavonol (63%), and polymer (83%) intake. Four foods contributed >10% of anthocyanin intake (grapes, 20%; pears, 24%; wine, 22%; and berries, 12%) and three foods to >10% of flavone intake (oranges, 27%; wine, 26%; and peppers, 14%).

Association between flavonoid intake (including subclasses) and diabetes markers

In the main analysis, a higher intake of anthocyanins was significantly associated with lower insulin resistance and lower fasting insulin levels. This came from a comparison between women with the highest 20% of flavonoid intake and those with the lowest 20%.

Intake of foods rich in anthocyanins and flavones were both significantly associated with less insulin resistance and lower insulin levels and there appeared to be a dose response relationship.

How did the researchers interpret the results?

The researchers’ main conclusion was that “the findings of the current study provide an insight into the potential mechanisms by which anthocyanins may act to reduce type 2 diabetes risk and are consistent with previous studies investigating intake of specific flavonoid subclasses and type 2 diabetes risk.”

They also mention that “it is plausible that increasing intakes of anthocyanin-rich foods, such as grapes, berries, and wine, would lead to greater improvements in insulin resistance because in vitro [in the laboratory] studies have shown previously that this is dose-dependent relation”.


This study found a link between levels of flavonoids and biomarkers of diabetes suggesting some flavonoid subclasses may have a potential role in lowering the risk of type 2 diabetes.

The strengths of the study include the large sample size and the range of flavonoid subclasses investigated. The food frequency questionnaire used in the study was validated previously and shown to both reflect habitual dietary intake and have the ability to rank participants according to their usual intake of flavonoid-rich foods. However, the questionnaire is still, ultimately, a subjective estimate and is reliant on accurate self-reporting.

The limitations to consider include:

  • As this was a cross sectional study, it cannot prove that flavonoids prevent type two diabetes. A randomised clinical trial would be needed to prove this.
  • Of the 5,119 participants eligible to participate, only 1,997 were analysed in the results, the rest were excluded because they either didn’t fill in the food questionnaire fully, did not attend clinical assessment to assess diabetes biomarkers and other reasons. It is possible this large number of exclusions biased the results.
  • This study did not look at whether flavonoid subclasses were associated with type 2 diabetes directly. Instead it took an indirect approach looking at markers associated with diabetes type 2. Some people with these markers won’t get the disease so this indirect approach is less reliable than finding out whether flavonoids were associated with a diagnosis of diabetes for instance.
  • The study included only women, the results in men may be different.

The study did not highlight chocolate as a big contributor to flavonoid levels in the women’s diet so the media was slightly wayward in their reporting of this. Wine and berries were mentioned as significant contributors for the women in the study.

The bottom line is that this study only highlights a possible link and cannot prove cause and effect. A clinical trial is needed before these results can be believed.

We would have no problem promoting a diet rich in fresh fruits such as berries and oranges. However, care should be taken with tea; excessive amounts of caffeine may trigger symptoms of irritability and insomnia in some people.

As with chocolate and wine, it could be the case that any potential benefit is outweighed by the risk, such as liver disease and obesity.

A proven method of reducing your risk of type 2 diabetes is to maintain a healthy weight, and contrary to media reports, a diet rich in wine and chocolate is not going to help you with that.

NHS Attribution