Richard Doll, smoking and moderate drinking

First published by Spectator Health in January 2016


Sally Davies, the Chief Medical Officer, recently described the belief that moderate alcohol consumption was good for the heart as an ‘old wives’ tale’. This was the culmination of a long-running campaign within a section of the public health lobby to cast doubt on the large body of evidence showing lower rates of heart disease and lower rates of mortality among moderate drinkers. A report from researchers at Sheffield University, released on the same day, claimed that the health benefits of drinking were ‘disputed’ and the subject of ‘substantial debate’.

It is difficult to imagine any amount of evidence persuading Sally Davies that moderate drinking is healthy. The protective effect of alcohol on the heart was first observed in 1926 and countless studies from all around the world have confirmed it in the 90 years since. It seems that Davies places a much greater burden of proof on scientists who find a positive effect from drinking than on those who find a negative effect. This can be illustrated by looking at two light bulb moments in the career of the legendary epidemiologist, Richard Doll.

In 1950, Doll, along with Austin Bradford Hill, published the first epidemiological study showing a link between cigarette smoking and lung cancer. Many people were sceptical and potential flaws were quickly flagged up. The study had been limited to hospital patients in and around London, the vast majority of whom were smokers. In response, Doll and Hill got back to work and published a further study in 1952 which expanded its geographic reach. It came to the same conclusion.

Doll then initiated a prospective study which tracked the health of smoking and non-smoking doctors around the UK. When the first full findings were published in 1956 they once again showed a clear link between smoking and lung cancer risk.

In the meantime, researchers from other parts of the world conducted similar epidemiological studies with similar results. Gradually, it came to be accepted that the relationship was causal — smoking caused lung cancer.

Scepticism did not disappear overnight but as the years went by it was increasingly confined to the vested interests of tobacco companies and to maverick scientists who put forward alternative theories which might explain the statistical association between smoking and ill health. For example, the great statistician Ronald Fisher suggested that the early stages of lung cancer gave people the urge to smoke. Implausible though they were, such theories were not rejected out of hand but were subject to rigorous empirical testing until they were found wanting.

By 1976, when Richard Doll and Richard Peto published another edition of the doctors’ study, there was a wealth of evidence to support ‘the smoking theory’ (as it had been known) and there was a broad scientific consensus that the relationship between cigarettes and cancer was causal and proven. Alternative explanations had been tested and debunked.

Some people still refused to believe it, but they were a dwindling minority. The sceptics now became ‘merchants of doubt’, cherry-picking individual studies that seemed to undermine the larger body of evidence and raising spurious objections that had already been addressed in the scientific literature. Some demanded an impossible burden of proof by calling for randomised control trials which could not possibly be conducted even if it were ethical (which it would not be). Others said they would not believe smoking caused lung cancer until the exact biological mechanism by which it did so was identified.

Then, in 1994, Doll (now Sir Richard Doll) identified another statistical association in the data from the doctors’ study. In a study published with Richard Peto, he found that all-cause mortality was lower among moderate drinkers than it was among non-drinkers and heavy drinkers. It was not the first time such an association had been observed but some people were doubtful — not about the risks of heavy drinking but about the apparent risks of not drinking. Alternative explanations were again put forward, notably the possibility that some non-drinkers may have been former drinkers who had put their health at risk and were therefore at greater risk of premature mortality. This came to be known as the ‘sick quitter’ hypothesis.

As he had done when conducting research into smoking, Doll addressed his critics by carrying out a new epidemiological study. He published an article in 1997 looking at the question of causality which rejected the hypothesis that the association was due to confounding factors. Then, a few months before his death in 2005, he published a study based on 23 years of data which replicated the results of his previous studies while disproving the sick quitter hypothesis by comparing lifelong non-drinkers with moderate drinkers. The latter had lower rates of heart disease and lower risk of premature mortality.

By this time, Doll concluded: ‘That the inverse relationship between ischemic heart disease and the consumption of small or moderate amounts of alcohol is, for the most part, causal should, I believe, now be regarded as proved’. A Department of Health working group appeared to agree, noting: ‘All the evidence we have received confirms that the relationship between all-cause mortality and alcohol consumption follows a J-shaped curve. Non-drinkers have higher all-cause mortality than light and moderate drinkers’.

Doll was not alone in this research. As with the smoking-lung cancer finding, he was supported by researchers from around the world whose studies came to the same conclusion: moderate drinking lowered the risk of all-cause mortality and of heart disease in particular. In 2006, a meta-analysis of 34 prospective studies concluded that men who drank up to four drinks a day and women who drank up to two drinks a day had a lower mortality risk than those who did not drink at all. The sick quitter hypothesis was repeatedly tested and found wanting. The protective effect on the heart was repeatedly shown to be real and not the result of unhealthy former drinks in the non-drinking group.

As before, a noisy minority continued to deny these findings. They insisted that the biological pathways were unproven, though plausible pathways had been identified. They made generic criticisms of epidemiology that could apply to any observational research, though they never made them of studies which showed negative effects from drinking. Above all, they treated the sick quitter hypothesis as an unanswered question, never acknowledging that it had been tested extensively.

Decades after the evidence on moderate alcohol consumption had first been identified, those who refused to accept it were embroiled in a campaign of doubt and denial similar to what Doll had witnessed in the mid-20th century, but this time the naysayers were on the inside of the public health establishment, albeit in its neo-temperance wing. For years, they chipped away at the science, repeating the same old criticisms, cherry-picking studies and demanding an impossible burden of proof from researchers. They received a sympathetic hearing from their public health colleagues who had long struggled with the nuanced message that heavy drinking was bad while moderate drinking was good. Preferring a simple, clear, strong message that alcohol was dangerous, they were similarly inclined to dismiss or downplay the epidemiology.

Taken as a whole, the evidence was too strong to overcome, but so long as the critics persisted the evidence would, by definition, be ‘disputed’ and there would still be a ‘debate’. That was the line taken by the authors of the Sheffield University report when the Chief Medical Officer commissioned fresh research for the new drinking guidelines. In the 20 years since the Department of Health’s working group had concluded that epidemiological studies ‘strongly indicate a direct causal relationship’ between moderate drinking and cardiovascular health, the evidence had grown and become stronger, but anyone reading the Sheffield report would have got the impression it was on the brink of falling apart. Anyone listening to the Chief Medical Officer on the Today programme would have assumed it had already fallen apart. ‘An old wives’ tale’, she said. And with that, the job was done.