As the buns continue to fly back and forth across Hadrian’s wall in a mock war of independence, cheeky commentators south of the border have dubbed face masks, now compulsory when out and about north of the border, jock straps, on the basis that they are indeed straps of cloth fitted to all Jocks, of either sex. Inevitably, the Scottish compulsion, which came into effect last week, has prompted a frenzy of hand wringing and bed wetting in England as to why we aren’t doing the same thing. More moderate minds wonder what the evidence is for the effectiveness of masks, given that any compulsion is also inevitably an infringement of liberty, and that, as such, any compulsion must be proportionate, insofar as the benefits outweigh the harms.
The crux of the matter is the compulsion. At the moment, an individual in England is free to decide whether or not to wear a mask in public. To move to a state of compulsion, with punishment for non-compliance, requires more than a cursory nod to the precautionary principle, it requires evidence of effectiveness. If there is no such evidence, the compulsion cannot be proportionate, because there is no way of establishing the benefit, and so no way of testing the proportionality. This approach, of first establishing whether there is evidence of effectiveness, has the added bonus that if we don’t find evidence of benefit, then we do not need to fret over potential harms, since the proposal has already failed the proportionality test.
What evidence do we have for the benefits of face masks? At present, very little. One leading authority, and it’s vocal president, that has come out in favour of wearing face masks in public, is the Royal Society, a cabal of our great and good scientists. It has produced a 37 page report which gets off to a good start by declaring itself to be both pre-print (only it has been published) and un-peer reviewed. A good many pages cover what other countries do (presumably to establish an Americans carry guns so we should too line of reasoning), behavioural mumbo jumbo, and more mumbo jumbo on the need for joined up public messaging on control measures. But what about the evidence for the benefits — reduced disease spread — of wearing face masks in public? After all, this is the Royal Society: if anyone can do a decent job of finding the evidence, then few are better equipped for such a challenge.
It turns out there is precious little evidence. Instead, it seems we have an exercise in political decision making (make masks mandatory), followed by a reworking of the evidence to underpin this political decision – science led by the politics, rather than the other way round. We can see evidence of this in the way the normal process of evidence appraisal gets turned on its head.
Section 2.2 ‘Evidence of effectiveness of public wearing of masks and coverings in community-based studies’ kicks off with ‘A repeated concern raised by some is that there are few randomised control trials (RCTs) with conclusive results examining the effectiveness of face masks conducted in community settings’. In fact there are virtually none, but these ‘repeated concerns’ are summarily dismissed as calls for unwarranted rigour in the face of awkward to do studies. Having at a stroke dumped the need for good quality evidence — no need to bother with that sort of nonsense round here, dear boy — and adding, for good measure, a poke in the eye of unwarranted rigour: ‘we note that there have also been no clinical trials of coughing into your elbow, social distancing and quarantine, yet these measures have been widely adopted’ (so that’s alright then), they then proceed to scrabble around in what evidence there is to prosecute their polemic. It does not go well.
First up is a pooled meta-analysis, in which ‘the authors [of the pooled meta-analysis] conclude that there was no significant reduction in influenza transmission with the use of face masks’. This awkward finding is summarily dismissed on the grounds that a number of the pooled studies were under-powered, implying there probably was a benefit, only it failed to materialise because of methodological flaws. Speculating there might have been a small effect, but it was not detected because of lack of statistical power, is speculation, not science.
Second up is a ‘non-peer reviewed medRxiv pre-print meta-analysis of around 20 studies’. Putting aside the fact we are now dealing with non-peer reviewed pre-prints to the power of two — one inside another — this study, or rather three RCTs within it, appeared to show a non-significant 6% reduction in the odds of developing influenze like illness/respiratory symptoms for mask wearing (OR 0.94, 95% CI 0.75 to 1.19). This finding, based small RCTs in unusual settings (university halls, Hajj pilgrimage), is small in effect size, and not significant — there is no evidence of benefit. Even if there was, it could not be generalised. Observational studies fared marginally better, but were plagued with both self-reporting biases and confounding. Though not reported verbatim by the Royal Society, the meta-analysis concludes ‘the evidence is not sufficiently strong to support widespread use of facemasks as a protective measure against COVID-19′.
And so on. ‘No significant reduction in influenza transmission with the use of face masks’; ‘the evidence is not sufficiently strong to support widespread use of facemasks as a protective measure against COVID-19′. There seems to be something of a pattern emerging. Nonetheless, the Royal Society authors soldier on with their polemic. Next up is their own meta-analysis of four healthcare based — ie not community based — case-control studies, mostly with small numbers of subjects, looking at inward (wearer protection) H1N1 and SARS transmission. By a miracle of numerology, the pooled results show a possible but questionable small effect, but it is meaningless. As the Royal Society authors themselves admit, ‘We also note that these are case-control studies and do not show causal relationships‘.
So far, the studies taken into account by the Royal Society have only looked at protection from infection for the mask wearer. What about protection for others? Here things take a turn for the worse, if that is possible: there are no clinical studies. Instead, the Royal Society authors turn to a proxy (instead of looking at the clinical outcome of real interest, you look instead at an alternative (and easier to study) factor that might (or might not) have an impact on the outcome of real interest — looking at cholesterol levels as a proxy for cardio-vascular events in statin studies is an example), in this case the in vitro filtration capabilities of various fabrics both on their own and in various combinations. The bottom line seems to be that if you wrap your face up in half a dozen or high quality cotton pillow cases and silk scarves, preferably wearing rubber soled shoes to maintain an electro-static charge, and stick you head in the upstream chamber of filtration testing apparatus, you might notice less salt in the air. Providing, that is, there are no mask leaks. Even small leaks, typically from poor fit, dramatically reduce effectiveness, and small leaks are hard to eliminate. But even with no leaks, we still have no clinical evidence that face masks reduce transmission in vivo. They might, or they might not: we simply do not know.
In summary, the best evidence the Royal Society can muster is:
1. a meta-analysis (Xiao et al) which found ‘no significant reduction in influenza transmission with the use of face masks’
2. a meta-analysis (Brainard et al) which concluded that ‘the evidence is not sufficiently strong to support widespread use of facemasks as a protective measure against COVID19’
3. the Royal Society’s own flawed (cannot show a causal relationship) case-control meta-analysis
4. an in-vitro study (Konda et al) of a proxy measure of transmission, which cannot be extended to a clinical setting
which is tantamount to there is, as yet (like Keynes, Dr No follows the ‘if the evidence changes, I change my mind’ dictum), no evidence for the effectiveness of face masks in reducing covid-19 transmission.
Given no evidence of benefit, no evidence that face masks reduce covid-19 transmission, the case for compulsion falls at the first hurdle of the proportionality test. There is, as things stand, no case for a legal requirement to wear a face mask in public. Instead, it should remain a matter of individual choice.