Alcohol and Cardioprotection: Enduring Truth or Myth That Won't Die?

Christopher Labos, MD, CM, MSc


July 14, 2020

Christopher Labos, MD, CM, MSc

If you're a cardiologist and you've ever been to a wedding, dinner party, or some other setting where alcohol is served, invariably someone will tell you that alcohol, particularly red wine, is good for your heart. In this setting, you have three options: You can agree with them and move on; disagree with them and get into a potentially contentious argument while you try to explain the complexity of inferring causality from nonrandomized data; or you can change the subject to less risky topics like religion or politics. I usually go with the third option.

Most physicians could be forgiven if they have trouble formulating a coherent recommendation about alcohol for patients. The 2015-2020 Dietary Guidelines for Americans states that "if alcohol is consumed, it should be in moderation." But at the same time, they do not recommend that people who abstain from alcohol start drinking for any reason. The American Heart Association has a similar recommendation and states that "moderation is key,” while also warning the public that if they do not drink, they should not start.

But some guidelines are more blunt. The World Health Organization (WHO) calls alcohol a toxic substance that contributes to 3.3 million deaths per year worldwide, citing cancer and other noncardiac risks. Alcohol can also increase blood pressure and in excessive amounts can result in alcoholic cardiomyopathy. Which begs the question: Is it time for guideline writers to reassess their tentative approval of "moderate" drinking.

What Is Moderate Drinking, Really?

One fundamental problem is that there is no clear consensus on what "moderate” drinking means. Definitions and guidelines can vary widely between countries. In the United States and Canada, one bottle of wine (12% alcohol, 750 mL or 25 oz) contains five servings of a standard drink, whereas in the United Kingdom it has between nine and 10 units of alcohol. Given that most of us split a bottle of wine between two or three people, it seems likely that we have trouble accurately gauging a standard drink, and consequently may underestimate our alcohol consumption and routinely exceed the guideline recommendations of up to two drinks per day (196 g of alcohol).

Adding to the complexity is a recent meta-analysis published in The Lancet suggesting that even these thresholds might be too high. This analysis of individual-participant data from three large cohorts included nearly 600,000 current drinkers and found that the risk of all-cause mortality began to increase after approximately 100 g of alcohol per week — nearly half the current guideline-recommended threshold. In fact, the researchers estimated that reducing alcohol consumption from 196 g/wk to 100 g/wk would result in 1-2 years of increased life expectancy at age 40.

In this meta-analysis, there was a roughly linear association between increasing alcohol consumption and increased risk for stroke, coronary disease, heart failure, fatal hypertensive disease, and aortic aneurysms. The lowest risk was seen in those who consumed the least alcohol (0-100 g/wk). Consequently, the authors concluded that their data support the "adoption of lower limits of alcohol consumption than are recommended in most current guidelines."

However, two caveats might suggest a mitigating factor. First, despite the linear association with many negative cardiovascular endpoints, alcohol consumption was inversely associated with myocardial infarction (MI). Second, the association between alcohol consumption and all-cause mortality was stronger in drinkers of beer and spirits rather than wine.

The French Paradox

The late Morley Safer cemented the potential cardioprotective effects of wine into the public consciousness in a 1991 episode of 60 Minutes. Since then, multiple studies have suggested the same and have led to the widespread conviction that some alcohol is indeed good for you.

The presumed mechanism of this benefit has generally centered around the presence of antioxidants in wine. Resveratrol was commonly put forth as the likely candidate, but the Aging in the Chianti Region study found that resveratrol levels in a community-dwelling cohort of Italian men and women over age 65 were not associated with any cardiovascular, cancer, or all-cause mortality benefit.

Nevertheless, red wine contains multiple molecules with antioxidant properties and potential cardiovascular benefits. Data from the PREDIMED study suggest that polyphenols in red wine might be the mechanism for its positive benefit. In a secondary observational analysis of the main PREDIMED trial (admittedly done prior to the discovery of concerns surrounding the study's randomization), patients with the highest quintile of flavonol intake (mainly from red wine and apples) had improved cardiovascular outcomes compared with those in the lowest quintile. PREDIMED principal investigator Ramon Estruch, MD, has suggested that 20% of the effects of the Mediterranean diet are due to a moderate intake of wine.

Other patient cohorts seem to offer up similar suggestions. The Seguimiento Universidad de Navarra (SUN) project analyzed the drinking habits of a cohort of Spanish university graduates. Adherence to a Mediterranean alcohol-drinking pattern (ie, moderate alcohol consumption, avoiding binge drinking, and a preference for red wine over spirits) was suggestively but not statistically or definitively associated with lower cardiovascular mortality.

