Mandates or recommendations — how strong is the evidence?
Disposable surgical masks are loose-fitting covers that were designed to be worn during surgery as a physical barrier to protect accidental contamination of patient wounds, and to protect the wearer against sprays of body fluids. They were not designed to inhibit viral transmission, and there is limited evidence that they do so in a meaningful way, at least in the case of influenza. This was the conclusion of the UK Influenza Pandemic Preparedness Strategy, 2011 (the latest). The document also concluded that: “There is very limited evidence that restrictions on mass gatherings will have any significant effect on influenza virus transmission”. Most countries had pandemic preparedness strategies in place prior to COVID-19 (C19), which had been formulated in a time of calm reason. They were discarded almost immediately.
Herein, I will limit myself to evidence surrounding the use of surgical or cloth masks for limiting the spread of C19 in community settings.
A note on laboratory studies
I’m not going to dwell on these studies because they have little relevance to the real world. Certainly, it can easily be shown that putting a mask of some sort over a person’s face will limit transmission of droplets of certain sizes through the mask for the short-term duration of the study. This is not rocket science. However, mask recommendations or mandates go well beyond laboratory-controlled conditions, and may introduce hazards, for example through mishandling and rebreathing of contamination.
Influenza viruses — where the data is
While these viruses are not SARS-CoV-2 (SC2), they are respiratory infections that share principal modes of transmission with SC2. We have substantial data on masking for influenza, which may guide us with our response to C19.
A recent 2020 review and meta-analysis of randomised controlled trials (RCTs) of surgical mask use in the community concluded: “In our systematic review, we identified 10 RCTs that reported estimates of the effectiveness of face masks in laboratory-confirmed influenza virus infections in the community. In pooled analysis, we found no significant reduction in influenza transmission with the use of face masks.”
Note that RCTs are considered the highest standard of evidence, and that these studies used surgical masks that might be expected to be more effective than cloth masks. Further, when they incorporated hand hygiene, there was still no benefit.
Whilst these studies will have had their limitations, overall it’s not a good track-record for masks and influenza.
Healthcare workers, a RCT incorporating cloth masks
There are more studies in healthcare workers than for the general public. As I’m focussing this post on mask mandates for the general public I won’t delve into this topic in depth, but again, it could still inform us and there’s a study comparing surgical and cloth masks of some interest.
A RCT of influenza in hospital workers compared usual surgical mask use (as per procedural guidelines, i.e as needed), with all-day surgical mask wearing and all-day wearing of reusable cloth masks. The study was carried out in Vietnam, where (as for much of Asia) there is compliance with face coverings. There was no difference in infection rates between the first 2 groups (procedural or continuous surgical masking). However, the cloth-mask wearers experienced an influenza-like illness that was on average 13-times more frequent than either of the surgical mask groups. That is a massive increase. The hypothesis was that reusable masks might be more likely to become contaminated and infect the wearer, especially if worn for long periods (such as the working day).
So, in this group of experienced health workers, working in a regulated and controlled environment, reusable cloth masks were a hazard. What then of cloth masks for the general public? While we don’t know to what extent we can extrapolate from influenza to C19, the study is a cautionary tale.
The one RCT for C19
We have less data on masks and C19, and I am only aware of one attempt at a RCT. In the beginning of 2020, Denmark was neither recommending nor mandating masks and only about 5% of the population chose to wear a mask when in the community outside the home. Researchers recruited about 6,000 non-mask wearing people, and randomly gave half a supply of surgical masks with instructions on how to use them properly (not common for the general public elsewhere). The other half were told to continue with their practice of not wearing a mask. After ~2 months, they compared infection rates in the two groups. Infection occurred in 42 participants in the mask group (1.8% of the group) and 53 control participants (2.1%). This difference was not statistically significant.
The investigators had considerable difficulty getting these results published in a scientific journal of standard — it was presumably not the result people were looking for. Their paper was rejected by 3 consecutive journals, before it was accepted by the 4th (Annals of Internal Medicine). As an academic scientist myself, I can see where the authors were almost certainly required to water-down their paper to suit reviewers and editors, and to keep open the hypothesis that masks are useful despite their findings.
Health authorities (1): The Centres for Disease Control (CDC)
The US CDC, an organisation with a long track-record in infectious disease control and prevention, did not immediately recommend masks at the start of the pandemic. Presumably because of the strong evidence coming from the influenza RCTs. However, they reversed this position in April 2020, perhaps because of emerging evidence (or political pressure). In May, 2021, the CDC updated a briefing document explaining what they saw as the scientific support underlying their mask advice. The document concerns itself with cloth masks.
