Even one or two drinks a day can cause brain damage to your hippocampus, according to a recent study in the British Medical Journal. But, how much faith should we put into these findings?
I’ve pointed out the flaws in studies that try to link benefits to supposedly bad behavior, such as the study that says eating cheese is good for you, or that chocolate is good for your heart. The former was funded by cheese and dairy industry groups, and the latter was not a controlled study that had lots of statistical limitations. We should subject this paper to the same scrutiny – and not just because it’s a message we don’t want to hear.
The BMJ is the same journal that published that “chocolate is good for your heart” study, and upon closer examination you’ll find that this one has many of the same caveats.
The study involved 550 Londoners who filled out surveys about their drinking habits and had an MRI every 5 years for 30 years. The paper concludes that one drink per week is associated with a 0.01 percent decline in the size of your hippocampus, which is associated with memory and navigation.
You don’t need to be a statistician to see the problems here:
- This study reports only on correlation; it is not a double-blind experiment and so we can say nothing about causation.
- 550 people, all of whom are part of the same social group and in London, probably aren’t representative of all humans to begin with.
- 0.01% is a very small change to detect, especially given this sample size. For comparison, your hippocampus shrinks 0.02% per year on its own, just due to aging.
- There is the potential of selection bias, as participants in the study had to have the means and willingness to travel to Oxford from London to undergo these periodic tests.
- Here’s the real kicker – for some reason, this change was only statistically significant for the right hemisphere of the hippocampus. It makes no sense that alcohol would affect it in such a selective manner. What would make more sense is that the arbitrary threshold chosen for “significance” just happened to be right on the cusp of the results for the right hemisphere observations – but “statistical significance” does not mean the effect is real!
In fact, a different way to report on this study is “having 8-12 drinks per week actually isn’t that bad for your brain.” If there is an effect, it’s barely detectable. That’s a more honest way to report these results. But it’s also true that claims of moderate drinking being beneficial are also probably sketchy.
To be fair, the authors are quite up front and transparent about these caveats, as they were in the chocolate study. But not all press coverage really presented these findings in their full context.
Hang on while I get a beer.
Image credit: iStock.com / Ezhukov