Testing the total consumption model of alcohol

I have a new peer-reviewed study out in Economic Affairs ($) this month looking at the total consumption model theory as it pertains to alcohol. It builds on some of the research I did for Lockdown Lessons in Health Economics in which I showed that greatly reducing the availability and advertising of alcohol during the pandemic did not lead to fewer alcohol-related deaths (as 'public health' theory would predict). Instead there was a large increase in such deaths.

Put simply, 'public health' theory inspired by the neo-temperance movement that emerged in the 1970s says that drinking patterns in society move in concert, so if moderate drinkers drink less, heavy drinkers will drink less. They believe that if per capita consumption of alcohol declines, the amount of heavy drinking - and therefore alcohol-related mortality - will also decline.

This leads them to think that reducing per capita consumption should be the goal of policy. In short, they think the tail wags the dog. As I write in the article:

 

It is important to stress that the single distribution theory is not presented as a rough rule of thumb, but as an iron law. Skog (1985, p. 90) asserted that “when mean consumption increases, the consumption level of all kinds of drinkers increases”. Rose wrote in The Strategy of Preventive Medicine that “from the average alcohol intake of a population one can precisely predict the number of heavy drinkers” (2008, p. 121; emphasis added).

Adherents of the total consumption model do not believe merely that shifting the curve to the left is one way of reducing heavy drinking, but that it is the only way. As Cook and Skog (1995, p. 31) put it, “heavy drinkers will reduce their drinking if – but only if – the others cut back too”. Rose and Day (1990, p. 1034) argued that “to help the minority the “normal” majority must change”.

 

It's easy to test an iron law. If there are exceptions, it's not an iron law. And there are loads of exceptions to the total consumption theory. The aforementioned lockdown experience, when per capita consumption fell but deaths spiked, is just one of them.

 

Far from moving ‘in concert’, there is evidence of polarisation in drinking patterns in Iceland, Sweden and UK, with heavy drinkers consuming more and moderate drinkers consuming less (Bjarnason, 2006; Meier, 2010; Zeebari et al., 2017).

Further evidence can be found in the alcohol-specific mortality data, which can be used as a proxy for heavy drinking. In the UK, an 18 per cent decline in per capita alcohol consumption between 2004 and 2016 was not accompanied by a decline in alcohol-related deaths or hospitalisations (Whittaker et al., 2020, p. 1987). In the WHO European Region, per capita consumption declined between 1990 and 2014 but alcohol-attributable mortality rose (Hallgren et al., 2018). In the Australian state of Victoria between 1999 and 2007, there were “significant increases in alcohol-related harm” despite “relatively stable alcohol consumption levels” (Livingston et al., 2010, p. 368). In Norway and Ireland, rates of liver cirrhosis fell between 1980 and 2000 as per capita consumption rose (Bentzen & Smith, 2011).

I add to this evidence base by looking at patterns of consumption and harm in 174 countries between 2010 and 2019. There is no consistent pattern and 64 of the countries saw trends moving in the opposite direction to what would be predicted by neo-temperance theory.

 

In conclusion:

Taken together, the evidence leads us to a conclusion that is almost diametrically opposed to the public health orthodoxy, but is likely to strike the reader as no more than common sense. The amount of alcohol consumed by the average drinker does not affect the amount consumed by heavy drinkers. Unless the average consumer is drinking a dangerous amount, there is nothing to be gained from making her drink less. If efforts are to be made to tackle heavy drinking, they should be directed towards heavy drinkers rather than the general population.