Effectiveness of face-masks in controlling COVID-19—and why PCR testing might be cheaper

I recently tweeted that most people in Sydney in New South Wales (NSW), Australia. don’t wear masks and they seem to be doing just fine—zero cases of local transmission in past two weeks despite 200,000 tests. One twitter user was bothered by my nonchalance and asked not unreasonably for links to the evidence …

NSW public health guidelines strongly recommend face-masks in public if physical distancing can’t be maintained, the NSW Premier makes it quite clear she wants people to wear masks, and public transport has plenty of signage asking passengers to wear masks, but most people don’t. Indeed, an informal survey suggests that 80% of most Sydney residents don’t bother. The ones who do tend to be Asian or the elderly. I think for a lot of people in the world, especially in anglophone Western countries outside the DC 20500 zip code, this state of affairs would seem shocking or completely crazy.

So here’s a quick summary of the evidence (and for links to the research, see this nice Nature review):

  1. There is solid evidence that barriers, such as face-masks, will reduce expulsion and inhalation of droplets that could potentially carry SARS-Cov-2 virus 
  2. There is good epidemiological evidence for other respiratory viruses that face-masks reduce transmission, both protecting the wearer as well as reducing onward transmission by infected persons.
  3. There are good small-scale studies that show that masks reduced COVID-19 transmission in specific cases.

As far as I am concerned, this is enough to justify general public health guidelines and to provide a basis for government mandates to require mask wearing in situations where other control measures are not sufficient to rapidly control COVID-19 spread.

Unfortunately, the epidemiological evidence quantifying the impact of face-masks on COVID-19 spread is not as strong. There are some good studies finding significant impacts in specific countries, such as the USA, but there is also evidence of countries, such as many in Latin America, that universal mask mandates have failed to stop rampant spread of the virus. Clearly, problems such as lack of compliance, not wearing masks properly and use of ineffective masks effect impact considerably.

There is one cross-country, ecological study that the Nature review mentions that did find a global impact. However, this study suffers from a number of limitations, including inadequate statistical controls to exclude the impact of other measures, and a focus on mask mandates as opposed to actual face-mask wearing. So I tend to rate its findings as of modest strength.

So bottom line is that masks almost certainly reduce spread, but the effects are probably not very large or enough to substantially slow an epidemic by themselves. 

Which brings me back to Sydney. It’s obvious that the other control measures, basically extensive PCR testing of coughs and colds plus efficient contact tracing, are working quite well, so they probably don’t need the additional security of masks. In the last two weeks, NSW did 200,000 tests and detected zero cases. They do get occasional local cases and some they even classify as cases of community transmission (unlike Sri Lanka), but they really have COVID-19 beaten:

NSW Government COVID-19 dashboard as of Saturday, 23 November 2020

I haven’t seen any modeling of what masks are doing or not doing in Sydney, but there is an official study from Norway that is relevant. Back in June, the Norwegian Institute of Public Health modeled what impact masks would have in Norway, which by European standards has low levels of COVID-19 transmission. They calculated that 200,000 people would need to wear a mask to prevent a single infection each week, and so they recommended against universal face-mask wearing. At very low levels of transmission, masks probably don’t make much difference, and in Sydney where COVID-19 incidence is many times lower than in Norway, they will probably have even less impact. 

So, I can say with some confidence that the lack of universal face-mask wearing in Sydney is not something to worry about, as long as their strategy of extensive PCR testing continues and prevents large outbreaks.

Which leads me to another thought. Mask mandates are not cost-free, and they impose significant costs on poor people, unless like Taiwan or Singapore, the government gives free masks to everyone. 

A back of the envelope calculation suggests to me that our current mask mandate imposes a cost on Sri Lankans of at least Rs 100 million/day (assuming half the population has to spend an average of ten rupees each day to comply). This compares with Rs 60 million/day that the government, i.e., we Sri Lankans, spends on PCR testing. With pooled testing, an extensive effort to routinely PCR test coughs and colds as in Sydney might cost Rs 50 million/day (Rs 2,000 per sample x 25,000 samples/day). So an extensive Sydney style testing programme would probably cost Sri Lankans less than our current mask mandate, and be a lot more effective at stopping large outbreaks.