The return of "no safe level of drinking"

In 2012, I wrote a post on this blog titled 'Towards Zero':

For the last couple of years I have been predicting that the time will come when the recommended drinking guidelines (colloquially known as "limits") will fall to zero. One day we will be told that the daily/weekly units system is too confusing for the plebs and that there is no safe level of alcohol consumption. The best advice, the experts will say, is to not drink at all.

I predicted this not because the evidence of health benefits from moderate consumption was getting weaker - it was getting stronger and has continued to do so - but because it would be politically expedient for the 'public health' lobby if the optimal level of drinking was zero.  

History told us this, as I said in a subsequent post:

The reason I have predicted that the 'no safe level of alcohol' and 'alcohol has no medicinal benefit' claims will gather pace in the years ahead—despite plenty of evidence that they are lies—is because they were two of the core beliefs of the nineteenth century temperance movement. They were integral to the nonsense of Scientific Temperance Instruction and they helped clear the way for full Prohibition in the twentieth century.

But it is not twentieth century prohibitionists who provide the blueprint for the twenty-first century war on alcohol. It is the modern anti-smoking lobby. Charles Moore was perceptive enough to see this after Sally Davies, the Chief Medical Officer, made her 'no safe level' remark in 2016...

The model in their minds is tobacco. Having succeeded in virtually outlawing smoking, they want to do the same with alcohol. If they can create the public “fact” that there is no such thing as safe drinking, they can then attack everyone who brews, distills, makes wine, or runs a pub, club or restaurant, for pushing something which is unsafe. They can also have a go at anyone who advertises any of the above, and insist on health warnings and, later, bans.

The health benefits of moderate consumption make no difference to the liberal and economic arguments against anti-alcohol policies. Informed adults should be free to drink as much as they like so long as any external costs are internalised through Pigouvian taxation.

But they are a real headache for anti-alcohol campaigners, hence the concerted effort to convince the public - and, more importantly, politicians - that they do not exist.

And yet the J-Curve (below) showing lower rates of mortality at moderate rates of drinking survives, and while efforts to undermine it have gained pace in recent years, it has been a stop-start affair. Every now and again, Tim Stockwell will pop up with a cherry-picked meta-analysis to 'debunk' it but then a string of new epidemiological studies will be published confirming the benefits of drinking for dementia, cardiovascular disease and overall mortality.

Modelling provides the campaigners with their solution. Let's be blunt. You can make a model say anything, as Public Health England proved when they told the Sheffield University Alcohol Research Group to change their model to justify lower guidelines.

A simple model can be useful for illustrating a theory and a predictive model is the best you can hope for when looking to the future, but recent years have seen the 'public health' lobby not only use models to predict the future (as with minimum pricing) but also to rewrite history (as when reviewing smoking bans and sugar taxes) and rewrite science itself (as with the guidelines).

Today's widely reported study in The Lancet is in the latter category. It represents the most concerted effort yet to erase the J-Curve and claim, as the prohibitionists once did, that every drink is deadly.

The study makes little attempt to hide the political agenda that inspired it. Here are some snippets...

These results suggest that alcohol control policies might need to be revised worldwide, refocusing on efforts to lower overall population-level consumption.

.. it is crucial for decision makers and government agencies to enact or maintain strong alcohol control policies today to prevent the potential for rising alcohol use in the future

Our results point to a need to revisit alcohol control policies and health programmes, and to consider recommendations for abstention.

Governments should consider how these recommendations can be implemented within their local contexts and broader policy platforms, including excise taxes on alcohol, controlling the physical availability of alcohol and the hours of sale, and controlling alcohol advertising.

The results themselves - even if they were true - would not imply a particular set of policy recommendations. This is pure editorialising, and there is more of it elsewhere in the journal with an op-ed which emphasises that the study provides 'strong support for the guideline published by the Chief Medical Officer of the UK who found that there is “no safe level of alcohol consumption"' and calls for 'a minimum price per unit (MUP), closely followed by marketing regulation, and restrictions on the physical availability of alcohol'.

Tellingly, the authors of the op-ed (who include Nick Sheron) claim that these are 'the most effective measures for curbing tobacco-related harms, another commercially mediated disease, with an increasing body of evidence showing that controlling obesity will require the same measures'. So much for there being no slippery slope.

The study virtually eliminates the J-Curve. I suspect we'll be seeing this graph a lot in the years ahead as decades of epidemiology are quietly discarded...

How did they do it? The study contains no new evidence and uses an unusual modelling approach based on population-wide data from various online sources. If you look at this massive appendix you can see the kind of data they were using. The figures are extremely crude.

The authors don't dispute the benefits of moderate drinking for heart disease but they claim that the benefits are matched by risks from other diseases at  low levels of consumption and are outweighed by the risks at higher levels of consumption. Some diseases which have been associated with benefits of drinking, such as dementia, are excluded from the analysis entirely. They also ignore overall mortality, which you might think was kind of important.

The study is essentially ecological (the weakest form of epidemiology), with the authors looking at alcohol consumption in different countries and comparing rates of disease. They say that one of the unique selling points of their study is that they adjust for under-reporting of alcohol in countries where there is a lot of illicit or illegally imported booze. They fail to note that illicit and surrogate alcohol, which is drunk in vast quantities in poor countries is inherently more dangerous than the stuff produced by Big Alcohol. Nevertheless, it is Big Alcohol they are gunning for.

The weakness of the methodology and the global perspective mean that the findings have little relevance to any individual country and have no relevance to developed countries like Britain. For example, the authors find a strong, linear association between alcohol consumption and tuberculosis which starts from the first drink. This makes moderate drinking seem riskier than previously suggested, but the risk of getting tuberculosis from drinking at any level in the UK is approximately zero so what relevance does it have to us?

If you want to know the effects of moderate drinking in western countries, look at the evidence from western countries (spoiler: it's good for you). If you want to know if drinkers live longer than teetotallers, you follow large groups of people over time and track their rates of disease and overall mortality. You don't try and work it out by going online to find national averages from vastly different societies, nor do you confine your analysis to alcohol-related diseases. 

The fact of the matter is that the modelled estimates in this study don't align with a mountain of epidemiological evidence (both cohort and case-control studies) and they don't even align with national estimates of disease prevalence (countries with higher rates of alcohol consumption have longer life expectancies). The study tries to do something with aggregate data from whole nations that is much better done by studying human beings. And it has been much better done by studying human beings, with the evidence showing time and time again that moderate drinkers live longer than teetotallers. 

This study is a fundamentally political piece of work and it is no surprise to see it published in the most political 'public health' journal. It is another step on the road to 'alcohol control' and is aimed at the delegates who will be attending next months UN meeting on 'non-communicable diseases'. Whatever its academic merits - and they are few - its ramifications for drinkers could yet be profound.

You can read my comment for the IEA about this here and David Spiegelhalter has some interesting thoughts here.