June 12, 2020

expert reaction to a study looking at mandatory face masks and number of COVID-19 infections in New York, Wuhan and Italy

A study, published in PNAS, looked at mandatory face masks and number of COVID-19 infections in New York, Wuhan and Italy.

Prof Paul Hunter, Professor of Medicine, UEA, said:

“In my opinion the study “Identifying airborne transmission as the dominant route for the spread of COVID-19” by Zhanga and colleagues is deeply flawed and its conclusions highly suspect.

My main criticisms are that 

1. the analysis was very simplistic fitting just a simple linear regression model to cumulative case numbers. Outbreaks never follow such a straight line for very long and it was inevitable that the curve would fall below the line eventually.

2. The authors have included no control countries/states so that they did not in any way check that such slackening of increase was due to facemasks or something else or consistent across multiple countries whether or not facemasks were being used.

3. The authors do not appear to account for the lag between date of infection and reporting. Typically there is a lag about a week or more for people to develop symptoms, and then for those symptomatic people to seek a test and then for the test to take a couple of days for that test to be reported on and positive results to be included in the summary figures. It is clear from figure 2 in the paper, that case numbers were declining before facemasks were recommended and certainly before facemasks use could possibly have influenced the trajectory of the outbreak in New York.

“This last point in particular severely undermines the author’s conclusions.

“So in conclusion, this paper does not provide any convincing evidence or analysis behind its suggestions that facemasks reduce disease by 40% and its conclusions are almost certainly false. The authors finish their abstract with the sentence “Our work also highlights the fact that sound science is essential in decision-making for the current and future public health pandemics”. I agree but this paper does not represent sound science in my view.”

Dr Louise Dyson, Assistant Professor in Epidemiology, University of Warwick said:

“This study has major methodological flaws which undermine the conclusions made. The authors observe a reduction in case numbers from the day of mandating face masks be used. Since it takes approximately 5 days for an infected person to become symptomatic, any changes in transmission will not be seen in case data for at least that long.”

Dr Mike Tildesley, Associate Professor, University of Warwick, said:

“The authors fit a straight line to the data and observe that case numbers are reduced compared to this linear trend. However epidemiological curves display exponential growth and decay, rather than a linear trend. Fitting a linear trend to any part of the decay curve will result in the phenomenon seen.”

Dr Ed Hill, Research Fellow in Mathematical Epidemiology, University of Warwick, said:

“Face masks may be an effective tool for reducing transmission, but this study does not produce robust evidence to support that. Since it is likely that a range of (mandated and voluntary) behavioural changes will influence ongoing transmission, further research is needed to adequately untangle these different effects.”

Prof Jonathan Reid, Professor of Physical Chemistry, University of Bristol, said:

“Although the paper explores some important correlations between implementing physical distancing guidance, the wearing of face coverings and reductions in rates of infections, to definitively identify causation is much more challenging.

“We know that face coverings can reduce the transmission of large droplets from someone when they sneeze and cough and could reduce the exposure of someone passing by. Although respirator masks clearly reduce exposure, the value of commonly used non-medical face coverings in reducing exposure to small respirable aerosol particles is much less clear. Although the viral RNA fingerprint of the coronavirus has been seen in airborne samples, no one has yet been able to detect infectious virus from aerosol particles and, as the authors recognise, this really must be a priority to establish airborne transmission as an important route or the dominant route. It is also remains unclear how long the virus survives while airborne, how the survival compares outdoors with indoors, and to what dose someone needs to be exposed to become infected. Answers are urgently needed to these questions to provide the scientific evidence base that the authors say is essential.”

Prof Patrick Royall, Professor of Chemical Physics, University of Bristol, said:

“The paper by Zhang et al analyses data from three epicentres (Wuhan, New York and Italy) and pinpoints the point at which the rate of increase of confirmed infections begins to fall as the point at which face covering is made mandatory. This is shown in Fig. 2 of the paper, for the latter two epicentres (it’s hard to assess Wuhan as face coverings were worn from the very start). In NYC and Italy, face coverings were only mandated some weeks into the outbreak enabling the rate of increase of infections to be assessed before and after the mandatory use of face coverings. So in short the rate of infections began to slow when face coverings became mandatory.

“The paper goes on to show (Fig. 3) that social distancing and stay-at-home measures can by themselves, be insufficient to lead to a drop in infection rate, though clearly this will be related to compliance by the population in question (the data is quoted from the US).

“It is consistent with other evidence as far I am aware and the work is sound. I do not see any overspeculation here, the statement in the conclusions

‘We conclude that wearing of face masks in public corresponds to the most effective means to prevent interhuman transmission, and this inexpensive practice, in conjunction with extensive testing, quarantine, and contact tracking, poses the most probable fighting opportunity to stop the COVID-19 pandemic, prior to the development of a vaccine.’

I think is a fair assessment of the situation.”

Prof Trish Greenhalgh FMedSci, Professor of Primary Care Health Services, University of Oxford, has revised her original quote as follows:

“I’ve revised my original comments on this paper in the light of expert comments from specialists in the techniques used. The authors conducted statistical analysis of transmission rates in natural experiments in four settings, all of which had high incidence of COVID-19: Wuhan, China (where several population measures, including aggressive contact tracing, quarantining, social distancing and mandatory face coverings, were all introduced in parallel); Lombardy, Italy (where mandatory face coverings was introduced 3 weeks after social distancing had failed to stem a rapidly-escalating local outbreak); New York City (where social distancing was introduced a month before mandatory face coverings); and the rest of USA (where social distancing but not face coverings was the norm). In every case, mandatory face coverings led to a rapid reduction in transmission – except in the rest of USA where transmission rates remain high.

