Last week colleagues at Bristol and I published a paper looking at how surgical and cloth masks filter out droplets, and this week I read a recent UK Government SAGE (Scientific Advisory Group For Emergencies) report: Masks for healthcare workers to mitigate airborne transmission of SARS-CoV-2. The two publications are like the proverbial ships passing in the night. There is no overlap, let alone a meeting of minds. For example, in our paper we use “filter”/”filtered” etc 34 times – because masks are basically air filters we wear on our face – the SAGE report uses “filter” etc 0 times. There is a complete absence of curiosity as to what masks actually are, and how they work.
The SAGE authors seem only to consider epidemiological evidence for mask effectiveness, i.e., they only to want to look at evidence that is in the form of “set of healthcare workers A wore masks” while “set B did not”, and “fewer in set A became infected”. But we have very little of this type of evidence*. However, we have excellent evidence on how masks of all types work. For example, we know that the proper PPE masks, called FFP2/FFP3/N95, offer much more protection than the surgical masks nurses and doctors are being advised to wear in many circumstances. We know this because the protection can be**, and has been, measured. This part of SAGE ignores this high quality evidence.
As far as I can tell the SAGE authors want to treat assessing the effectiveness of masks just as they would a drug or a vaccine – i.e., in a way they are comfortable with. I don’t think that is the best way because masks are not like a drug or vaccine. With a drug or vaccine there is basically no good way for testing if they work, other than the sort of clinical trials medical doctors are familiar with. For a mask we can just measure how good a filter it is. We can straightforwardly do the measurements and show that a well fit FFP3 mask filters out 99%+ of droplets, while a surgical mask with a decent fit filters out around 80% at best. This shows FFP3 masks offer much more protection. There are no equivalent measurements you can do for a drug or vaccine.
I think here SAGE needs to look at a broader range of evidence, although to be fair we could have looked at bit harder at the epidemiological data in our paper.
* We don’t consider this epidemiological evidence in our paper, partly I guess because we have the opposite bias to the medics who wrote the SAGE report – we are more comfortable with measurements than epidemiological data – and partly because the epidemiological data is so limited and inconclusive.
** Actually, we can only measure how efficiently droplets are filtered out, but as we know the virus is in these droplets this is a good enough proxy here.
Chance & Necessity 