Safe alcohol drinking guidelines

You may have been rather surprised if you saw the story in today's news that "experts" are now recommending that we drink no more than 3 units of alcohol, or about 1 large glass of wine, per week. Yes, per week. Not per day, as is recommended in the government's current safe drinking guidelines.

So what's going on? Well, it's all down to some rather subtle differences between individual levels of alcohol consumption and population levels of alcohol consumption. So subtle, in fact, that the distinction seems to have been completely missed by the mainstream media that have picked up on the study.

The story is based on a paper published in BMJ Open. This was not a new piece of epidemiological research investigating the risks of different levels of individual alcohol consumption. It was instead a modelling study designed to investigate how different population levels of alcohol consumption would have different effects on mortality.

What the researchers did was to look at the number of non-drinkers and of drinkers in the overall English population, and at the distribution of alcohol consumption within the drinkers. Obviously, among the drinkers, not everyone drinks the same amount of alcohol. Some drink just 1 or 2 units of alcohol per week, while others drink large amounts every day, with most people coming somewhere in the middle.

The modelling study assumed that the variability of the distribution of alcohol consumption would stay the same, and then modelled various different scenarios of population average alcohol consumption. So if you set the population average to a high level, most drinkers will drink a lot, some drinkers will drink frighteningly large amounts, and only a small number of drinkers will drink moderately. In contrast, if you set the population average to a low level (such as 1 glass of wine a week!), most drinkers will drink small amounts, and only a few will drink heavily.

You could argue about how realistic those assumptions of constant variability in the distribution (on a log-transformed scale, if you want to get technical about it) are, but I don't think that's really the point.

Anyway, we know that drinking heavily is bad for you. We also know that drinking a small amount is good for you (mainly because of a reduction in cardiovascular risk), but the benefits of drinking a small amount are fairly modest, and the harms of drinking excessively are large. So it's not surprising that the researchers found that a low average level of alcohol consumption minimised mortality in the population, as a low average level would reduce the number of those drinking excessively.

This may be of some interest to policy makers. It is, however, completely irrelevant to any individuals wondering how much alcohol they can drink without having to worry about the effects it will have on their health.

The optimum population average alcohol consumption may well be 1 glass of wine a week, if we assume that there will be some variability around that and some people will drink more. This does not mean that the optimum amount for an individual to drink is the same.

I do not see anything in this latest research that suggests that current recommended guideline limits of 3-4 units of alcohol per day for men (and slightly less for women) are in any way inappropriate for individuals.

I think I might even have a little drink this evening to celebrate.