When the Canadian Centre on Substance Use and Addiction released its Guidance on Alcohol and Health (opens in a new page so you can follow along) it was after a period of public consultation. I spent a lot of time reading the report released in August, talking about it with colleague and friends, and generally becoming frustrated at the heavy handedness and distortions in the data.
When I read the final report, my frustration increased, but as a teacher it also provided a remarkable “teaching moment” about knowledge translation (ie: how to make complicated information comprehensible to the general public) and manipulation.
The funny thing about a scientific body releasing a report for “public consultation” is that it suggests that there needs to be some kind of general consensus on the “truth” of this data before it becomes a firm guideline. However, what was actually going on was consultation on how to make this message more powerful.
The data speaks for itself, one says, but sometimes the data has a rag in its mouth, is tied to a chair, and is shouting something you can’t quite understand:
“What’s that? Sorry? You’re saying ‘its really bad’?” says the temperance minded expert. “No,” thinks the hostage data “I’m saying ‘it’s not really bad’! Get this damn rag out of my mouth you terrorist.”
Ok, maybe that wasn’t fair, but you tell me. Last time I talked about relative versus absolute risk. this time I want to point out some ways that data can seem pretty freaking bad if you don’t know what they’re talking about, and if they are banking on that.
First: proportions. I mentioned last time about the way the data on pp 25-26 showed increased number of drinks per week and adjusted the scale so that after 7 drinks the number increased by 7, then 14 (this is a distortion of a logarithmic scale, which is used by data minded folks to compress time scales to show change over time when it takes a long time to change.
I’ll give the CCSA some credit. It was way better to use the idea of drinks per week than the earlier data that showed grams of alcohol per day, because most people don’t measure their alcohol by grams (!) and also the standard drink the CCSA mentions is 13.45 grams, so it’s tough to make a quick calculation to know what that means. Thank you for clearer information.
They also removed the deaths per year data which was not a good comparison. If you’re talking about likelihood of death over a lifetime, deaths per year is not useful (however, they may have dropped it because things like tuberculosis, which is strangely top on their list of harms, accounts for very few deaths per year. I’ll be talking about Tb another time)
But flip the page. Here we have a new table, something called “lifetime years of life lost (YLL).” This is probably new to you. It was new to me and I consulted an epidemiologist who does not do booze research and who said this is an unusual way to use this data, but maybe in this field it’s standard.
YLL per 1000 people basically means if you take 1000 people, over the course of their lives their lives will be reduced by that many years. So 500 YLL per 1000= 500/1000 years lost per person = half a year lost.
They don’t tell you this. But that is what it means.
So now let’s look at the table. According to the data (and again, I’m taking the data at its face value, I don’t have the wherewithal or the funding to dig deep into it to contradict the findings as data) a woman who has 14 drinks per week (still within the previous guidelines of 2-3 drinks a day but at the upper end) might lose 54 years per 1000 people. In other words, 0.05% of a year… carry the one… that’s 1.36 days.
This means she would die 1.36 days earlier.
And I can’t help but say if she doesn’t drink at all, she might have wanted to die even sooner. Sorry, dark humour is my drug of choice.
Apply this to every disease and you see that the risks are so incredibly low that we have to wonder why the hell they even generated this report. (I will try to answer that question but I can’t really assume I know their motivations)
The other thing to consider is that scale. I showed it last blog but I want to show it again:
(Source: Canada’s Guidance on Alcohol and Health, Final report, p. 27.)
It is weird to me. The report, er, guidance, is trying to make an argument about low risk drinking. It is also trying to argue that the previous limits of 2-3 drinks a day 14-21 drinks a week, (which is a lot, I’ll admit, and a heroic commitment to drinking) are too high. But they include 35 drinks a week. I am not sure why. If you are drinking 35 drinks a week, I applaud your boldness and hope you have egg salad sandwiches and those little pickles at your funeral (but seriously, that is a lot and I shouldn’t be making jokes, but I do love those sandwiches).
To me it seems to be a sort of predictive model. It suggests that people at 14 drinks/week might be on their way to 35. (5 drinks a day, wow). Or it might be another way of distorting data in order to make a broader policy statement about the dangers of drink.
Here is the problem. If you ask someone who likes to have a few drinks with friends, and might drink, say, 14 drinks a week, if they’d cut back to, say, seven in order to extend their life by, I dunno, four days, what do you think they’d say? Have you heard of a “spit take?”
By the way, on p30 the CCSA indicates the YLL based on the old guidelines. It is 755/1000 for men and 336/1000 for women. So nine months less for men and 4 months less for women. It also talks about “acceptable risk” which suggests that someone else should tell you what risk you should take with your life. That’s patronizing. There are all sorts of risks in life. It is your choice to take them, and as long as you don’t hurt other people in the process, have at er.
Next blog: temperance in historical perspective. I promise it will be short and not boring.
(c) Dan Malleck 2023