Wednesday, 29 October 2014

The risks of drinking

There's a nice illustration of the gulf between 'public health' and reality in BMC Medicine this month in an essay entitled:

Why does society accept a higher risk for alcohol than for other voluntary or involuntary risks?

When the authors (some Canadian folk who credit temperance stalwarts Ian Gilmore and Tim Stockwell for providing 'helpful comments') say 'society' they really mean 'the government', but their question becomes much easier to answer if it is framed properly as:

Why do people accept a higher risk for alcohol than for other voluntary or involuntary risks?

The answer, quite obviously, is that drinking is highly enjoyable, the benefits are large and the risks are small to non-existent for moderate drinkers and tolerable for heavier drinkers.

Remarkably, this answer barely occurs to the authors. Instead they point the finger at things like alcohol's "cultural acceptance among elites in most western societies". They suggest that "the risks of alcohol may not be fully understood by the general public" and that "the addictive properties of alcohol cloud the consumers’ ability to assess information and make a free choice" And, inevitably, they blame "the strong political influence of global alcohol producers".

The methodology employed in this paper is rather odd. The whole article seems designed to revive the old temperance belief that there is no such thing as a safe drink and it hinges on this graph, which purports to show the absolute risk of dying as a result of alcohol consumption before the age of 70.

This graph bears no resemblance to the actual relationship between alcohol and all-cause mortality which, as regular readers will know, is a J-curve. (Note that one drink is about 15 grammes of alcohol.)

Their assertion that drinking just over one drink a day gives you a one in 100 chance of dying from an alcohol-related disease before the age of 70 flies in the face of so much evidence that there is little point in continuing. Nevertheless, this is how they proceed:

How do these risks compare to other acceptable risks in society? Many of the fully involuntary risks, such as unsafe water provided to a household, have risk thresholds set at one in one million. Indeed, the one in one million has become something of a gold standard of acceptable risk for involuntary exposure and has been used in different areas such as water safety in Australia and the US, or for increases of exposure to carcinogens in air, sediment or soil. 

Fine. It makes sense to be highly risk averse when it comes to water pollution or carcinogen exposure so long as there are low-cost ways to eliminate risk and there are no benefits lost by doing so. Nobody wants to drink unclean water. Lots of people want to drink alcohol.

Involuntary risks are associated with activities, conditions or events to which individuals might be exposed without their consent. Examples of involuntary risks include the risks of natural disasters (earthquakes, floods, and so on), or technology-related risks such as bad air quality or contaminated water.  

People who live on fault-lines and flood plains are exposed to earthquakes and flood with their consent. Millions of people live in California with the sure knowledge that the Big One will come sooner or later. The benefits they enjoy from living in California (God knows what they are) outweigh the risk of having their home razed to the ground and the possibility of being killed.

It should be noted that other standards have been used, and sometimes we see ranges, such as one in a million to one in 100,000. Starr found that the public seems to be willing to accept voluntary risks roughly 1,000 times greater than involuntary risks. By this standard, an acceptable risk for voluntary risks experienced by the drinkers themselves is one in 1,000 deaths for the pattern of behaviour over a lifetime. 

Do you know what else Starr said in his classic 1969 essay? He said this:

In the case of "voluntary" activities, the individual uses his own value system to evaluate his experiences. Although his eventual trade-off may not be consciously or analytically determined, or based upon objective knowledge, it nevertheless is likely to represent, for that individual, a crude optimization appropriate to his value system...
"Involuntary" activities differ in that the criteria and options are determined not by the individuals affected but by a controlling body.  

He was right. Involuntary risks that affect whole populations may require action by 'a controlling body'. Voluntary risks, like drinking, don't.

They continue:

If we accept the stated acceptable risk of one in 1,000 deaths, drinking 20 g pure alcohol per day (equivalent to 1.5 to 2.5 standard drinks dependent on the national standard drink: 8 g pure alcohol per drink in the UK, between 10 and 14 g in other European countries) exceeds this threshold ... drinking 20 g pure alcohol per day seems to exceed a threshold of one in 100 for death on a lifetime basis

Firstly, drinking 20g of alcohol per day does not exceed that 'threshold'.

Secondly, drinking 20g of alcohol every day for decades is not a single action. It is many thousands of individual actions over a course of a lifetime.

Thirdly, insofar as the one in a thousand shot is a meaningful threshold, it applies to voluntary behaviour and is therefore a matter for individuals, not 'a controlling body'.

To put this in perspective, the average level of daily consumption in EU countries in 2012 was about 31 g pure alcohol per day among drinkers, entailing a mortality risk beyond this threshold. This level of drinking has led to a situation where every seventh death in men and every 13th death in women before age 65 in the EU is caused by alcohol. Clearly, this level of risk is not acceptable by usual standards. 

This is willfully misleading. Aside from the fact that the statistics are questionable, the reference to average consumption of 31g of alcohol clearly aims to reinforce the false impression that this is a dangerous amount to be drinking. In fact, alcohol-related deaths before the age of 65 are dominated by liver cirrhosis—which requires a much, much larger alcohol intake than that—and various causes such as suicide, drink-driving and violence which tend to be the result of acute intoxication, not persistent but moderate consumption.
Only at the very end of the article do we get this brief, bewildered acknowledgement:

Or may the actual or perceived pleasurable effects of alcohol consumption (that is, benefits) be so high that the informed choice of a mortality risk in the 1:100 range is seen as “reasonable risk”, so that there is no pressure from the public for government action, and governments are dissuaded from effective alcohol policies such as raising taxes?

That's exactly what it is. Well done. Even if drinking at that level led to a 1 in 100 risk of premature death rather than - as is actually the case - a reduction in the risk of premature death, it would be entirely reasonable for people to do it if they enjoy it. See also driving a motorbike, going skiing, climbing mountains, eating cream buns or any of the other comparable activities that are brought into these debates.

If, however, someone doesn't feel that the benefits are worth it—either because they don't really enjoy drinking or because they are particularly neurotic about their health—then they are free to abstain. What they don't do, unless they work in public health, is demand "government action" and higher taxes for people who don't share their preferences.


Jonathan Bagley said...


DiCastelnuova defines, very vaguely and surprisingly only in passing, one drink as 10.5g of alcohol.

"The association with a lower mortality was apparent up to 42 g/d (about 4 drinks per day), and the lowest mortality was seen at 6 g/d, or about half a drink daily (RR, 0.81 [95% CI, 0.80-0.83])".

Chris Oakley said...

Rehm has a long track record of political public health activism. He is an East German statist loathed by many in his own country but taken in in Canada where healthism is more socially acceptable. He has no talent and no redeeming features. I assume that he is paid for by the Canadian tax payer these days.

Vova said...

No real disagreement about Rehm - except he's not an Ossie he's a Wessie - born in Augsburg, university in Munich. Not that it makes any difference to him being an over-rated chancer. The natural heir to Room.

Chris Oakley said...

Thanks Vova. I am guilty of assumption based on Rehm's current positions in Dresden /Toronto. Hopefully we are agreed on his track record and his qualities.