Working out if masks really work is hard

I am both teaching and shopping in a mask, and the UK is one of many governments that require people to wear masks in many situations. The obvious question is: Do masks work, i.e., reduce the transmission of SARS-Cov-2? Answering this question is hard, and to be honest we don’t have a good answer at the moment.

In medicine, it is the often said that the ‘gold standard’ in determining if an intervention, here mask use, works, is the Randomised Controlled Trial (RCT). Here this would involve taking a large number of people, assigning everyone randomly to two groups, then getting one group to weak a mask and the other to not wear a mask, but somehow getting the two groups to behave identically (apart from the mask wearing). Then you just look for any difference in infection rates between the two groups.

If the group wearing masks has a statistically smaller number of infections, you conclude that the RCT has shown the masks work. I can’t see such a trial being conducted, I am not sure it would be ethical or even lawful, given that COVID-19 is such a nasty disease.

But trials of this sort have been done for milder diseases such as flu, where ethics is much less of a problem as the potential for harm is much less. For example, ten years ago Aiello and coworkers split a cohort of students in halls of residence into mask-wearing and non-mask-wearing groups, and found perhaps a small reduction in infections with the flu in the mask-wearing group.

Brainaird and coworkers reviewed this literature earlier this year. They concluded that there are very few RCTs on transmission of respiratory diseases such as flu, SARS-CoV-2 etc, and so they provide a very weak evidence base. The RCTs suggest that it is unlikely that mask use dramatically reduces the transmission rate, although this may depend a lot on context, eg, mask use may (or may not) be more effective for say transmission in a work place or a restaurant where people are just sharing a room for a limited amount of time, than when they live together.

So, if RCTs are not going to be possible for SARS-CoV-2, what are the alternatives? One alternative is what are called observational data, where you just use gather data on people going about their business. The problem is what are called confounding factors. An example will hopefully make this clear.

I observe, via news websites, that over in Washington DC many more Republicans, who in many cases don’t wear masks, have been infected, than Democrats, who are much better at mask wearing. Can I conclude from this that mask wearing reduces transmission?

Not really, as there are many differences between Republicans and Democrats, in addition to mask wearing. These extra differences are confounding factors. For example, the Republicans are holding many indoor events in which people mix, such as political rallies etc. And it could be these differences that are driving most of the difference, not mask use.

So, we are in a pickle. I think we can’t do RCTs for SARS-CoV-2, and it is very hard to get strong conclusions from observational data, for something as complex as virus transmission. It may be that the best we can do is try and get as much observational data as possible, and in parallel build physical models both of how virus is carried through the air and how masks work. In essence, cobble together as much diverse observational data as we can, and rely on basic physics to both help us understand the data, and use it to make predictions. Predictions for how to reduce transmission as much as possible, with as little disruption as possible to how we want to live.

Leave a Comment

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

You are commenting using your 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