COVID-19: a commentary on face coverings in public places
There is much misinformation about the wearing of face coverings in the context of the COVID-19 pandemic. I have written this commentary to try to reduce the misinformation and to explain why I am opposed to a legal obligation requiring the general public to wear a face covering in a public place as of July 24th 2020.
Three recent papers (Chu et alia, 2020; Mitze et alia, 2020; Royal Society, 2020) have generated significant publicity in the general press and the medical press and they serve to explain much of the misinformation and the reasons I have reached my conclusions opposing mandatory face coverings.
New evidence might arise in which case I shall be happy to review my conclusions. After all when a genuine expert like John Maynard Keynes is known for his second thoughts (Davenport-Hines, 2015) then I shall be in good company.
My analyses need to be seen in the context of the prime ethical principle of autonomy for the individual and in population health. The guiding principles in my medical and wider practices are based on andragogy and humanism.
I value physical distancing which means being physically apart; one metre is recommended by the World Health Organisation. Sadly poor use of language has meant that social distancing is often used in error. Social distancing is not what we should be seeking to do at all.
In this commentary I have standardised the language by using the term “face covering” instead of “face mask” or some other terms.
1) Chu et alia (2020): this paper was published after peer review in the Lancet online on June 1st 2020.
It was a systematic review and a meta-analysis of physical distancing, face coverings, and eye protection. Focusing on the subject of face coverings in community settings, i.e. outside of health settings, there were no randomised control trials. There were 10 case control studies comparing face coverings with no face coverings and virus infection rates in the pandemics of SARS, MERS, and SARS-CoV-2.
These studies came from diverse countries: China five; Saudi Arabia two; Canada, USA, Vietnam one each, which makes it questionable whether it is appropriate to group them together. The community settings were heterogeneous including household contacts and close contacts as well as one study from an airplane and one from a dormitory which raises further questions about the transferability of the results to public places.
The authors estimated a reduction in the chance of viral transmission of about 14% in those using face coverings. However the authors confirm the grade of certainty about this effect size is low and the true effect could be substantially different from the estimated effect. Moreover the authors identify limitations to the results including recall and measurement biases particularly as much of the evidence comes from cases within households and from the contacts of cases and not from cases in general public settings.
In my opinion this study has been carried out to high standards of systematic review, meta-analysis and critical thinking. The authors recognise some of the disadvantages from face coverings including discomfort, damage to the skin, and difficulties communicating with others. I would add there are increased risks of the virus transferring to the wearer’s fingers through frequent touching of the covering.
I think the 10 studies were too heterogeneous to permit their aggregation and the results provided only weak evidence for the wearing of a face covering in a public place.
2) Mitze et alia (2020): this paper is a discussion paper published by the Institute of Labor Economics (IZA) on about June 12th 2020. The IZA emphasise that discussion papers “often represent preliminary work and are circulated to encourage discussion.” The paper does not appear to have been peer reviewed.
The authors used the synthetic control method to estimate how the use of face coverings affected the incidence of registered infections of COVID-19 in Jena, Germany and the four other regions of Rottweil, Wolfsburg, Nordhausen and Main-Kinsig.
In the context of this paper, the synthetic control method (Abadie, 2020) observes a region where an intervention occurs and compares what happens there with an artificial or synthetic region drawing on data from other matched places.
In essence, the authors studied the region of Jena where the intervention of mandatory face coverings was introduced on April 6th 2020. They compared the incidence of registered infections of COVID-19 in Jena with the incidence in a synthetic Jena derived from data in other places where face coverings were not mandatory.
They also studied four other regions in Germany who mandated face coverings between April 14th and April 20th 2020 and compared the incidence rates with their synthetic regions before the requirement from the German federal government to wear face coverings outside took effect on April 29th 2020.
The synthetic control method could be described crudely as a method to estimate the effect of an intervention, e.g. face coverings, on infection rates in the population in a region and to compare the intervention region with rates in a synthetic control population that was matched to be similar but without experiencing the intervention.
The results were presented in several different ways which can be confusing. However, the results for Jena compared to the synthetic Jena found a reduction of 13% in the cumulative incidence of COVID-19 infections after 10 days and 23% after 20 days.
Other regions compared to their synthetic counterparts showed a mixed picture of benefits/detriments. Thus Rottweil and Wolfsburg regions both had similar results to Jena; Nordhausen had equivocal results, not clearly beneficial or detrimental; and Main-Kinsig showed an increase in infections relative to the synthetic control region, i.e. face coverings appeared detrimental leading to an increase in infections.
Overall the authors found a reduction of between 2.3% and 13% in the cumulative incidence of COVID-19 infections after 10 days.
In my opinion this study has been carried out to high academic standards and the authors recognise the limitations of the synthetic control method. The reasons I am not persuaded about the validity and reliability of the results are –
a) the intervention region of Jena seems to be something of an outlier so it is challenging to identify a good fit with the synthetic control region, e.g. the cumulative incidence rate of infection was 6% higher in Jena than in Germany (cases per 100,000 inhabitants), Jena has the densest population (residents per square kilometre) of all the cities in Thuringia state and four times the density of Germany. Also Jena was an early adopter of mandatory coverings and while some other regions showed similar effects of benefit from face coverings, some regions showed smaller effects and even detriments from face coverings.
b) other events could have affected behaviour and infection rates in the intervention regions compared to the synthetic control regions, e.g. the spread of information about some aspects of COVID-19 with contamination of the regions leading to changes in behaviour. Also there could have been differences in the climate or pollution that affected the behaviour either of the populations in the regions or of the SARS-CoV-2 virus e.g. humidity, ambient temperatures.
