top of page

Raise a glass to the right experts

The temperance movement hasn’t had a boost like this since prohibition.

A major new international study on the health risks of alcohol, by the Institute for Health Metrics and Evaluation, in Seattle, was published last week in The Lancet. “The safest level of drinking is none,” the authors concluded. “Our results point to a need to revisit alcohol control policies and health programmes, and to consider recommendations for abstention.”

But if nonsense was graded like alcohol, this would be 100 proof. The authors are making statements and policy recommendations their own findings do not support.

The study’s results show the risk of developing an alcohol associated health illness such as heart disease or diabetes over the course of a year for non drinkers, is 0.914 per cent. For those who consume a single drink daily, the risk increases to 0.918 per cent. Two drinks per day raises the risk to…0.977 per cent.

So, among 100,000 people consuming two drinks per day, 63 more would be expected to develop health problems, relative to a group of 100,000 teetotallers.

The implication of the results is not that we should embrace teetotalism, but that the health risks of moderate daily alcohol consumption are, in fact, extremely small. One of the report’s authors, Professor Sonia Saxena of Imperial College London, reflected none of this in her comments to the media.

“Most of us in the UK drink well in excess of safe limits and as this study shows, there is no safe limit. The recommendations need to come down further, and the government needs to rethink its policy,” she said.

Let’s skip over the Lewis Carroll style paradox of being able to drink more than a nonexistent safe limit, and consider what UK government health recommendations on alcohol actually are.

In 2016, the government cut the levels of recommended alcohol for men and women to no more than 14 units per week, that’s six pints of average strength beer or seven glasses of wine. So roughly around one drink a day, which is what the study suggests is barely more risky than not drinking at all. That means government policy is, in fact, appropriate. One could argue it’s excessive, given the small relative risk increase associated with two drinks per day.

What lessons can we learn from this debacle? The lesson for researchers is, obviously, to be extremely careful not to make claims that are not clearly supported by their work.

The likelihood is these authors, focusing on the unquestionably severe health harms associated with high alcohol consumption, believed they had a moral responsibility to overegg their findings.

Perhaps they were worried by mixed messages coming out of other recent research suggesting moderate alcohol intake was beneficial. Perhaps they feared a slippery slope to mass binge drinking.

Whatever their motivation, the decision to call for total abstention was a mistake. And there seems to be a particular problem with this kind of overreaching in public health research – where the “x causes cancer” story has become ubiquitous, often based on rather tenuous or ambiguous results.

The line separating science from advocacy is inevitably blurry and we shouldn’t be overly strict about policing it. Yet these authors pretty clearly went too far and, sadly, have probably inflicted damage on the credibility of the wider scientific research community as a result.

What about the lesson for the public? There will be, doubtless, those who say this shows one shouldn’t trust scientists, because they have a hidden social or political agenda. So ignore the overwhelming consensus on climate change. Don’t listen to economists on Brexit. Follow your guts instead, or whichever non expert politician speaks most convincingly to your prejudices.

That would be folly. The right conclusion is not to disregard expert advice, but to make greater efforts to heed relevant experts, specifically those who can put raw results in an appropriate context.

The Lancet authors might be experts on the epidemiology of alcohol harm but they are not experts on risk and its evaluation. The fact their paper does not even state the absolute risks of alcohol consumption, and only publishes the relative risks – which is bad statistical practice and a breach of The Lancet’s guidelines – was a warning sign.

David Spiegelhalter, professor for the public understanding of risk at Cambridge University, has led criticism of the alcohol study’s scaremongering, improving what would otherwise have been woefully misleading media coverage.

The risks of getting bad expert advice are real enough. But the risks of listening to no expert advice at all, are far greater.


bottom of page