Distortions: Relative risk and cirrhosis of the liver

In the new “Guidance on Alcohol and Health” the Canadian Centre on Substance Use and Addiction includes some shocking tables of risks of death by certain conditions (on pp25-26), dividing the data by biological sex (male and female) and colour coding higher risk in screaming red shading. No attempt at panic building here (I’m being facetious), but it is somewhat patronizing to assume people can’t see 55% and think it is high (I’m not being facetious).

The problem is, 55% increase might not be all that high. The table shows relative risk, so if you have a very low risk of a condition and it increases by 55%, you still have a pretty low risk.

But one thing I want to note is the information on liver cirrhosis. According to the CCSA, Men who drink 14 drinks a week have a 114% increase in liver cirrhosis; for women it is 444.7% increase. Holy shit! Let’s imagine this is absolutely accurate. You’d immediately stop drinking, right?

Hold my beer for a second. Let’s go back to the source: The study upon which this conclusion is made studied likelihood of cirrhosis as an outcome by comparing people who drank to people who did not drank. Taking this information, the CCSA then developed a formula (notice how the increase is gradual? For example, for women: one drink= 61.5%, two drinks=94.3 (32.8% difference), three drinks= 124.3 (34% difference) etc.

From Guidelines, p 25.

By the way, as an aside this is presented differently from the previous report which was released for public consultation. In that report the CCSA presented grams of alcohol per day and the scale increased by 5 grams. In this one the CCSA presents standard drinks per week, and the scale is distorted as follows:

This is why the 225.5% increase is followed by nearly double. It’s double the number of drinks per day.

Ok, that ham fisted data interpretation aside, what does this mean? Do women really have a 444.7% chance of developing cirrhosis of the liver if they drink 14 drinks a week?

I don’t want to make a mealy mouthed politicians answer, but I will start with context: this is a comparison of drinkers with non drinkers. In other words, it is a number relative to a non-drinker’s risk of developing cirrhosis of the liver.

We know that the main cause of liver cirrhosis is alcohol consumption. But you gotta drink a lot. If you compare the likelihood of someone who drinks alcohol to someone who does not, yes you are going to see a higher likelihood (aka risk) for a drinker. That’s because a non-drinker has almost NO chance of developing cirrhosis.

Ok so take a breath. Let’s continue.

The other thing to think about when looking at this particular data is the process of developing cirrhosis. Liver cirrhosis is irreversible. Once those scars develop, they don’t undevelop. However, unlike cancer, which can develop with few signs, cirrhosis comes after other changes that are reversible. The most discussed one is “fatty liver.” If you are diagnosed with fatty liver, yes, you probably should back off on your drinking and do other things that will help (talk to your physician for goodness sake, I’m not that kind of doctor).

This is the power of generating panic using relative versus absolute risk. A moderate drinker’s absolute risk of developing cirrhosis is not 125% higher if they drink 3 drinks a week. It’s nearly impossible to do that unless you have a liver that has some other problems in processing alcohol. The data presented here is the worst example of distortion of data, and an indicator of an attempt to manipulate the public in an unconscionable way.

Next blog entry: how to adjust your messaging and blur your data to freak people out.

(c) Dan Malleck 2023

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