Teetotallers still dropping like flies

Another day, another large cohort study confirming the benefits of moderate drinking. This time it's a 14 year study involving 333,247 Americans which concludes:

Compared with lifetime abstainers, those who were light or moderate alcohol consumers were at a reduced risk of mortality for all causes (light—hazard ratio [HR]: 0.79; 95% confidence interval [CI]: 0.76 to 0.82; moderate—HR: 0.78; 95% CI: 0.74 to 0.82) and CVD (light—HR: 0.74; 95% CI: 0.69 to 0.80; moderate—HR: 0.71; 95% CI: 0.64 to 0.78), respectively. In contrast, there was a significantly increased risk of mortality for all causes (HR: 1.11; 95% CI: 1.04 to 1.19) and cancer (HR: 1.27; 95% CI: 1.13 to 1.42) in adults with heavy alcohol consumption.

So that's a reduction in mortality risk of more than a fifth for both light and moderate drinkers compared to teetotallers. A similar magnitude of risk reduction was found in a BMJ study earlier this year. Numerous studies have been showing the same thing since the benefits of moderate drinking were first noticed (and covered up) 45 years ago.

The usual alcohol policy tweeps from the 'public health' lobby have been quiet today - perhaps because they are on holiday - and so the task of muddying he water has fallen to Buzzfeed. Its article, by Tom Chivers, mentions the usual excuse that non-drinkers are all terribly ill because they used to be alcoholics. Although Chivers acknowledges that this study only looked at lifetime abstainers, he implies that most studies of this kind do not. In fact, most studies conducted in the last fifteen year have only looked at lifetime abstainers. The 'sick quitter' hypothesis is a thoroughly debunked zombie argument that should have been dropped long ago.

With the sick quitters put to one side, he takes the fall back position of suggesting that there is something else about teetotallers that makes them less healthy:

But people who've never drunk in their lives are fairly unusual, in Western society, as well. So there may be some other factor that we haven't thought of.

There are several problems with this line of argument.

Firstly, saying 'teetotallers are less healthy because, er, reasons' isn't much of an argument to begin with.

Secondly, if it is an argument, it is an argument against all epidemiology. Different groups are usually different in several ways. That is why epidemiologists adjust for confounding factors. Of confounding factors, Chivers says: 'You can try to avoid these problems, but you can never do it perfectly.' This is trivially true, but there is a hint of the Nirvana fallacy about it. Perfection is impossible in observational epidemiology. The question is not whether it is perfect, but whether it is good enough. In this study, the researchers took account of a wide range of factors, including physical activity, smoking, race, body weight and education. They also adjusted for the prevalence of various diseases. The association between teetotalism and death remained.

Thirdly, it is not at all clear that teetotalism correlates with unhealthy behaviour. On the contrary, it often correlates with healthy behaviour (whereas heavy drinking correlates with unhealthy behaviour).

Fourthly, teetotallers are not 'fairly unusual' in the United States. A third of Americans are teetotal. Forty per cent of the people in this study did not drink and 23 per cent of them had never drunk. By contrast, only 5 per cent were heavy drinkers. It is therefore heavy drinking that is fairly unusual, but Chivers doesn't fault the evidence on alcohol harm from heavy drinking on the basis that there might be something weird about heavy drinkers.

A few other points are worth mentioning because they have been generally absent from the news reports about this study:

Firstly, the authors found a reduction in risk for cancer mortality among light drinkers, and no increase in cancer mortality risk for moderate drinkers. So much for there being 'no safe level'.

Secondly, they found no increased risk of heart disease mortality among heavy drinkers (but a large reduction among the other drinkers).

Thirdly, moderate drinking was defined as up to 14 drinks per week for men and up to 7 drinks a week for women. A 'drink' is not a unit, however. In this study, a drink is 14 grams of alcohol whereas a British unit is 8 grams of alcohol. The male moderate drinkers were therefore consuming up to 24.5 units and the female moderate drinkers were consuming up to 12.25 units.

The Buzzfeed piece claims that the statistician David Spiegelhalter 'says [the study is] a vindication for the new, low-risk NHS guidelines'. It's hard to see how. The authors clearly define moderate drinking differently to the NHS and there was a reduction in mortality in the group that included men who drink much more than the 14 units now recommended by the government. (NB. the only change to the guidelines was the reduction in the male limit.)

Spiegelhalter (who was on the guidelines committee) is quoted as saying: 'Once you get above the NHS guidelines of 14 units a week, that's when risk starts taking off.' Nothing in this study supports that statement. Is he getting confused between a US 'drink' and a UK 'unit'?

Fourthly, the health outcomes for light drinkers and moderate drinkers were virtually identical across the board, so there is no evidence for the claim that is sometimes made about the benefits of drinking only existing for people who drink tiny amounts.

Finally, an accompanying editorial notes that...

For most older persons, the overall benefits of light drinking, especially the reduced cardiovascular disease risk, clearly outweigh possible cancer risk.

Can we expect the 'public health' lobby to recommend that teetotallers start having a few drinks? Don't hold your breath.