Distortions: What is a rational actor to do?

One of the challenges facing the Canadian Centre on Substance Use and Addiction relating to its Low Risk Drinking Guidelines is its hope to provide clear information to the general public. In an article in the Toronto Star on 29 January (behind a pay wall, sorry), the co-chair of the project to develop the guidelines, Catherine Paradis, noted that they wanted to provide information “that allow Canadians “that people can decide to integrate into their way of life, if they’d like to.”

This is a nice sentiment, rooted in the language of individual autonomy and freedom (choose to do this or not) and is at the core of some of the less radical language and distortions in the report. I applaud people who want to provide information; I challenge people whose information is ideological or at least distorted to the point of creating a sense of good and bad behaviour.

When we talk about freedom we need to consider its limits. You have probably heard a lot of talk about liberty and freedom these days. One of the key elements of classic liberty is the “harm principle” articulated by John Stuart Mill, philosopher and liberal in Victorian Britain. The Harm principle is the idea that you can do what you want as long as you don’t hurt other people. It was key to the ability of public health campaigners to get smoking out of public places, arguing convincingly that second hand smoke (or passive smoking) affected other people who had little choice but to be in those spaces (like employees at bars).

However, another key aspect of Mill’s idea of liberty lies in the ability of the individual to act in an informed way. Mills talks about the “rational actor” someone who has the ability to act rationally. The assertions of Dr. Paradis and others at the CCSA that they are just providing information upon which Canadian can act or not is a good example of the articulation of liberty for the rational actor.

Yet there is one thing a rational actor requires: clear information to inform their action. And that is where the CCSA information becomes problematic.

I have rattled on about relative risk versus absolute risk before. The tables on pp 25-26 of the Report show relative risk, but without indicating absolute risk. When I wrote this in an op ed in the Globe and Mail (no longer behind a pay wall), the CCSA pushed back, saying that the second set of tables, pp 27-28, show absolute risk and that my statement was, in their words “False.”

So, not wanting to misrepresent anyone, and hoping to provide you with clear information, I revisited those tables. Let’s take a look. Here is the table for “females” (I imagine they are talking about human females):

Source: CCSA Canada’s Guidance on Alcohol and Health: Final Report, p 27.

These tables show estimates of Years of Life Lost (YLL) per 1000 lives. It is a weird way of representing risk, because it is not representing the reality of any individual. We have only one life, we might expect to live 100 years if we are lucky, so a number of YLL/1000 is hard to comprehend.

Before we proceed I’ll say that I was not familiar with YLL data before these reports came out, so I spoke to epidemiologists who understand this stuff better. One told me that he had never seen YLL used in this way and that it was “Unusual.” Take that as you will.

One way to interpret this (and it has limitations I’ll discuss in a minute) is to divide the number in the table by 1000 to indicate an average of how many years an individual might lose. So taking the highest number in this table, we would say women who drink 35 drinks per week have a risk of losing 536/1000 years of life (due to intentional injuries, so I guess this is self harm). So 0.536 years=196 days (rounded up).

Ok, so how’s your risk tolerance so far? Is dying half a year early worth it?

Before you answer, consider the qualifications. According to the Toronto Star, which did a (typical for the Star) generally uncritical article on the new limits, averaging is not how you make use of YLL data. Considering the estimate that 17.5 YLL per 1000 lives is believed a tolerable risk for most people, the Star says that this does not mean everyone will have 17.5/1000 years lost, but in a group of 1000 people, “one unlucky person in 1,000 [will] die 17.5 years sooner.” You can’t just think in averages.

Well that’s scary.

(I’ll note that, in their “Limitations” (pp46-47) the CCSA notes that most of their data applies only at the population health level, which means it draws upon “Population averages” (p47) . I wrote about this in a previous post).

But, you might ask rightly, then how do we account for something like 536 YLL? I mean, I don’t have 536 years to live. I’m not Methuselah for goodness sake!

What’s a rational actor seeking clear information to do?

This is where my contention that the CCSA does not provide absolute risk data persists. Because a YLL/1000 calculation gives nobody a clear idea of their risk. It might have some meaning if you take it as an average, but you’re not supposed to use it just as an average. (Unless the Star’s reporter was mistaken, but I’ll give her and the Star credit for being smart and having fact checkers.)

Let’s continue with this table because it has been the subject of a distorting thread on twitter by people who seem to be critical of the CCSA report like me. We are in an ongoing discussion about how to interpret an make use of YLL data.

To their credit, the CCSA does not sum the risks at the bottom of each column. You might think such a sum would be a good idea, because wouldn’t it be logical to add up the number of years of life lost due to each condition and then get a full idea of how many years of life you can lose if you drink that many drinks per week?

Short answer: not at all.

Longer answer: these data show years of life lost and the end point of life is death. And you only die once. Unlike something that might show “number of brain cells lost” due, say to repeated hits in a boxing ring, YLL is a predictor of death. You cannot have cumulative deaths.

Although it might be possible to argue that someone could have a stroke which will weaken them, then cancer that might further limit their life, then say a heart attack and it’s game over, it is a distortion to say all of those numbers in each column indicates a real or even theoretical total loss of life. So a woman who drinks 7 drinks per week should not think she will lose on average (summing that column) 216.7/1000 years (=79.1 days) of her life. It don’t work that way. Excepting those who come back after their heart stops, most of us can only die once (I think–I haven’t died yet).

So again, what is an informed citizen, a rational actor looking for clear information, to do?

That’s a really tough question. Repeatedly I’ve said if you know you have a high likelihood of contracting one of these diseases (say you have a history of breast cancer in your family, or other conditions that are known to be hereditary) you might want to consider the information, talk to your physician, and act accordingly. (I’m not a physician, so don’t ask me). I say “Might,” because there are a lot of factors that go into causing any one disease.

At the same time, you might want to stop worrying. This data is drawn, as I’ve said many times, from a narrow band of information that does not give any consideration of the positive social role of alcohol in our lives. On top of that, this ignores the different effects of different forms of alcohol despite ongoing debates about the value, say, of red wine on heart health. I know, I know, the CCSA says there is no effect but they also claim that their recommendations are drawn from research that does not consider different forms of alcohol. So their denial is concerning–they’re not considering red wine, but are adamant that it has no protective effects.

I sympathize with what the CCSA has set out to do: provide good guidelines for people who are concerned about drinking. Give a rational actor information on which to base their activities.

But I challenge both the outcome of this project and the underlying ideological position of some of the researchers who have provided this data which has infected the entire project. I challenge many of the statements in the report — especially as I’ve noted earlier, the assertion that 7 drinks can increase your risk of heart disease and stroke when the data shows exactly the opposite to be the case, and the assertion that anything after 7 drinks “radically” increases your risk, when their data doesn’t show a radical increase (although their data also jumps from 7 drinks per week to 14, so any increase is distorted by such a jump in quantity).

And most of all I very strongly challenge the CCSA’s policy recommendations. Not only was it outside the scope of their mandate (the idea was to update the guidelines, not recommend policy) but it will serve to further distort the information they provide to rational actors thirsty for clear information.

One of the things they want to see are labels on alcohol. I’ll talk about that in another post.

(c) 2023 Dan Malleck

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s