“The analysis appears to support the hypothesis that you can catch COVID-19 by airborne transmission (inhaling viral particles directly from the air) as well as by contact (e.g. touching surfaces such as doorknobs) and droplets (e.g. via a wet cough or sneeze). However, there are several flaws in the methods used which should make us cautious about the study’s findings. First, the key statistical method used (linear regression) was inadequate to deal with the potential effects of confounding factors. Second, the authors’ conclusions appear to depend on selecting a particular set of dates to begin and end the analysis; if the conclusions were robust they should hold across different time periods. Finally, the decrease in infection rate in some of the case studies appears to date from the date the policy was introduced, which is implausible since SARS-CoV-2 takes several days to produce illness in a contact.

“In sum, we cannot confidently conclude anything from this paper. It has not convincingly confirmed that the fall in transmission in the case sites was due to the introduction of masking. Whilst the findings are consistent with such a conclusion, there may be other explanations.”

and below in the original roundup:

Prof Trish Greenhalgh FMedSci, Professor of Primary Care Health Services, University of Oxford, said:

“This study appears to have been well-conducted using appropriate methods. The authors conducted statistical analysis of transmission rates in natural experiments in four settings, all of which had high incidence of COVID-19: Wuhan, China (where several population measures, including aggressive contact tracing, quarantining, social distancing and mandatory face coverings, were all introduced in parallel); Lombardy, Italy (where mandatory face coverings was introduced 3 weeks after social distancing had failed to stem a rapidly-escalating local outbreak); New York City (where social distancing was introduced a month before mandatory face coverings); and the rest of USA (where social distancing but not face coverings was the norm). In every case, mandatory face coverings led to a rapid reduction in transmission – except in the rest of USA where transmission rates remain high.

“The study is important because the analysis strongly supports the hypothesis that you can catch COVID-19 by airborne transmission (inhaling viral particles directly from the air) as well as by contact (e.g. touching surfaces such as doorknobs) and droplets (e.g. via a wet cough or sneeze). This is a new and policy-critical finding. Washing hands protects against contact transmission. Social distancing protects against droplet transmission and will help reduce airborne transmission, especially out of doors. But the only measure to provide substantial protection against airborne transmission, especially indoors in poorly-ventilated settings, is covering the face so that viral-laden droplets are stopped before the virus becomes airborne. Face coverings will also help reduce contact transmission, since droplets will be less likely to contaminate surfaces.

“This study lends support to the message I’ve been posting for the past two months:

COVER YOUR FACE
IN A PUBLIC PLACE
IT’S EVIDENCE-BASED
IT MAKES US SAFE”

Prof Rowland Kao, Professor of Veterinary Epidemiology and Data Science, University of Edinburgh, said:

“By comparing trends in infection rates in different geographical locations, this study shows that, after mandatory introduction of face masks there is a statistically significant deviation from the trend prior to adoption. This suggests that the masks cause a substantial reduction in transmission. This result is in line with previous experimental evidence of the potential role of face masks, however it must be remembered that the relationships compared do not directly take into account the potentially substantial nonlinearities associated with transmission patterns, and as it is a statistical model – establishing correlation, not causation. Thus there may be other factors that influence this trend, such as behavioural changes that may be related to the use of masks. Despite these caution not to take the paper on its own as concrete evidence of face mask efficacy, it nevertheless adds to the growing body of evidence that face masks can play an important role in preventing COVID-19 spread”

Commenting on both the PNAS paper and a study from the German Institute of Labor Economics:

Prof Peter Openshaw, Professor of Experimental Medicine, Imperial College London, said:

“These are interesting and potentially important studies, but it’s important to recognise that they are not controlled interventions that prove once and for all that masks are the key to controlling spread. They are observational and retrospective- looking back over the evidence and trying to work out what causes effects. However, they fit with the emerging direct evidence that face coverings help reduce spread of coronavirus.

“The study from the Institute of Labor Economics (supported by the Deutsche Post Foundation) concludes that introduction of face masks on 6 April reduced new infections by almost 25% over next 20 days. Comparing the daily growth rate in control group, the rate in the Jena region shrank by around 60%. In larger cities, reduction in the growth rate due to introduction of masks is around 40%.

“The study from the USA examined mitigation measures in Wuhan, China, Italy, and New York City to identify impacts of various mitigation measures. Mandated face coverings appeared to have major effects in the three instances. They estimate that face masks prevented 78,000 transmissions in Italy from April 6 to May 9 and over 66,000 in New York City from April 17 to May 9. Other mitigation measures (e.g. social distancing, track and trace) are insufficient unless combined with face mask wearing. They conclude that wearing face masks in public is the most effective single means to prevent transmission, but should be combined with other methods to prevent spread.

“These findings and conclusions seem plausible and have significant implications for public health, reinforcing the trend towards mask wearing even by those who feel well and have no symptoms.

‘Identifying airborne transmission as the dominant route for the spread of COVID-19’ by Zhang et al was published in PNAS on Thursday 11^th June 2020.

All our previous output on this subject can be seen at this weblink: www.sciencemediacentre.org/tag/covid-19

Conflicts of interest:

None received.