The MACH Study That Wasn't and Mendelian Randomization

While observational research provides some interesting insights, clearly only a randomized trial could settle the issue of alcohol's benefit or lack thereof. The National Institutes of Health (NIH)-funded Moderate Alcohol and Cardiovascular Health (MACH) trial was supposed to be that trial when it began randomly assigning 7800 middle-aged men and women to zero or one drink per day. However, the NIH pulled the plug after reports surfaced that the alcohol industry was partially funding the study. There were also issues surrounding trial design. An NIH report was critical that heart failure was excluded from the composite cardiovascular endpoint, calling it a "serious shortcoming."

Without a traditional randomized study to settle the issue of alcohol's beneficial cardiovascular effects, genetic studies might offer up the next best thing. Mendelian randomization studies, sometimes referred to as nature's randomized trials, exploit a particularly useful quirk of genetic inheritance.

At conception, an embryo is equally likely to receive either allele A or allele B of a gene, essentially mimicking the randomization seen in clinical trials. As long as the gene in question doesn't have any pleiotropic effects (ie, it doesn't also affect an important confounder like blood pressure, diabetes risk, or propensity to smoke), people with allele A or allele B should be similar in every other respect. Therefore, any difference in outcomes between the two groups can be attributed to their genetic difference.

Researchers from Oxford University used just such a technique in a large Chinese cohort. They examined individuals with a genetic variant in the aldehyde dehydrogenase gene, ALDH2. Individuals with a loss-of-function mutation in that gene metabolize acetaldehyde less quickly, develop symptoms like flushing when they drink alcohol, and drink less alcohol on average.

Another advantage in this Chinese cohort was that women drank very little alcohol, with only 2% of women reporting some alcohol consumption in most weeks compared with 33% of men. This difference provided researchers with a natural control group. Any effects that were seen in men but not women who had the same genetic variants could be attributed to the alcohol rather than some other pleiotropic effect of the genetic variant. Finally, they performed one last control check in their analysis: They presented the results not only of their genetic analysis but also of a conventional epidemiologic analysis that focused only on alcohol consumption patterns.

The conventional epidemiologic analysis showed a U-shaped association for alcohol consumption, with moderate drinking (100 g of alcohol per week) being associated with a lower risk for stroke, intracerebral hemorrhage, and MI. But the genetic analysis showed no such association. It demonstrated a dose-response relationship where higher alcohol consumption was tied to increased risks for both ischemic and hemorrhagic stroke.


"We got quite clear results for stroke," Iona Millwood of the Nuffield Department of Population Health at the University of Oxford, and joint first author of the study, told me via Skype. "We could demonstrate clearly that there was no protective effect on moderate intake for stroke." However, the results for MIs were more equivocal. Because the number of MIs was limited compared with the number of strokes in the population, the study may have been underpowered to detect a clear effect.

Another limitation is that most people in this cohort consumed spirits rather than red wine. But Millwood points out that their study, just like prior studies measuring red wine consumption, showed a J-shaped association where moderate consumption was protective relative to no consumption. Or at least it did until they did the genetic analysis.

"Our study shows quite nicely that for stroke, no matter what you do with these conventional epidemiology studies, it's very hard to remove those J-shaped curves that you see," explains Millwood. "We tried adjusting for many factors, we excluded ex-drinkers from our nondrinking group, we did a series of sensitivity analyses to try to remove early follow-up where we might have seen the effects of reverse causation, and we still couldn't get rid of the J-shaped curve when doing this conventional study."

But when they ran their genetic analysis, the J-shaped curve disappeared and the beneficial effect was gone. Despite their best efforts, residual confounding was still skewing the results of the conventional analysis, and they would have gotten the wrong result if they had not used their genetic analysis.

"We saw such a different pattern of association in our genetic study, and it shows that sometimes you really do need to try different methods in your research," says Millwood.

It may not be easy to dislodge the popular idea that moderate alcohol consumption is good for the heart. But given the WHO assessment that alcohol is the leading risk factor for premature death among 15- to 49-year-olds, and the mounting evidence that the protective effect may be the result of residual confounding, maybe we should try a little harder.

Christopher Labos is a cardiologist with a degree in epidemiology. He spends most of his time doing things that he doesn't get paid for, like research, teaching, podcasting, and writing for the newspaper. Occasionally he finds time to practice as a cardiologist so that he can pay his rent. He realizes that half of his research findings will be disproved in 5 years; he just doesn't know which half. He is a regular contributor to the Montreal Gazette, CJAD radio, and CTV television in Montreal. To date, no one has recognized him on the street or asked for his autograph.

Follow | Medscape Cardiology on Twitter

Follow Medscape on Facebook, Twitter, Instagram, and YouTube


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.