Is moderate alcohol consumption really healthy?

I'm going to start this article by revealing my own biases. I'm not sure where the idea that alcohol might be healthy comes from. It's pretty well established that alcohol is poisonous to all living organisms. That's why we use it to disinfect surfaces, and why I lather my hands in it several times per hour when I'm working in the hospital. It interferes with the functioning of cell membranes, and at high enough doses it causes the cell membranes to fall apart completely, killing the immersed organisms. That's the reason it is such an effective disinfectant. It doesn't take a great leap to think that something that interferes with the functioning of our cell membranes might not be too good for us.

As an interesting aside, the alcohol disinfectant I use to rub my hands in at work is perfectly drinkable, if you're desperate enough. Which is why the nurses generally remove all the disinfectant containers from the patient's room if they know the patient has an alcohol addiction.

Alcohol is a very energy dense molecule. One gram of alcohol (a.k.a. ethanol) contains 7 calories. This can be compared to carbohydrates and protein, which both provide 4 calories per gram, and fat, which provides 9 calories. This is probably why we've evolved the ability to metabolize alcohol and turn it in to energy in the first place. Those of our ancestors who could eat any fermented fruit they happened to come across were less likely to starve. Which is the reason why it's possible to subsist (for a while at least) on a diet consisting entirely of alcohol. It's also the reason alcoholics generally develop severe vitamin deficiencies - they're getting most of their energy needs met from a source that is completely devoid of other nutrients. I don't recommend.

So where does the idea that moderate alcohol consumption is healthy come from? My guess would be that the companies that sell alcohol have something to do with it. I wouldn't be the least bit surprised to find that they'd seeded the scientific literature with articles supporting the healthiness of alcohol, just as the sugar industry has seeded the literature with articles supporting the healthiness of sugar. It could also have something to do with the fact that moderate alcohol consumption is part of the mediterranean diet, which we've been told for decades is one of the healthiest ways to eat.

Anyway, enough about my various preconceptions. Let's get to the science. I'm going to focus here on overall mortality, because it doesn't matter if alcohol for example were to decrease your risk of getting cardiovascular disease but at the same time increased your risk of getting cancer by as much or more. What matters is the overall effect on longevity.

A systematic review and meta-analysis was published in The Journal of Studies on Alcohol and Drugs in March 2016. It sought to answer the question of whether moderate alcohol consumption has a beneficial effect on overall mortality.

The study did one thing in particular that was very important. It sought to correct a mistake that is common in studies of the effects of alcohol on health, which is to lump former alcoholics who are now abstainers in to the same category as people who have been abstainers their whole lives. Obviously, lumping the two groups together will tend to make abstinence look worse than it is, because former alcoholics generally have pretty bad health to begin with, and will tend to have worse than average outcomes going forward, regardless of how much (or little) they now drink.

Support for thinking that this might be a big problem comes from the fact that the J-shaped curves (i.e. the curves that slope down first, and then up, thereby suggesting that moderate alcohol consumption is healthier than both abstinence and heavy consumption) can be seen in observational studies correlating alcohol with a wide range of different disease states. It defies belief that the same curve would be seen for pretty much any disease you care to look at, from deafness to the common cold to dementia, which suggests strongly that the J-curve is caused by underlying poor health in general, and not by the alcohol itself.

Even liver cirrhosis, a disease primarily caused by heavy alcohol consumption, has been found in observational studies to be more prevalent among abstainers than among moderate drinkers, a clear sign that former alcoholics are being lumped in with life-long abstainers, and that this is confounding the results. So in order to truly see if there is any benefit to moderate consumption, these groups need to be separated out from each other, which is what this systematic review did.

The review included 87 studies with a total of four million people. Participants were followed for an average of 13 years. Note that all the studies were observational in nature. Unfortunately, people aren't generally willing to be randomized to a certain amount of alcohol consumption, so that is the type of data we have to work with. Most likely there never will be a randomized controlled trial in which people are randomly assigned to a certain amount of alcohol consumption.

Unlike randomized trials, observational studies cannot prove the existence of cause and effect relationships, all they can do is show correlations and thereby provide clues. However, if observational studies fail to show any correlation, then you can usually be pretty certain that no meaningful cause and effect relationship exists.

Participants with less than moderate alcohol consumption were divided in to three separate categories. One of these categories consisted of people who had never consumed any alcohol. One consisted of people who had previously consumed alcohol but had now stopped completely - a category that likely to a large extent consisted of recovered alcoholics. And one consisted of people who drank alcohol occasionally, but consumed less than one drink per week.

These categories were compared with people with a "moderate" consumption, here defined as more than one drink per week but not more than two drinks per day. They were also compared with three additional categories representing ever higher alcohol intakes. Just to clarify what we mean by "one drink" - In this case, one drink was defined as around 10-12 grams of alcohol, which is equivalent to a small glass of wine. So, when we say one drink, we don't mean a gigantic American size wine glass filled to the brim.

So, what were the results?

After adjusting for confounders, the mortality rate was the same among life-time abstainers, people with a low alcohol consumption, people with a moderate alcohol consumption, and even with people with slightly higher than moderate consumption (up to four drinks per day).

When I say it was the same, I mean there were no statistically significant differences in mortality between the groups. Remember that we're talking about four million people here who were followed for an average of 13 years, so if there is no meaningful difference in a study with that much data, then there is no meaningful difference in reality.

Former drinkers who had now stopped, on the other hand, had a mortality rate in line with that seen in current heavy drinkers (roughly a 40% increased relative risk of dying during follow-up, as compared to lifetime abstinence).

So, what can we conclude?

The J-curve commonly seen in studies of alcohol consumption, suggesting that moderate alcohol consumption is beneficial, is purely a statistical artifact, produced by lumping former heavy drinkers in with people who have never drunk. Moderate alcohol consumption is not healthier than abstinence. So, if you've been feeling pressured to drink alcohol because you think it's good for your health, you can relax. It isn't.

Interestingly, never drinking any alcohol and drinking up to four drinks per day are equivalent from a longevity perspective. So, if you are one of those people who drink a glass of wine or two with dinner, and you're concerned that it might be harming your health, you can also relax. The evidence suggests you don't need to worry.

This is quite surprising to me. I've been trained to view someone who drinks four drinks per day as borderline alcoholic, and certainly at high risk of long term health issues. The current official NHS guidelines, for example, say that you should consume at most 10 drinks per week, which is far lower than this data suggests you can safely consume without having to worry about harming your long term health.

So who does need to worry according to this review? Everyone consuming more than four drinks per day on average. As alcohol consumption increases beyond that point, there is a sharp increase in mortality.

Unfortunately, former drinkers who stop drinking appear to remain in the same risk bracket as continuing heavy drinkers, suggesting that the ability to roll back the damage done by prolonged heavy drinking is limited. To be fair though, this is probably another statistical artifact, as many heavy drinkers won't stop drinking until they're really sick and quite close to the end of life.

So, to sum up: moderate consumption of alcohol is not correlated with any beneficial effect on longevity, as compared with avoiding alcohol completely. Nor is it correlated with any harmful effect on longevity.