Is Drinking Bad for Your Brain?
Most Americans drink. According to a 2019 survey, 86% of American adults have drunk alcohol at some point in their lives, and a new study now claims to show that drinking any amount of alcohol can harm your brain.
This study first made the news in late May. CNN, Fox, Yahoo, The Guardian... every organization reporting on it used roughly the same headline – i.e., some version of the study's title and conclusion ("No safe level of alcohol consumption for brain health...").
Startling news.
It surprised me too that the study wasn't reported by highly reputed agencies such as the New York Times. Why not? Presumably because it hasn't been peer reviewed yet.
Ordinarily, I wouldn't read a study that's not reported in a peer-reviewed journal. However, this particular study was conducted by University of Oxford researchers. Studies coming out of prestigious institutions like Oxford may or may not be credible, but they do merit a close look, if only because the rest of the world will be looking.
I won't keep you in suspense about my conclusion. I don't think the data support the conclusion that there's no safe level of alcohol consumption. So, pour yourself a (small) glass of your favorite beverage and let's have a look...
The Oxford study focused on 25,378 participants in the UK Biobank, a subset of the largest brain imaging sample in the world and an enormous repository of data. The researchers examined the relationship between alcohol consumption and brain health, while controlling for a variety of demographic and health variables. (For those of you who are stats-savvy: Appropriate corrections were made to ward off Type I error, and statistical power seems good.)
The main finding was a linear relationship between alcohol consumption and brain health – the more alcohol consumed per week, the greater the harm to the brain, as indicated by diminished volumes of grey and white matter. Although the relationship was especially strong for people with high blood pressure, high BMI, and/or a history of binge drinking, it was observed regardless of physical health status and drinking behavior. And so, the researchers concluded that there's "no" safe level of alcohol consumption for brain health. That's a strong statement, and one that might alarm you if you're indeed sipping on something right now.
I believe it's clear from this study, and others, that drinking a lot can be harmful – for yourself, for others, for the flowerbed you park your car on after a night on the town. However, this study does not show that small amounts of alcohol consumption harm your brain. Rather, in some respects the findings are inconclusive, while in others the data seem to show that small amounts of alcohol are safe – and even beneficial.
Let's start with a simple question: How did the researchers measure alcohol consumption? (In technical terms, how was alcohol consumption operationally defined?) On the surface, the approach to measurement was straightforward: Participants were asked how many glasses of alcohol they drank in a typical week and month. Here's the rest of what the researchers said about that:
"Numbers of glasses were asked for, and subjects were given guidance about how many glasses in the normal bottle. Glasses were converted to UK units as follows: red or white wine = 1.7 units; fortified wine = 1.2 units; pint = 2.4 units; spirits = 1 unit; other (e.g. alcopops) = 1.2 units. Amounts were also converted to grams pure ethanol (1 UK unit = 8g) to aid international comprehension and comparison."
That's it. That's all the researchers tell us about what they asked participants. This leads me to my first concern about their conclusion:
1. The measurement of alcohol consumption was likely to be inaccurate.
I'm sure you can see some sources of inaccuracy. Participants not understanding the "guidance". Participants reporting consumption incorrectly. Participants drinking wine or spirits differing in alcohol content. Participants not using glasses. Etc.
My concern here is not a systematic bias toward, say, overestimating amount of alcohol consumed. I'm just referring to measurement errors that would undermine our confidence in the findings. (If you've taken stats, you may recall the distinction between random and systematic measurement error. What I'm claiming here is random error, which can lead to Type I or Type II error).
Here's a second concern:
2. The minimum amount of alcohol consumption is not specified.
Non-drinkers were excluded from all analyses, so there must be a minimum amount or range of consumption. However, neither text, tables, nor figures contain that number.
