There's One Thing We Love to Argue About When It Comes to Alcohol
Sign up for the Slatest to get the most insightful analysis, criticism, and advice out there, delivered to your inbox daily.
The debate about alcohol and health has been going on for decades. Everyone knows that drinking a lot of booze is bad for you—or they should!—and researchers know that pretty much any regular intake of alcohol harms your liver long term. That said, it turns out that the people who are healthiest do tend to drink, though sparingly. This has led to endless arguments over whether a little bit of alcohol is good for your health or whether healthy people's tendency to raise a glass from time to time is more of a statistical artifact.
The latest broadside in this interminable war: It seems, according to headlines, that champagne is good for your health . It can prevent sudden cardiac arrest , a condition in which a person has a heart attack without any warning. (It's often caused by underlying heart conditions, but it can happen entirely out of the blue.)
For people like me who work in epidemiology, or have a good understanding of confounding , this is already pretty funny. I laughed a bit when I saw the news stories. Let me explain.
The study that those articles are based on is a fairly standard epidemiological analysis. The authors took a massive dataset of people from the British Biobank study, which includes an array of data such as genetics, eating habits, and health outcomes. They used this dataset to look at 125 separate risk factors and assess how they were associated with sudden cardiac arrest, correcting for age, sex, race, and place of recruitment in their statistical model. (We'll come back to this in a bit.)
Of these 125 risk factors, 56 had some association with sudden cardiac arrest. They were all fairly commonplace findings: Smoking is bad, exercising is good, eating more fruit and vegetables is good, etc. The authors then did something less common, called a Mendelian randomization analysis. For this process, scientists look at genes that are associated with certain behaviors—ie, genes that may predispose people to, say, have a sweet tooth—and use those genetic associations to see if things are likely causally related rather than associated. It's one thing to ask participants how much sugar they consume, but if everyone with the sweet-tooth gene ends up having sudden cardiac arrest, well, that's a better indicator that there is truly a link.
The final analysis showed that there were nine risk factors that might have a causal relationship with sudden cardiac arrest, including intake of champagne and white wine. In this case, genes associated with an affinity for these beverages seemed to be linked to reduced risk.
There are a few problems here—first, Mendelian randomization analysis. It uses what are called single-nucleotide polymorphisms, which are specific variations in a gene that can affect how your body functions. They work really well in situations where you've got a simple genetic association—like a single SNP that increases or lowers blood cholesterol if it's turned on or off—but are much less useful when you have lots of SNPs that are associated with a single outcome. In this paper, the authors used up to 524 SNPs for one outcome, which requires a massive load of assumptions and makes it harder to trust that the results demonstrate a causal impact. The biggest assumption is that there is no way the SNPs are associated with the outcome—here, heart attacks—other than through the mechanism of interest.
In the case of champagne, there are six SNPs related to how much of the stuff you might tend to drink. According to the authors' logic, the only way those SNPs modify heart attack risk is through the medium of champagne. This is incredibly unlikely, because the main thing that changes your champagne intake—aside from your cash on hand—are taste preferences, and genes that modify taste have numerous effects on health. Basically: It's not that you have “champagne-drinking genes.” You have genes that help determine your general preferences when it comes to food and beverages, which could, in addition to driving other eating and drinking habits, make you more likely to reach for a glass of champagne. But the idea that you can draw a straight line from genes to champagne intake to heart attacks is laughable.
The other problem is what made me laugh when I first saw the headlines: simple confounding. The easiest way to explain confusing is through a famous example: ice cream and drownings. When more people eat ice cream, drownings go up. These two events are strongly correlated. But ice cream isn't causing the drownings; it's just that people eat more ice cream when the weather is hot, which is also when they tend to go swimming more often. More swimming means more drownings, and so the amount of ice cream being consumed and the amount of drownings that occur always appear hand in hand. The warm weather is the underlying factor—the confounding variable—that causes both the exposure and the outcome.
The same idea holds for many other things in epidemiological research. In this case, it's pretty obvious what has happened. Although the authors corrected for a few things in their statistical analysis, there was no good correction for wealth or income. Rich people are more likely to drink champagne—a relatively expensive treat—and they are also more likely to have better health in all sorts of ways. The authors didn't correct for this problem, and even the Mendelian randomization doesn't fully solve the issue. Saying that champagne is preventing sudden cardiac arrest is a bit like arguing that buying a Ferrari makes you less likely to default on your mortgage.
The study itself is interesting in an epidemiological sense. It gives us some useful information about associations that people may want to look into in further study. To be brutally honest, we already knew most of these things—it's no surprise that exposure to air pollution and a higher body mass index may be bad for your health—but it is, nevertheless, a new way of looking at the data that may be useful for researchers like me.
But to the average person, these results are essentially meaningless. It's possible that drinking champagne could improve your health—anything is possible—but it's far more likely that this is just another spurious correlation in the ongoing battle over moderate alcohol consumption. In any case, pop a bottle because you want to, not because it's good for your heart.
Sign up for Slate's evening newsletter.
