Humans have been drinking fermented concoctions since the beginning of recorded time. But despite that long relationship with alcohol, we still don’t know what exactly the molecule does to our brains to create a feeling of intoxication. Likewise, though the health harms of heavy drinking are fairly obvious, scientists have struggled to identify what negative impacts lesser volumes may lead to. Last September, the prestigious peer-reviewed British medical journal The Lancet published a study that is thought to be the most comprehensive global analysis of the risks of alcohol consumption. Its conclusion, which the media widely reported, sounded unequivocal: “The safest level of drinking is none.”
Sorting through the latest research on how to optimize your well-being is a constant and confounding feature of modern life. A scientific study becomes a press release becomes a news alert, shedding context at each stage. Often, it’s a steady stream of resulting headlines that seem to contradict one another, which makes it easy to justify ignoring them. “There’s so much information on chocolate, coffee, alcohol,” says Nicholas Steneck, a former consultant to the Office of Research Integrity for the U.S. Department of Health and Human Services. “You basically believe what you want to believe unless people are dropping dead all over the place.”
Scientific studies are written primarily for other scientists. But to make informed decisions, members of the general public have to engage with them, too. Does our current method of doing so — study by study, conclusion by conclusion — make us more informed as readers or simply more mistrustful? As Steneck asks: “If we turn our back on all research results, how do we make decisions? How do you know what research to trust?” It’s a question this new monthly column aims to explore: What can, and can’t, studies tell us when it comes to our health?
The truth is, putting alcohol research in context is tricky even for scientists. The Lancet study is epidemiological, which means it looks for patterns in data related to the health of entire populations. That data might come from surveys or public records that describe how people behave in their everyday environments, settings that scientists cannot absolutely control. Epidemiological studies are a crucial means of discovering possible relationships between variables and how they change over time. (Hippocrates founded the field when he posited an environmental rather than a supernatural cause for malaria, which, he noted, occurred most often in swampy areas.) They can include millions of people, far more than could be entered into a randomized-control trial. And they are an ethical way to study risky behaviors: You can’t experiment by randomly assigning groups of people to drive drunk or sober for a year. But because epidemiologists can only observe — not control — the conditions in which their subjects behave, there are also a vast and an unknown number of variables acting on those subjects, which means such studies can’t say for certain that one variable causes another.
CreditIllustration by Celia Jacobs
Modern epidemiology took off in the 1950s and ’60s, when public-health researchers in the United States and Britain began long-term studies tracking a wide variety of health factors in thousands of people over decades and surveying them about their behavior to try to identify risks. What they found when they looked at alcohol consumption in particular was puzzling: People who reported being moderate drinkers tended to have a lower risk of mortality and many specific health problems than abstainers did. Did this mean that a certain amount of alcohol offered a “protective” effect? And if so, how much? In 1992, an influential study in The Lancet observed that the French had a much lower risk of death from coronary heart disease than people in other developed countries, even though they all consumed high levels of saturated fat. The reason, the authors proposed, was partly that the French drank significantly more wine.
The notion that alcohol may improve heart health has persisted ever since, even as further research has revealed that it can cause cancer and other health problems and increase the risk of injury and death. But equally plausible counterhypotheses also emerged to explain why teetotalers fared worse than moderate drinkers. For instance, people might abstain from alcohol because they are already in poor health, and most studies can’t distinguish between people who have never had a drink and those who drank heavily earlier in their lives and then quit. Indeed, over the years, compared with abstinence, moderate drinking has been associated with conditions it couldn’t logically protect against: a lower risk of deafness, hip fractures, the common cold and even alcoholic liver cirrhosis. All of which advances a conclusion that health determines drinking rather than the other way around. If that’s the case, and abstainers are predisposed toward ill health, then comparing drinkers to them will underestimate any negative effects that alcohol has. “This problem of the reference group in alcohol epidemiology affects everything,” says Tim Stockwell, director of the Canadian Institute for Substance Use Research at the University of Victoria in British Columbia. “It’s urgent to establish, What is the comparison point? All we know is that risk goes up the more you drink for all of these conditions.” But without a reliable comparison group, it is impossible to say precisely how dire those risks are.
The authors of the recent study in The Lancet endeavored to address this problem, at least in part, by removing former drinkers from their reference group, leaving only never-drinkers. To do so, they spent two years searching for every epidemiological study of alcohol ever done that met certain criteria and then extracting the original data. They marked those that already excluded former drinkers, which they thought would make the comparison group more accurate; to those that didn’t, they applied a mathematical model that would control for differences between their comparison group and that of the preferred studies.
The results — which are broken down by age, sex, 195 geographical locations and 23 health problems previously associated with alcohol — show that over all, compared with having zero drinks per day, having one drink per day increases the risk of developing most of those health problems. They include infections like tuberculosis, chronic conditions like diabetes, eight kinds of cancer, accidents and self-harm. (The more you drank, the higher those risks became.) This suggests that, on the whole, the benefits of abstaining actually outweigh the loss of any health improvements moderate drinking has to offer. The results, however, also show that a serving of alcohol every day slightly lowers the risk of certain types of heart disease — especially in developed countries, where people are much more likely to live long enough to get it. So, theoretically, if you are a daily drinker who survives the increased risk of accidents or cancers that are more likely to strike young to middle-aged people, by 80, when heart disease becomes a major cause of death, your moderate drinking could prolong your life. Then again, it might be your innate biological resilience that kept you healthy enough to drink. The data still can’t say.
Keep in mind that population studies like these are not meant to directly change individual behavior. They offer generalizations — in the case of the Lancet study, that alcohol consumption is probably riskier and less potentially beneficial than we thought — that may eventually influence policies, like higher taxes on alcohol or warning labels on bottles. Paradoxically, only if those policies, in turn, reduce the amount that millions of individuals drink will we know whether doing so improved their overall health.
In the immediate term, a better way of understanding the value of scientific studies might be to think of each as a slight adjustment of an eyeglass-lens prescription. Each one answers the question “Is it clearer like this, or like this?” and in doing so, brings our view of reality — our understanding of ourselves and the world around us — into sharper focus. If we dwell too much on the conclusions studies seem to offer, instead of also considering how they were reached, we risk missing out on one of the great benefits of the scientific process: its ability to reveal all that we don’t know.