A description of one of the analyses hints that the 5th percentile of consumption was 0.2 UK units of alcohol per week, or about 1.6 grams. This is just over one tenth of a standard drink (as defined here). If that's correct (and I'm not 100% sure) then the smallest amount of alcohol consumption in this study is indeed small – just a few swallows of wine per week, on average. However, clarity is needed; we shouldn't have to guess. And we still face the measurement issues I mentioned earlier. Worst of all:
3. The actual data show no adverse effects for the lowest levels of alcohol consumed.
I've pasted in one of the researchers' graphs below. Notice that at the far left, the line seems flat! (If this graph is too small, have a look at the third-from-last page of the study.) In other words, for some range of values, an increase in alcohol consumption seems unrelated to a decrease in gray matter. Unfortunately, we don't know what that range of values is, because the details aren't given on the x-axis (or in the article).
All the graphs presented by the researchers look like the one above (except for a few in which brain health improves as alcohol consumption increases from the lowest levels to some undefined point where the line starts to trend downward).
So, even though the line in the graph above shows an overall relationship between alcohol consumption and volume of grey matter, a close look at the line doesn't show such a relationship for the smallest amounts of alcohol consumed (whatever that happens to be). This alone calls into question the title and conclusion of the study.
There's also a pesky little problem for the researchers concerning brain function. Along with brain structure (grey and white matter), the researchers also examined the functional connectivity of the brain at rest. Here, the main finding was that greater alcohol consumption was associated with greater functional connectivity and better cognitive performance! To be clear, the researchers weren't saying that the more you drink, the better your brain works while drinking. Rather, they found that the more participants drank during a typical week, the better the functional connectivity of their brains (while sober), and the better their cognitive performance (while sober).
The researchers noted that interpreting this finding is "difficult". Well... perhaps it's difficult because they were so committed to the notion that alcohol consumption harms the brain!
I could stop here, but I won't. This is a study coming out of Oxford University that got international media attention, and the findings are relevant for many people. More concerns can be raised. I will close with just one:
4. The researchers didn’t rule out all plausible confounds.
The researchers were careful to examine the impact of demographic and health variables, but here's an important variable that wasn't included: Stress. Stress has been linked to diminished brain health (e.g., lower grey matter volume – see here). Decades of research has also linked stress to alcohol consumption (see here). Conceivably, for some range of values, greater stress could've led Oxford study participants to drink more, but the increasing loss of grey matter they experienced may been attributable to stress rather than alcohol. (Some of you will recognize this as a simple example of a third-variable problem.)
So, what's the take-home message?
First, the results of the study do not show that with respect to brain health there's no safe level of alcohol consumption. Rather, the results seem to show that the lowest levels of alcohol consumption either have no effects or actually confer benefits on brain structure and function. (I say "seem" because of the possibility of measurement error and other problems.)
Second, the results of the study do show overall linear relationships between alcohol consumption and brain health. Specifically, it looks like greater alcohol consumption leads to diminished volumes of grey and white matter, along with better functional connectivity and cognitive performance.
Should we be concerned about the link between drinking alcohol and reduced grey/white matter? Well, sort of. Although I've discussed problems such as measurement error, the researchers' findings with respect to grey/white matter are consistent with other studies on the topic. Drinking alcohol does kill brain cells, to put it crudely. But the magnitude of the effects in this study are quite small. For instance, across the entire dataset, alcohol consumption only accounted for 0.8% of variance in grey matter volume. Normal aging, in contrast, accounted for 27% of the variance. (Stats note: these are R-squared values.) Informally speaking, in a sample consisting of people who range from not drinking much to drinking a lot during an average week, normal aging was roughly 30 times stronger than amount of alcohol consumption as a predictor of differences in grey matter volume. The only rationale for even mentioning that 0.8% figure is that (a) it's significantly greater than zero (even though it's a tiny number), and (b) unlike aging, alcohol consumption is controllable.
In sum, drinking small amounts of alcohol each week does not seem to damage your brain, according to this study. It might even do you good. Although "small" can't be defined very precisely, I may still enjoy a glass of wine tonight.
Cheers, and thanks for reading!