No, One or Two Alcoholic Drinks a Day Is Not Unsafe or Unhealthy


Here we go again. A new round of news headlines implies any level of alcohol consumption is bad for you, based on the findings of a single study that contradicts decades of research. “No amount of alcohol is safe, health experts warn” as a CNBC headline put it, with others like CNN, CNBC, The Guardian, The Daily Beast, and many more repeating some variation on that theme. The problem, once again, is that the study on which these headlines are based found no such thing.

Published last week in the influential scientific journal The Lancet, the research calculated the risk of various levels of drinking on twenty-three different diseases and conditions and found that the benefits of moderate alcohol consumption were outweighed by the risks when considering the health of a nation as a whole. It did not examine the risk drinking poses to individuals and it certainly did not find that drinking was “unsafe.”

Bigger isn’t Always Better

When it comes to research most of us think that bigger is better, but that is not necessarily the case. Without going into the scientific weeds, when researchers have a ton of data they can find reliable evidence that two things (e.g. alcohol and depression) are somehow linked, but rarely how they are linked. For example, a study like that in the Lancet might find that eating a slice of cake once a month is associated with a higher risk of Type II Diabetes versus no cake, but that increased risk might be so small that it is functionally meaningless.

Diseases have Many Factors

Another big issue with the Lancet study is its inability to separate the effect alcohol might have on any of these 23 conditions from all the other factors that influence their development. This is because all the data was plucked from nearly 600 studies that other researchers around the world completed since 1990. Called a systematic analysis, such studies allow modern research access to huge amounts of data, but only the data collected by those conducting the original studies; as a result, authors of systematic analysis have a limited ability to account for the many factors that might impact their results. Sticking with the diabetes hypothetical, if a study from 2003 collected information on both peoples’ alcohol intake and the rate at which they developed diabetes, but failed to collect information about their level of education (which is independently linked both to diabetes and separately to increased risk of alcohol abuse) the authors of a systematic review in a 2018 study would have no way to tell if the link found in the 2003 study might be caused by alcohol intake or some other, unrelated, factor like education.

This is just one possible confounding factor for one particular disease. Diabetes may be impacted by weight, age, family history, physical activity, smoking, level of education, and socio-economic status, among other things, and most diseases have similarly long lists of contributing factors. For the 23 diseases studied by the Lancet authors, the only confounding factors they could account for were location, age, and sex. As such, it is impossible and—more importantly—inappropriate to conclude based on their study that alcohol causes any of the increased risk the purportedly found.

Increased Risk ≠ Danger

Even if we ignore these and the many other methodological issues with the Lancet study, it still only demonstrates that alcohol intake is associated with a slight increased risk of disease-development within a population. Specifically, one drink a day raises the risk of developing one of the 23 conditions by 0.5% compared to zero drinks a day. In context, this means if 100,000 people consumed zero alcohol a day, an expected 914 of them would develop one of the 23 conditions in a given year. If that same group drank one alcoholic beverage a day, an expected 918—four more people—would develop one of the 23 conditions in a given year. Four. With two drinks a day, the risk rises to a more significant seven percent, but that still leads to a very small 63 additional cases of disease per 100,000 people compared to no alcohol at all. 

At the highest alcohol intake—five drinks a day—the researches still found only a 37 percent rise in disease risk. That is the exact same increase in risk that a study published in June found between high-stress jobs and atrial fibrillation. That is not an insignificant increase, but still we would never call the choice to pursue a high-pressure profession a “dangerous” decision and call on government to recommend people avoid high-stress work. That is because, in this case at least, we implicitly understand that just because a choice increases a certain risk, it doesn’t mean that choice is unsafe.

The distinction, which both the Lancet authors and many reporters failed to make, is that danger, like chugging rat poison or swan-diving off of a 12-story building, is universal. Risk, on the other hand, is profoundly personal.

We make choices every day that alter our exposure to risk; how we travel to work, whether we drink that second (or fourth) cup of coffee, whether we stay up late or go to sleep early. All of these choices change our risks up or down and between types of risk. We can never find a choice that is free of risk entirely, including abstaining from alcohol (which increases the risk of cardiovascular disease). Instead we can only choose which risks we’re willing to take: a decision based entirely on individual circumstances and preferences (e.g. family history, values, and risk-tolerance).

The study in the Lancet is not concerned with individual risk, despite the headlines it generated. As lead author Emmanuela Gakidou told Time “we don’t have the information for specific individuals…[but] if you’re running a health system in a country, it’s better overall for the population of your country to not drink at all than to drink a little bit.” Put in different terms, Gakidou’s opinion is that zero alcohol consumption is better for the overall health statistics of a nation even if it’s worse for individuals.

The authors do not put it as bluntly, probably because that’s a poor way to achieve their desired end of dispelling the “myth” (established by decades of research) that moderate drinking can benefit health.

In fact, this paper reads less like a scientific inquiry than a coalition letter with the astounding 506 collaborators taking up the cause célèbre among public health advocates these days—the idea that no level of alcohol is safe—and then designing a study that would come to that conclusion in order to recommend governments should advise against any alcohol consumption. Or, as the study more diplomatically put it: 

Our results point to a need to revisit alcohol control policies and health programmes, and to consider recommendations for abstention.

Whatever its quality, it should be obvious that this study is not sufficient to justify that recommendation nor many of the conclusions to which the authors came. Unfortunately, leaping from study results to unsupported conclusions and policy prescriptions is fairly common in health research these days, perhaps because they know that news media will rely only on these easily-digestible sound bites and incredulously parrot press release and opinion piece. While that did happen at some major news outlets, it wasn’t as widespread as it was with the Lancet’s earlier alcohol study. But it is a sad day when the quality of health journalism has sunk so low that a handful of outlets not grossly misrepresent science seems like a reason to cheer.