A recent study found that moderate alcohol consumption — even one drink a day — could shrink your brain. The explosion of context-free headlines predictably followed. Let's dive a little deeper and examine what most reporters missed.
Are you still afraid of COVID-19? Perhaps it's the prospect of an escalating conflict in Ukraine that has you worried. Forget all that. The real threat to your health might be the IPA or glass of wine you had with dinner last night. That's because, according to a study recently published in Nature Communications, even moderate alcohol intake may harm your brain. The hipster-oriented “I F****** Love Science”—get it? Edgy because of the profanity—put it this way:
Bad news for casual drinkers – just one to two alcohol units a day results in lower brain volume, only getting worse as consumption increases, a new study suggests.
The study continues to challenge the belief that small amounts of alcohol are of no consequence – or are even healthy – finding that those with significant brain alterations from alcohol consumption may have cognitive impairments as a result.
This is nonsense. The study was little more than a comparison between the self-reported alcohol consumption and MRI scans of 36,000 people. It poses no serious "challenge" to the prevailing wisdom about moderate alcohol consumption. If you can responsibly enjoy a few drinks with your friends or family, feel free to ignore IFL's hyperventilating. [1]
A tiny effect
The press release for the paper and the copycat news articles that drew from it were almost entirely devoid of numbers, but loaded with caveman-speak like "More alcohol, less brain." That's probably because even taking the study's results at face value reveals a relatively small effect. “Alcohol intake explains 1% of the variance in global [gray matter volume] GMV and 0.3% of the variance in global [white matter volume] WMV across individuals beyond all other control variables,” the researchers wrote.
When they excluded “abstainers” and heavy drinkers (i.e., women who consume more than 18 units a week and men who consume more than 24 units a week), the results were mostly unchanged. However, “the variance explained by alcohol intake beyond other control variables is reduced to 0.4% for GMV and 0.1% for WMV when individuals who consume a high level of alcohol are excluded.”
Previous studies have estimated that we lose five percent of our brain volume every decade after 40, which can accelerate as we reach our golden years. The possibly alcohol-induced changes reported in the present paper are relatively insignificant against such a backdrop.
But that's only if we take the results at face value, which would be a mistake. Here's why: the researchers noted that
"An additional limitation stems from the self-reported alcohol intake measures in the UK Biobank, which cover only the year prior to participation. Such estimates may not adequately reflect drinking prior to the past year and are susceptible to reporting and recall bias."
Biobank study participants range in age from 40 to 69 years old; the mean age of participants in the current study was 63.72. As a result, the researchers had no idea how much alcohol these individuals consumed throughout most of their lives. One possible problem that stems from this massive gap in the data, as the authors acknowledged, is that their “analyses do not account for individuals with a past diagnosis of [alcohol use disorder] AUD.” Alcoholism is characterized by undulating periods of heavy drinking and partial or complete abstinence, accompanied by declines and improvements in brain health. That kind of fluctuation almost certainly complicates the association.
I can't be certain of this, but I suspect that most people who consume alcohol follow a similar though less extreme pattern, drinking more or less at different points in life depending on their circumstances. For instance, single, highly educated men may drink more than those who are older and less educated. Women with less education and poorer health also report less alcohol consumption.
Self-reported data is meaningless
But we have a problem even when we consider the year's worth of data the researchers had to work with. The simple fact is that people don't reliably report their exposures in epidemiological studies, either because of faulty memory or dishonesty. In fact, we lie about everything from church attendance and voting to exercise and eating habits.
For example, in a 2013 assessment of self-reported data from the National Health and Nutrition Examination Survey (NHANES), researchers noted that participants underreported their calorie consumption to levels that were “physiologically implausible (i.e., incompatible with survival).” In other words, they'd be dead if they ate as little as they claimed. Do people more accurately report their alcohol consumption? I don't know, and neither do the authors of this new paper. That's precisely the problem.
The problem with epidemiology
None of these variables is the researchers' fault. All the sexy epidemiological work was completed years ago: we know about the deadly effects of smoking, the causes of heart disease, and the link between HPV and cervical cancer. Many epidemiologists have therefore turned their attention to the often insignificant influence of individual risk factors on specific health outcomes. Some experts have tried to push the field away from this methodology, fearing marginalization since many “public health successes of the past century owe little to our methodological developments as a discipline.”
The bigger problem is that this ongoing dispute in epidemiology has done nothing to discourage the researchers, universities, scientific journals, and media outlets that benefit from exaggerated headlines warning that “just half a beer” could have serious consequences. Until I see better evidence to the contrary, I'll keep enjoying the occasional cocktail without fear of brain shrinkage. Feel free to do the same.
[1] Dr. Alex Berezow called IFL "the Daily Mail of science journalism" in a 2020 ACSH piece. Hopefully, it sticks.
