Masks and Marketing in Medicine

Whether masks or meds, beware when someone sells it to you by quoting relative risk reduction

First - many apologies! I accidentally hit “send” and this got sent out in proto-form, unedited by Julie, which means typos and other idiocy completely of my making. This is the edited version…

There are 2 related but importantly different concepts in epidemiology.

Shark Attacks in the Bathtub

Relative risk reduction is the decrease in a risk caused by a certain intervention. Let’s say you tested a new electronic pulse shark repellent device and found it reduced shark attacks by 95%. That is pretty damn good. But it doesn’t mean that shark repellent devices should be mandated for universal use.

Wear a shark repellent in this situation. In fact, better to just get out of the water.

It might make a lot of sense to use this device spear-fishing on a deep-water reef off Hawaii at dusk. Or filming the next “Shark Week” documentary. (Or swimming off of western Cape Breton recently! Yeesh!).

But if you were to wear that same shark repellent every time you went for a dip at your local beach, you are likely wasting money and being paranoid. If you wear it swimming in a freshwater lake you are a clearly unbalanced. If you are terrified to bathe without it, you should be immediately committed to a mental facility.

No need for a shark repellent. That would just be weird.

We all understand this intuitively. To define it scientifically, the difference between wearing a shark repellent in Hawaii versus your bathtub, is understood by the difference between relative versus absolute risk reduction - RRR vs ARR.

RRR is easy to conceptualize - 95% decrease in shark attacks. It’s a nice, simple, fixed number. But ARR takes into account how high the risk is in the first place. The effectiveness of the same intervention changes depending on the circumstances in which you deploy it. You have a real risk of being attacked by a shark on that Hawaiian reef at dusk when the blood is streaming out of the fish you just speared. You have zero risk in your bathtub. If a risk is tiny to begin with, even reducing it greatly is a tiny benefit. And you can’t reduce a risk of zero. It’s ARR, not RRR, that actually matters to know if you should recommend an intervention in medicine.

How to know when you’re about to be a victim of slick marketing

A sales pitch that uses RRR is a red flag for scientifically literate docs. Salespeople prefer RRR because it sounds much more impressive than ARR.

For instance, a cholesterol-lowering drug might “lower your risk of heart attack by 25%” over 5 years. But if your risk was only 1 in 10,000 to begin with, that risk goes from 1 in 10,000 to 0.75 in 10,000. Or to be more clear, 4 in 40,000 to 3 in 40,000, (since as far as I understand it, we can’t save 1/4 of a life). If you are at low risk for heart attack, you can’t benefit much from this drug, even though 25% is actually quite a large RRR compared to most of the drugs we use in medicine today. The ARR is 1 in 40,000 or 0.0025%. Furthermore, if the drug had even a small risk - say 1 in 5000 - of serious side effects, these will outweigh the benefit.

On the other hand, if you were in a high-risk group with a 10% chance of a heart attack over the next 5 years, that same 25% reduction means you go from 10 chances in 100 to 7.5 chances in 100 of dying of a heart attack, or again to use round numbers 20 in 200 to 15 in 200. The odds of this drug helping you are 5 in 200, or 2.5% ARR. That sounds a lot more useful, and the drug might be worth the costs and side effects.

Many of us would be willing to pay for, and take, a drug every day for 5 years to lower our risk of heart attack by 25%. But surveys have shown that the vast majority of us wouldn’t be willing to take a drug that only reduced their odds of heart attack by 1 in 40,000 over 5 years. But this is actually the same stat, stated 2 different ways, if talking about a low risk group.

(PS: These are not real numbers about cholesterol drugs, just fictional examples. So don’t stop your meds, or refuse to start taking meds, based on this! Talk to your doc. But do ask about ARR vs. RRR. PS: and try to eat properly and get plenty of exercise.)

ARR vs. RRR with masking

Anyway,... masks! Right, masks. This does actually relate...

Therein lies the problem with masks. There has been some modelling and laboratory work (done with computer simulations, or in some cases, dummy heads with masks attached) showing that masks may reduce aerosol or droplet (more on that later) transmission.

Despite being super-cool, computer simulations don’t always predict real-world results.

But even if these studies are correct, reduction in particle transmission pertains to RRR. And remember it’s ARR that we need to focus on. And to calculate ARR, we must factor in that at any given time the vast majority of us don’t have COVID, nor are around someone who does. Masks can’t reduce COVID if there is no COVID around to reduce. The shark repellent will not help in a bathtub. And if most people who use the repellent are in bathtubs, the overall effectiveness of them is very poor.

There is a concept called “case density” - how many cases are there per 1000, or 10,000 in the general population. Conceptually it’s easy to understand that at very low case densities, mask mandates are ridiculous. My province of NS has 1 million people. If there is only 1 case in the whole province, can a mask mandate really be justified? 2 cases? 10? 100? Even at 1000 active cases, only 1 in 1000 people has COVID, and presumably the majority will be isolating at home. (As an aside, if most are not isolating, how do we then justify spending massive amounts of money on an aggressive testing program?). Thus the odds of anyone you pass at the grocery store having COVID are actually very low. Public health has never set a “case density” threshold for mask mandates. Instead, they imposed a mandate without scientific justification, and The Science™ says it will be safe to stop masking on March 20th at midnight. It all feels arbitrary.

Having played a wind instrument, and having cleaned it, I can attest that the stuff you are blowing into it is blowing out of it. Much better to put the mask on the instrument instead.

So there lies problem #1 with mask mandates. There could be a potential benefit of masks worn when around someone who actually has COVID, or by the person who has COVID. But these are likely small to begin with (more later!). And these benefits, and thus the ARR, get watered down by the fact that at any given time most of us are unlikely to be in a COVID-positive environment. And this is why on a macro scale, data does not seem to show an effect of mask wearing. (more on that later too…)

NEXT UP: Aerosols vs. Droplets and plausibility of mask efficacy