Low Amounts of Drinking are Healthy
Subscribe to Ja3k
That's right this blog is now a newsletter.
Thanks for subscribing!
Why are people so eager to claim small amounts of alcohol are bad for you? For example see:
- This World Heart Federation Policy Brief which doesn't say anything about alcohol's effects on cardiovascular health, presumably because they are good, but does say "There are no safe recommended levels".
- This interview with Matt Walker on sleep.
As far as I can tell there are no studies that show a drink a day have a significant negative impact on mortality or sleep. And in fact many studies which show some health benefits. For instance see:
I did a little digging into the World Health Federation's references to see how they arrived at their recommendation. From the results of their citation 11, Thun et al.:
The rates of death from all cardiovascular diseases were 30 to 40 percent lower among men (relative risk, 0.7; 95 percent confidence interval, 0.7 to 0.8) and women (relative risk, 0.6; 95 percent confidence interval, 0.6 to 0.7) reporting at least one drink daily than among nondrinkers, with little relation to the level of consumption.
To be fair that paper also says in its abstract that alcohol increases the rate of breast and other cancers and the policy brief echos this claim. But most importantly, again from Thun et al.: "The overall death rates were lowest among men and women reporting about one drink daily."
This result seems quite well replicated. See this other large n study which also investigates potential mechanism and this meta analysis.
From the abstract of a metastudy, Irsaad et al.:
At all dosages, alcohol causes a reduction in sleep onset latency, a more consolidated first half sleep and an increase in sleep disruption in the second half of sleep.
This is all good except for that last thing. But it seems to be a wash at low doses:
Total night REM sleep percentage is decreased in the majority of studies at moderate and high doses with no clear trend apparent at low doses.
I've noticed since I've started having a glass of wine everyday that I get up easier in the morning. I've attributed it to being excited to see my baby but maybe the second half disruption is helping.
- This study was posted to hackernews with a title like "Even a small amount of alcohol is negatively associated with brain size"  but if you open the paper you see this:
Which shows a small amount of alcohol has beneficial to no effect on brain size.
But TheSe aRe AssoCiaTIons 
To pre-empt the obvious criticism: these are associative results, here are a million reasons not to trust them:
- The "abstinence" category is full of people who are recovered alcoholics or have underlying health conditions which prevent them from drinking.
- Problem drinkers die young so aren't represented in these mortality studies.
- Richer people drink more and have better health outcomes.
- This whole paper which lists a bunch of reasons not to trust the results but embarrassingly can't actually cite a study in which the effect disappears through controls because such a thing does not exist.
First there are experiments. The sleep results are experiments. Unfortunately there are no mortality terminated randomized alcohol experiments in humans. But there are experiments in pigs and humans which show alcohol has an effect on the cardiovascular system I gather the researchers think is healthy. This is relatively weak evidence of the original claim that low levels of drinking are healthy but I think it's a problem for people who want to explain away the effect as non-causal.
To the claim that the abstinence category in these studies is confounded with sick people: in the Thun et al.'s study they separate out the categories "less than once daily" and "once daily" and you can see the once daily category has a very slightly lower all cause mortality than the less than once daily category. Their study also only puts people who have indicated abstinence for the past 10 years in the abstention category.
I couldn't find a good study that controlled for wealth unfortunately. Looking at a Gallup poll the difference does seem potentially large enough to explain the effect. But how do we know that this isn't one of the causes of rich people living longer? It seems true that richer people drink more on average but all the studies I found wanted to compare rich drinkers to poor drinkers not rich drinkers to rich non-drinkers and poor drinker to poor non-drinkers.
A meta note on how I choose how much stock to put in an associative study: If an associative study says "Thing everyone knows is healthy is healthy" I trust it a lot less than if it says "Thing everyone knows is unhealthy is healthy" because there's a huge correlation between health practices. People who exercise are more likely to eat right etc.
I've got (Grape)skin in the Game
I actually am convinced by my own claim here and started to drink a glass of wine every night about a year and a half ago . I did a (poorly designed) self sleep study in which I concluded it did improve my sleep . I do think wine is healthier than other choices of alcohol, see this literature review, but I'm less convinced by that then my core claim that one unit of alcohol a day is good for you. I mostly drink wine for aesthetic reasons.
 I think hn changes the title when it's particularly misrepresentative so now the submission is just the paper title.
 Sorry if this is rude to my opposition. I just thought it was a clean way to give my subtitle an opposing explanation while making clear I disagree with it. I'm almost always the guy doubting associations as non-causal. But in this case I think that would be a mistake.
 I'm often making contrarian claims but I really do believe them. I actually do use Android, I actually prefer to eat chicken to beef on ethical grounds and I actually never got boosted.
 The beauty of sleep science is you can test most of the interventions on yourself and pick and choose what works for you. There's no reason to rely on what works for the median person except as a starting point. Unfortunately you can't do this for mortality.