What’s the best way to protect yourself when you’re at risk of exposure to SARS-CoV-2? It seems like a simple question, but many of the options—face masks, lockdowns, social distancing, etc.—have been politically controversial. In addition, it has been difficult for public health authorities to maintain a consistent message, given our changing state of knowledge and their need to balance things like maintaining supplies of protective equipment for health care workers.
But several months into the pandemic, we’ve started to get a clear indication that social isolation rules are helping, providing support for those policies. So, where do we stand on the use of masks?
Two recent events hint at where the evidence is running. The first involves the retraction of a paper that appeared to show that mask use was ineffective. And the second is a meta-analysis of all recent studies on the use of protective gear against SARS-CoV-2 and its relatives SARS and MERS. It finds support for a protective effect of masks—as well as eye protection—although the underlying evidence isn’t as strong as we might like.
So, how do you test that?
It turns out that testing the effectiveness of masks is harder than expected. A recent study in the Annals of Internal Medicine appeared to be the sort of well-designed experiment that you might think would be decisive. The researchers took patients with a SARS-CoV-2 infection, put masks on them, asked them to cough, and collected any material that passed through the masks.
The paper had concluded that all masks were ineffective, but it has since been retracted, as the authors failed to account for the sensitivity of the equipment they used to detect the virus. (Retraction Watch has more details.) It’s also notable that the paper has only four infected individuals and no control coughers, so it shouldn’t have been viewed as decisive anyway. But, in an environment where there’s so little quality information, the study had already appeared in dozens of news reports.
To get around the issue of small, underpowered studies like this, the World Health Organization asked a group of researchers at McMaster University to undertake an exhaustive review of the medical literature. The team included studies of the related coronaviruses SARS and MERS, as many studies had been completed with these earlier viruses.
But even with these standards, the researchers struggled to find detailed studies of the use of protective gear. Despite identifying results from a total of over 25,000 individuals involved in various studies, there were no randomized controlled trials among the studies they identified. A few of the studies didn’t even use the WHO’s standards of determining who ended up infected.
So, while a meta-analysis can provide a better sense of what’s going on even though it relies on smaller studies that might be inconclusive on their own, it’s important to acknowledge that the starting material here isn’t exactly high-quality.
All told, the authors found 172 observational studies that looked at issues related to the prevention of coronavirus transmission. Sixty-six of these focused on the distance at which virus could be transmitted, thus providing information on social-distancing effectiveness. Another 30 looked at different types of face masks; 13 focused specifically on eye protection. Others either looked at multiple issues or didn’t address any of the protective measures focused on here. Fewer than 10 of these studies looked at COVID-19 cases; the rest focused on SARS or MERS, caused by related coronaviruses.
For the effects of distancing on transmission, the underlying studies used various measures of distance and infection. The authors accounted for this by running over 10,000 randomized models to determine what was needed to produce the results of earlier papers. These indicated that there was strong evidence that staying at least a meter away from infected individuals provided significant protection. There was weaker evidence that even greater distancing was more effective.
Overall, this is in line with what we’re learning at the population levels, where there’s strong evidence that various social-distancing rules are effective.
For face masks, the researchers found that the overall protective effect appeared substantial, but the underlying evidence was weak. Putting that differently, the data is consistent with a variety of possible degrees of protection, but the most likely answer is that masks are very protective. Part of the reason for this is that N95 masks provide superior protection to multi-layered masks, which do better than single-layered masks.
This also influenced the results regarding the context of where the masks were effective. Since medical workers had greater access to N95 masks, face mask use appeared to be more effective there. But if this was adjusted for, then mask use by the public also appeared to be protective. Given the severe shortages in N95 masks in many locations, however, it’s not clear when the public would be able to use this information for their protection.
The final piece of protective equipment they look at is eyewear, which also reduced coronavirus transmission. This is something that hasn’t been emphasized much, at least once medical workers got sufficient access to face shields. But eye protection is something that a lot of the public probably already has access to.
Not the final word
The study has some obvious limitations: it’s trying to integrate a huge amount of individual bits of research that may use different methods and measures of success. One thing that the authors acknowledge failing to account for is any measure of the duration of exposure, which will undoubtedly influence the effectiveness of different forms of protection. They also acknowledge that the context of exposure—such as in hospitals or public transit—may influence the effectiveness of different forms of protection.
The other big issue is that it treats all three of the major emerging coronaviruses—SARS, MERS, and SARS-CoV-2—as equivalent. While they’re all members of the same family of coronaviruses (the beta coronaviruses), they differ in substantial ways. Most notably, one of them latches on to a completely different protein on cells in order to gain entry and start an infection.
Despite these limitations, it’ll probably be a while until we get better information, given that the focus has been on providing any protection possible in the early stages of the outbreak. And at least some of the relevant factors, like how well protection limits the spread of particles expelled in a cough, aren’t specific to any one virus.
But a key fact noted by the authors is that none of these means of protection are complete; any form of contact with infected individuals presents a risk. So our focus should be on reducing the overall risk that infections spread, in the hope that we can eventually reach the point where contact tracing and quarantines can severely limit further infections. And, for that purpose, a less effective form of protection used by large numbers of people might have more of an impact than a highly effective form available only to a few.
Figuring that out is beyond the scope of this study. But at least people now have some information to inform the models that could test that.