Wine and (my) Sleep
Based on the results of a Twitter poll, I've been drinking a glass of wine an hour or two before bed every night. I've heard from 'big sleep' expert Matthew Walker  that alcohol is bad for your sleep. Even just one glass of wine. I figured I'd look into my personal sleep data and see what I could learn.
A note on studies before looking into my data. I could not find many studies which indicate one drink is harmful for sleep. I found many that say three drinks or more is bad. Here's one observational study which finds a very small effect size for one drink negatively effecting the first three hours of sleep. I'd be interested in seeing more definitive studies.
Just to get ahead of the obvious critiques, keep in mind:
- My experiment is an n of 1. This makes it the most relevant study for me possible but perhaps my personal experiment is less relevant to you than the literature. However I'd argue an underrated aspect of health/life optimizations is the variance. If something isn't helping you you can just stop, but if it does, you can continue. It's more productive to try things half of people hate and the other half love than to try things everyone just slightly likes. Generally I find when one looks at any kind of psychology, nutrition or sleep study the effect size is so small that it wouldn't be worth changing your lifestyle to guarantee the average benefit. So just take this as a data point that alcohol before bed actually can be good for some people, so maybe you should try, and if it's not good for you you get to stop.
- I didn't randomize my intake so maybe January was just a different time for me than October through December.
- I didn't blind my intake so maybe it's the ritual or the enjoyment of the wine which makes me sleep better not the alcohol. Maybe I could get an even larger effect with some other pre-sleep habit.
You can see my data here. Note I cut it off the day we took our daughter home from the NICU because things got a little wonky then. And I started in October sort of arbitrarily, I've been wearing the fitbit since August. I also track my daily habits with daylio and deleted a couple days from after December 24th when I started my new habit when for whatever reason I didn't have a drink. The day by day is too noisy to look at and see anything but the 7 day average shows a very noticeable increase in REM around December 24th when I started drinking wine.
The data showed an increase in my REM sleep of 2.9% which a t-test indicates is significant at the 0.0002 level. My deep sleep was virtually unchanged with a truncated average of 15.5% for both time periods. My time awake decreased by 1% but only at the 0.08 significance level.
Note I didn't use Fitbit's sleep score because it's bullshit. They're in the pocket of 'big sleep'. They always give you higher scores for sleeping longer but I'd rather have 6.5 hours with no major wakes and 25% rem and 16% deep than 9 hours with multiple major wakes and 18% and 12% for instance. But Fitbit prefers the later.
These results are definitely good enough for me to keep up my habit. It's harder to quantify but I also feel I fall asleep faster and wake up feeling subjectively better rested now. I was surprised by the increase in REM because most studies I've seen say that's among the more negatively impacted parts of sleep. I'd encourage readers to try it out for themselves. Though definitely monitor how you're feeling for yourself and stop if you're getting different results than me.
- There does seem to be a temptation to have a second glass, so it's sort of a dangerous habit.
- There might be other benefits. I've heard resveratrol is a healthy compound in red wine. It does seem like most studies show red wine drinkers are actually healthier. I have my usual problem with those: Richer people and people who care more about health are more likely to drink red wine, are the studies really properly controlling for this? But it seems plausible.
- I find beer can give me a little acid reflux and gas which I think makes it a little worse for pre-sleep consumption.