Soon after the second world war, Richard Doll helped to prove that smoking caused lung cancer. He was a smoker who suspected that it was the tar on Britain's roads that was to blame. Surprised by his own results, he gave up cigarettes. Doll then spent the rest of his long career figuring out the reasons people got cancer. But he enjoyed a drink, and he was intrigued by the effect alcohol might have on health. This, however, isn't easy to figure out: experiments in test tubes or on animals aren't particularly reliable, and you can't conduct the long-term randomised trials that work so well for other drugs. So in 1978, Doll sent out a questionnaire to thousands of British doctors, asking them about their drinking habits. He did the same thing to the same people in 1989. He repeated it a third time in 1991.
By 2004, he had acquired data on more than 12,000 men, 7,000 of whom had died. There was a clear link between alcohol and health: doctors who drank alcohol lived longer. Their chances of dying in any given year was about 20% less than their teetotal classmates. Doll suspected that there might have been a statistical error: he wondered if the results could have been twisted by high death rates among the ex-drinkers who may have quit after alcoholism or ill health. He went over the results a second time, this time comparing all the doctors who had ever regularly drunk alcohol with those who had never touched the stuff. The difference came out smaller, but the teetotallers still came off worse. Doctors drinking a unit of alcohol a day lived longest, but even those drinking an average of 40 units each week - the equivalent of 20 pints of beer - still lived longer than those who never drank at all.
Observations like these - unlike randomised trials - can muddle up causation with coincidence. Perhaps happy doctors happen to drink more and their happiness causes their good health. But Doll's study fits with other smaller ones. Binge drinking and alcoholism are bad, but regular drinking consistently comes out good. The only gloom is that the evidence is far less promising for women: it may be that the health benefits of alcohol only apply to men.
I heard Doll present his findings. Someone asked him if alcohol could somehow be reformulated as a pill. Puzzlement was clear in his reply. "Why would you want to do that?".