They draw on several published studies, none of them RCTs, to support their position.
The first study they addressed was of 2 US hair stylists who became symptomatic with C19. They saw 139 clients. Both the stylists and the clients were masked. The investigators rang the clients, and about half (67) consented to be interviewed. None of those that consented had developed symptoms. The bias is that the half that didn’t consent to an interview may have included symptomatic cases. We also don’t know what the outcome might have been without masks. Furthermore, anecdotal data from these hair stylists doesn’t easily generalise to broad community settings (e.g. from a hair salon to a restaurant or a footpath).
Other studies cited, that I won’t go into in detail (on the basis that giving priority to a study of 2 hair stylists is bad enough), include a study of 124 Beijing households, a retrospective survey of masks as part of contact tracing investigations, a study of an outbreak aboard the aircraft carrier USS Theodore Roosevelt, and investigations involving infected passengers aboard flights longer than 10 hours.
While none of these studies was a RCT, the briefing document goes on to disparage the two RCTs that I described earlier.All studies have their limitations of course, however their arguments show bias. For example the Denmark study is dismissed for only enrolling 0.1% of the population. It’s ironic that 2 hair stylists represent 0.0000006% of the US population. Likewise one aircraft carrier of heathy individuals is not a substantial or balanced representation of the general US population. I will leave it to the reader to calculate what percentage 124 Being households make up of the China population.
The cloth-mask study was also dismissed for various reasons, including “the risk of infection from self-washing [cloth masks]”. Which of course, is what the general public are obliged to do. They then refer to a “follow up study in 2020” that found no increased risk for the healthcare workers who had their cloth masks laundered by the hospital. This was not a follow up study. Rather it was the same authors re-analysing the data they had collected in their first study, 5 years earlier. Presumably they were under pressure to re-examine and reverse their findings. They identified 18 participants (out of ~2,000) who had their masks laundered by the hospital, and the frequency of infection was not greater for them than the surgical mask groups. Even if this finding could be shown to hold up with meaningful participant numbers, the point is that hospital laundering is likely to be more sanitised than possible in the home (e.g. higher water temperatures or chemical agents), and it is not an option for the general public. The potential still remains for reusable cloth masks to be hazardous in the real world. Nonetheless, the CDC used this study to dismiss the original findings.
Overall, this briefing document is a concern for the reputation of the CDC, which has been painstakingly built up since 1946. The document should be read as a political document, and previous CDC directors have expressed their disquiet over undermining the scientific credibility of the CDC for political purposes. Dr Richard Besser, a former CDC director: “I find it concerning that the [current] CDC director has not been outspoken when there have been instances of clear political interference in the interpretation of science.” He’s not alone.
Health authorities (2): World Health Organisation (WHO)
In the first 6 months of C19 the WHO also recommended against the need for, or benefit of, face masks, presumably informed to a large extent by the influenza studies or pre-C19 pandemic planning. The WHO held out longer than the CDC, however in June 2020 they changed their advice and recommended that the general public should wear cloth masks where there was widespread transmission and where physical distancing was not possible. Note the caveats. Concern over asymptomatic transmission was given as the justification. It could also have been the WHO was overwhelmed by unilateral action taken by governments (that fund WHO) to mandate masks.
We still have indoor mask mandates where I live. We don’t have mandates around the quality of the masks, how well they are fitted, how the wearer breathes through or around them, or how masks are handled both on and off the face, all of which are considerations for effectiveness. Mask wearing is more about theatre than effectiveness.
There’s a phrase I have heard repeatedly from authorities during this period — “we’re just following the science”. There’s a weasel word in that phrase, can you identify it?
It’s the word ‘the’. There is no ‘the science’. not even any science of significance. Science is an incomplete, often controversial and evolving human activity. I accept that health policymakers work with uncertain and imperfect information and make their decisions on the balance of probability. Still, I would have hoped for stronger evidence to justify mandating masks rather than simply recommending them.
Analysis of 10 RCTs for influenza. This was published in May 2020 in the journal Emerging Infectious Diseases, which perhaps ironically is a journal of the CDC.
Cloth mask vs. surgical mask RCT (2015) and the second data analysis of the 18 who used the hospital laundry (2020).
CDC briefing document (May 2021). This document was used for this post. However, I am now aware of a December 2021 update. The 2 hairstylist study has been moved down the presumed ranking list. Now top of the list is a study from Bangladesh in which some villages were given masks and some not. The difference in infection rates between villages was negligible, but when expressed as a relative risk could be made to appear meaningful.