Alcohol and your health: Is none better than a little?

Robert H. Shmerling, MD

Senior Faculty Editor, Harvard Health Publishing

It’s complicated.

That’s the best way to describe the relationship between alcohol and health. As I’ve written about before, a number of studies have demonstrated health benefits with lower amounts of drinking. But if you drink too much alcohol (especially at inopportune times), there may be significant harms as well. Just how these balance out remains a matter of some debate and controversy.

While it’s easy to say “too much alcohol is bad for you” (and then point out the litany of harms caused by alcohol, such as liver disease and motor vehicle accidents), it’s harder to answer these simple but important questions:

  • Just how much is too much?
  • Is there a health benefit to some drinking compared with none?

These are more than just interesting questions for researchers to study. The answers could guide recommendations of doctors, public health officials, and policy makers throughout the world — and they could save millions of lives.

But so far, the answers vary depending on the study. And perhaps that should not be too surprising since study methods differ widely. For example, the definition of “one drink” in the US is 14 grams of alcohol, as found in a 12-ounce bottle of beer, 5-ounce glass of wine, or 1.5-ounce shot glass of distilled spirits. In other countries, and in many research studies, a different definition is used.

Recent studies on alcohol and health

In June of 2018, a study published in the journal PLOS Medicine found that among older adults, light drinking (in the range of one to four drinks per week) was associated with a slightly lower risk of death compared with zero consumption.

In August of 2018, two larger studies examined the impact of alcohol. The first one, published in The Lancet, included only people who drank at least some alcohol. It concluded that common recommendations regarding “moderate” drinking (one drink a day or less for women, and two drinks per day or less for men) might be too much.

The second study, also published in The Lancet, was even bigger. It examined data from hundreds of studies and other sources (including sales of alcohol, home-brewed alcoholic beverage consumption, and even estimates of tourist consumption) in 195 locations. And it analyzed the overall health impact related to alcohol consumption, including death and disability due to automobile accidents, infectious diseases, cancer, and cardiovascular disease. It concluded that the best option for overall health was no drinking at all. Of note, the definition of “a drink” in this study was 10 grams of alcohol — that’s 30% less than a standard drink in the US, but 25% more than a standard drink in the UK.

Here are more details about what they found:

  • Alcohol use was the seventh leading cause of death and disability worldwide in 2016; about 2% of female deaths and 7% of male deaths (2.8 million deaths in total) were considered alcohol-related.
  • For those ages 15 to 49, alcohol was the leading risk factor for death and disability worldwide. Tuberculosis, road injuries, and self-harm were the top causes (the risk of each of these conditions is higher if you drink enough).
  • For older adults, cancers related to alcohol use were the top causes of death.
  • In general, health risks rose with rising amounts of alcohol use. However, some protective effect related to light drinking (less than one drink/day) was observed for heart disease and diabetes in some groups. For example, the risk of heart attack and related cardiovascular disease was 14% lower for men drinking 0.8 drinks/day, and 18% lower for women drinking 0.9 drinks/day compared with none.

From this, the study’s authors concluded that while light drinking might have a modest protective effect for certain conditions among certain people, “Our results show that the safest level of drinking is none.”

Is there another way to look at this?

As I look at the study data, I interpret it differently. True, the data does not confirm a protective effect of light drinking. But the health risks were low, and quite similar at levels between zero to one drink per day. That suggests that zero consumption may not necessarily be best, or any better than several drinks per week. In addition, this study (and others like it) is based on a large number of people, which is helpful to detect trends but can overlook important individual factors. In other words, some people may be harmed or helped more by alcohol consumption than others.

Is zero alcohol a realistic option?

It’s worth acknowledging that regardless of how you interpret this study or whether researchers encourage “zero alcohol” as the best health option, the findings are quite unlikely to lead to zero alcohol consumption. After all, many people are more than willing to accept some health risks associated with drinking because they like to drink! Alcohol can encourage social interaction (which is why it’s often called a “social lubricant”), it is part of many religious traditions, and it’s a source of regular enjoyment for millions of people. And the fact is, most people “get away” with moderate drinking without suffering any major health consequences. As one expert said when interviewed about this study: “There is no safe level of driving, but governments do not recommend that people avoid driving.”