On balance I recognise the high quality of the authors’ use of the synthetic control method including the sensitivity analyses seeking to identify the best fit for the synthetic Jena, and to mitigate for spill over effects (Bouttell et alia, 2017). However more detailed research is needed to confirm the results and thereby to inform decisions that could lead to legislation.
3) Royal Society (2020): this paper was published by the Royal Society and the British Academy on June 26th 2020. The publishers note that the paper is a preprint and it has not been subjected to a formal peer review.
Focusing on the subject of face coverings, part of the paper was a meta-analysis of studies that researched infection rates amongst people wearing face coverings in health care settings. The authors identified four case control studies, all from China, comparing face coverings with no face coverings and virus infection rates in the pandemics of SARS and H1N1.
The authors estimated a reduction in the chance of viral infection of about 54% in those wearing face coverings made of cloth and a reduction of about 39% in those wearing paper coverings.
However the authors confirm that these results were from health care settings and the authors hypothesised that the results might translate to community settings where there might be protection for both wearers and people with whom they are in contact. The authors recognised that case control studies can demonstrate associations but not causes.
In my opinion these results and hypotheses are lacking in validity and reliability. The authors did not provide a justification for extrapolating results from health care settings to the community. They cited the paper by Chu et alia (2020) but they did not draw on many of the papers in Chu’s review or from other reputable sources e.g. Mitze et alia (2020). They did not use research from the SARS-CoV-2 pandemic.
I look forward to reading the peer review of this part of the paper.
The section of the paper that has some merit is the section on behavioural factors related to the use of face coverings. There is a much better literature review which includes papers from experiences with COVID-19 and there is useful guidance about public health messages and building trust in the sources of those messages.
Overall these three oft-cited papers do not provide persuasive evidence for a legal obligation on the general public to wear a face covering in a public place in my opinion.
The observational studies cited here, both case control studies and synthetic control studies, observe situations and data and they show associations but they do not show cause and effect.
Some people may choose to wear a face covering because they believe it will protect them from infection by SARS-CoV-2 or they believe it will protect others from being infected by SARS-CoV-2 by them despite the weak evidence for protection in either direction.
I can see the allure of making a legal obligation that requires people to wear a face covering in a public place i.e. for face coverings to be mandatory. It would be a reflection of the disciplinary society described by Horton (2020) drawing on Michel Foucault’s work. However the disciplinary society is in opposition to individual autonomy and humanism and without appropriate safeguards especially transparency about information and decision-making (Horton, 2020), I reject it.
For some people wearing a face covering could be a form of virtue signalling whereby the wearers demonstrates their virtue in protecting others from the SARS-CoV-2 even though the wearers do not believe they are carrying the virus or that coverings are effective.
For other people, particularly some of the reactionary doctors who relish telling patients and the public what are best for them, requiring others to wear a face covering suits their paternalistic view of the world (Gawande, 2014).
A government goes too far when it mandates face coverings and the medical profession has no right to support mandatory face coverings.
For my part, in the context of the SARS-CoV-2 and the COVID-19 disease, I advocate recognising our responsibilities and rights, including respect for the autonomous choices others make. Hence I recommend –
- isolating if you have symptoms
- quarantining if you have had a high risk of exposure to the virus
- shielding based on your level of vulnerability
- physical distancing and not social distancing
- hand washing.
Abadie, A. (2020). Using synthetic controls: feasibility, data requirements, and methodological aspects. Journal Economic Literature Retrieved July 21st 2020, from https://economics.mit.edu/files/17847
Bouttell, J., et alia. (2018). Synthetic control methodology as a tool for evaluating population-level health interventions. Journal Epidemiology Community Health, 72, 672–638
Chu, D.K., et alia. (2020). Physical distancing, face masks, and eye protection to prevent person-to-person transmission of SARS-CoV-2 and COVID-19: a systematic review and meta-analysis. Lancet https://doi.org/10.1016/S0140-6736(20)31142-9
Davenport-Hines, R. (2015). The seven lives of John Maynard Keynes. London, UK: William Collins
Federal Statistics Office. (2020). Retrieved July 22nd 2020, from https://www.destatis.de/EN/Home/_node.html
Gawande, A. (2014). Being mortal. London, UK: Profile Books
Horton, R. (2020). The COVID-19 catastrophe. Cambridge, UK: Polity Press
Mitze, T., et alia. (2020). Face masks considerably reduce COVID-19 cases in Germany: a synthetic control method approach. Retrieved July 8th 2020, from http://ftp.iza.org/dp13319.pdf
Royal Society & British Academy. (2020) Face masks and coverings for the general public: behavioural knowledge, effectiveness of cloth coverings and public messaging. Retrieved July 9th 2020, from https://royalsociety.org/-/media/policy/projects/set-c/set-c-facemasks.pdf
Pingback: COVID-19 and trust | Dr Iain J Robbé