Advising Under the Influence: Alcohol Guidance Demands Full Disclosure

Harpreet S. Bajaj, MD, MPH


February 15, 2018

One of my patients who has prediabetes approached me last week: "I read this news story about two new studies that found that drinking wine can reduce my risk of developing diabetes. Is that true? Should I start drinking?"

Before I could even begin my educational spiel on the limitations of such research, he asked me another question: "Do you drink, doc?"

Harpreet S. Bajaj, MD, MPH

My patient was referring to two widely publicized studies published recently.[1,2] The first was a Danish study that looked at alcohol drinking patterns and diabetes risk, based on a 5-year follow-up of 76,000 individuals.[2] The second was a French study that examined the link between type 2 diabetes and a diet rich in antioxidants, including wine, based on a cohort of 64,000 middle-aged women who were followed for 15 years.

The French study, like most previous observational studies on alcohol-related risk, suffers from multiple confounding issues and biases. For one thing, it used just one baseline dietary assessment. Hence, any previous history of antioxidant intake—especially in regard to alcoholic beverages, where people are known to fluctuate in terms of frequency, quantity, and choice of drink—could have been missed (eg, former drinkers who quit could have been classified as non-drinkers).

In the Danish cohort study, the authors tried to overcome a similar baseline bias by analyzing lifelong abstainers separately, but a misclassification related to change in alcohol habits during the 5-year follow-up still could have skewed the results. In addition, information on food and antioxidant intake was limited, and confounding from diet-related variables was unmeasured and could have affected the study conclusions.

In both instances, researchers failed to adequately adjust for socioeconomic status, which is known to be associated with both alcohol intake and preference, as well as independently with the incidence of diabetes.

Furthermore, age stratification as an effect modifier was not evaluated in either study; note that the effect of alcohol on so-called cardioprotection is thought to fade away with graying of hair.[3] Finally, one of the pillars in the famous Bradford Hill criteria for determining causality[4]—biological plausibility—is questionable (especially given that each 150 mL of wine contains around 120 kcal, ie, comparable to a can of soda, so you'd have to balance the benefit, if any, from antioxidant content against the caloric content).

To get a clear answer to a question, we always like to have a randomized controlled trial, but in this case we'd have to ask:

  • Is it feasible to randomly assign patients to wine drinking versus abstinence for a prolonged period to investigate health associations?

  • Is it ethical to randomly assign anyone to alcohol, given the overwhelming literature that alcohol causes liver disease and pancreatitis, and may account for an estimated 5.5% of all cancers worldwide annually?[5]

We don't really need to answer those questions, because at least one randomized trial has already been done.[6] A 2-year study conducted in Israel and published in 2015 found that wine improved cardiometabolic risk in people with type 2 diabetes, with a statistically significant increase in HDL cholesterol for the wine-assigned group. But the small magnitude of HDL rise at the end of 2 years has questionable clinical significance.

There was a variable effect of the alcohol on glucose levels in this trial, depending on the type of wine and the rate of metabolization in an individual patient, so when you take this into account too, these results are insufficient to endorse wine as a uniformly healthy option in people with diabetes.

Are We Advising Under the Influence?

Although some clinical guidelines recommend public health interventions to reduce population-level consumption of alcohol, most diabetes and cardiology societies continue to perpetuate the age-old "drink with moderation" pseudo-endorsement.

In an era when alcohol consumption recommendations seem dependent on a wide range of variables, I can't help but wonder whether guideline recommendations should be harmonized to focus on overall health effects of alcohol, rather than targeting available evidence on each organ system separately.

And do the guideline differences reflect clinical experiences of the respective authors, or perhaps their own personal alcohol-related prejudices?

The lay media further polarizes any harm-benefit analyses about alcohol intake by highlighting news stories that cater to their prime audiences' preconceived notions and habits. Most media outlets covering the French study, for example, highlighted just the alcohol-related piece rather than discussing the other dietary sources of antioxidants that were evaluated in the same study.

What's the reason for this biased reporting? I doubt that this is a conspiracy hatched by the alcohol industry to increase their business, although this cannot be ruled out completely.

The bias probably lies within each of us. The public's perception, similar to that of healthcare providers, is filtered through the prism of our own alcohol-related behaviors.

For instance, there is widespread denial (and hence only short-lived news coverage) about the alcohol-cancer link recently touted by the American Society of Clinical Oncology,[5] probably because we don't want to give up our wine, beer, vodka, or whatever that favorite drink may be.

We'd rather believe the older studies which found that red wine reduces heart disease, or the latest ones which suggest a reduced risk for diabetes, without acknowledging the skewing effects of confounders and biases in observational research.

So, What Should I Advise My Clinic Patient?

Now that I have thought this through, I have come to this conclusion: In addition to acknowledging pharmaceutical conflicts of interests, researchers and guideline authors (and medical providers like us) should publicly declare our own drinking habits, along with a clear statement of whether our ethanol proclivity influences our alcohol prescription for patients.

This is my prepared spiel for patients moving forward:

I occasionally drink wine or beer. I try to limit the quantity because of caloric content. Importantly, this does not cloud my assessment of the published research, which, in my opinion, is not strong enough to make any conclusive judgement on the effects of alcohol on diabetes, heart disease, or mortality.

On the other hand, the evidence documenting the harms caused by alcohol in terms of liver disease, pancreatitis, gout, mood disorders, and certain cancers is much stronger.

So, here's my advice: If you're a teetotaler, it may be best for your health to remain that way. If you drink, do it for pleasure (and try to minimize both the quantity and frequency of consumption) rather than with the assumption that it may provide health benefits.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.