Should you stop drinking?

My take on these new studies is this: if you don’t like to drink alcohol, this latest research gives you no “medicinal” reason to start. But, if you drink lightly (and responsibly) and you have no health problems related to it, this study and other recent research is reassuring.

Clearly there are good reasons to discourage excessive alcohol consumption, driving drunk, and other avoidable alcohol-related trouble. But is “zero consumption” really where we should be aiming? I’m not so sure. I think it’s much more complicated than that.

If your interpretation of this research is different, let me know!

Follow me on Twitter @RobShmerling


  1. Dorothy

    How does one have .09 drink a day? Is that wine or brandy, scotch or vodka?

  2. Stephen C Row

    Robert – You know me as you lived right above me in Winthrop. You also know me as a member of the championship crew (1974) and two-time Olympic camp draftee. What you don’t know is I lived through a horrific childhood with an alcoholic mother who was emotionally and physically abusive to all of us. My sisters probably had the worse end, and both died in the past ten years from complications resulting from alcohol use, abuse and pills. My mom died from alcoholism 33 years ago. My childhood was one long saga of DT’s, hallucinations, convulsions, being driven around in the back seat of a car while mom went bar hopping with a drinking buddy. I never talked about this at Harvard and to be quite honest, had a few good years when my mother was in recovery, albeit shaky and littered with serious relapses. She was tremendously supportive of my Harvard education and though I wanted to go to Stanford, where I was also accepted. she encouraged me to go to Harvard and row on the crew team. I stopped drinking 33 years ago and believe I have the “gene” for alcohol addiction. I don’t drink at all and despite the possible health benefits of an occasional glass of red wine. have no desire to experiment. I watched one Christmas as mom drank a glass of red wine and said she could drink like the rest of us. The result was a horrific Christmas which ended with my brother and dad taking her to detox. I spent Xmas eve at a friends and came home to a house filled with shattered glass, the Christmas tree in the front yard, and a family in shock.

  3. Robert Shmerling

    I agree with the comments/replies here – yes, “association is not causation” – that’s a valid concern but to get a more definitive answer, you’d need a long-term, controlled trial with a large number of people randomized to different amounts of alcohol intake while keeping as many other variables as possible constant – that’s not a trial happening anytime soon! And, as mentioned, many of the outcomes (drunk-driving deaths) are likely causal.

    As for confounders, no, they did not control for diet, exercise and a myriad of other potentially relevant factors. It’s certainly possible that some of the findings are tainted by confounding (e.g., the heaviest drinkers probably smoke more) but the size of the study and the magnitude of the “dose effect” between alcohol and health outcomes argues that alcohol consumption matters.

  4. Manuel Pena

    I think Dr. Shmerling’s article is quite sensible, particularly his view that when it comes to alcohol, there probably isn’t a “one-size-fits- all” conclusion. I particularly wonder whether the large studies took into account possible confounders that might enter into the good-bad discussion. For example, in determining death rates among drinkers and teetotalers, did the researchers take regular exercise into account? Diet? Sleep patterns? Differences in individuals’ stress levels? It seems to me that we cannot isolate alcohol consumption from all of these other variables in determining what’s healthful and what’s harmful.

  5. Ian Graham

    @AlanLeger – could it be that the majority of teetotalers do not drink for religious reasons? and are likely to be no more or less health-conscious than the rest of the population?
    Statistics by association are questionable, but there is nothing to suggest these are simply statistics by association – the links between alcohol and deaths in these studies are causative. The outcomes and findings of well designed studies are more likely to be valid than not; and alternative explanations of the outcomes based on no evidence at all are very much less credible.

  6. Alain Leger

    Could it be that complete teetotallers are likely to be more health
    conscious than other people, and therefore more likely to outlive even moderate drinkers ?
    Statistics by association are always questionable

  7. AF Leger

    Could it be that complete teetotallers are likely to be more health
    conscious than other people, and therefore more likely to outlive even moderate drinkers ?
    Statistics by association are always questionable

Commenting has been closed for this post.