There has been a lot of controversy over whether face masks decrease the spread of respiratory infections during the covid-19 pandemic. And what is the most sensible thing to do when a topic is controversial? Look at what the evidence says!
An umbrella systematic review (a systematic review of systematic reviews) was published in Canadian Family Physician in July looking to answer this question. It included 11 systematic reviews, which were in turn based on 18 randomized controlled trials, with a total of 26,444 participants. The authors declared no conflicts of interest.
These trials were all carried out before the covid-era, so they were looking at other respiratory viruses, which is just something to be aware of, although covid does not appear to be markedly different from other respiratory viruses in terms of how it spreads or how infectious it is, so it should be possible to generalize from these studies. The data all come from randomized controlled trials, the highest quality type of evidence we have, and should therefore be more scientifically valid than the purely observational data that has often been used to justify mask use during the last few months. The problem with observational data is, as has been discussed before on this blog, that there is a huge scope for confounding effects. For example, people who choose to wear masks are probably taking more precautions in other ways than people who don’t, which will tend to make masks look more effective than they are.
Six of the trials looked at the use of face masks to prevent respiratory infections in the hospital, while the remaining twelve looked at the use of face masks in the community setting. We will begin by looking at the six hospital trials. Only one of them looked at masks vs no masks. The rest looked at different types of mask (mostly N95 masks vs surgical masks, although one also looked at cloth masks). It was always the hospital staff wearing the masks, not the patients, and the purpose was to see if the masks had any effect on probability of developing a respiratory infection. When the staff were supposed to wear the masks varied slightly – some of the studies mandated continuous use throughout the work shift, while others only required that the masks be worn when within six feet of patients or when caring for patients with respiratory infections. In general, around 60% to 80% of participants wore the masks as directed. This could of course be a problem, since it could make the results appear weaker if not everyone is following the study protocol, but at the same time probably makes the results more realistic, since not everyone is going to do as told in the real world either.
The one study that looked at masks vs no masks in the hospital consisted of 32 participants who were followed for 77 days. In all, one participant in the face mask group developed a respiratory infection, and one participant in the control group developed a respiratory infection, so there was no difference between the groupd. However, the study was so small that it’s really impossible to draw any conclusions from it.
Then we have four studies that were comparing N95-masks with surgical masks. These studies had a total of 8,712 participants, which is a big enough number that it should be able to find a difference if there is one. Overall, 45% of participants in the N95 group developed respiratory infections, compared with 55% in the surgical face mask group. This is an absolute difference of 10% , which isn’t too bad if the effect is real, but the difference wasn’t statistically significant.
What can we conclude? N95 masks seem to reduce the frequency of respiratory infections slightly more when compared with surgical masks, although the difference could also just be the result of chance.
Finally we can look at the one study that compared surgical masks with cloth masks. This study had 1,607 participants, of which 580 were in the surgical mask group, 569 people were in the cloth mask group, and 458 were controls. Unfortunately “control” in this study didn’t mean no mask, but rather that people in the control group could do whatever they felt like, sometimes wearing a cloth mask, sometimes wearing a surgical mask, sometimes wearing an N95, and sometimes wearing no mask, so the control group doesn’t really help us to understand anything. What were the researchers thinking?!
The people in the surgical mask group were given two new masks at the beginning of each shift, while the people in the cloth mask group were given five cloth masks to use for the full duration of the study, and asked to wash them with soap and water at the end of every shift.
At the end of the trial, 7,6% of people in the cloth mask group had develop a respiratory infection, compared with 7,0% in the control group, and 4,8% in the surgical mask group. As explained above, the control group in this study doesn’t tell us anything. The difference in performance between the surgical masks and the cloth masks was statistically significant however, and the absolute reduction was 2,8%, so surgical masks definitely seem to be better than cloth masks.
Ok, so what conclusion can we draw so far? N-95’s are possibly better than surgical masks, and surgical masks are probably better than cloth masks. Whether cloth masks are better than nothing (or for that matter, worse than nothing) is infortunately something we don’t know from these studies, since none of the researchers thought it would be a good idea to have a control group that wasn’t wearing any masks. Doh!
Anyway, let’s get to the twelve studies looking at how effective masks are in a community setting. That is after all what matters most to all the people who don’t spend their days in a hospital. Seven of these studies, with a total of 5,535 participants, were looking at families in which one person had a respiratory infection. In some of them the sick person wore a mask, in some of them the other family members wore a mask, and in some of them, everybody wore masks.
Unfortunately, it didn’t seem to matter who was wearing the mask, none of these studies found any difference in rate of infection between those households in which people were wearing a face mask and the households in which noone was wearing a mask. One explanation could be the low rate of adherence. Only 30-50% of participants were wearing the masks as directed, which I guess is understandable. People want to be able to relax when they’re in their own homes, and they want to be intimate with sick loved ones. Wearing a mask in your own home fits badly with thise priorities. Another explanation could be that if you’re spending hours per day in close proximity to a highly infectious sick person, the fact that you’re wearing a mask, or that they’re wearing a mask, probably isn’t going to make much difference. So all in all, these seven studies don’t tell us that much, but they suggest that there isn’t any point in anyone wearing a mask at home when a member of the household is sick.
I’ve intentionally left the best for last. Of all the studies in the review, there were two that looked at healthy university students in dorm rooms during influenza season. The two studies both lasted for six weeks and included 1,683 people. 765 were directed to wear surgical masks as much as possible, and the other 918 were a control group that didn’t wear masks at all. In practice, “as much as possible” meant four hours per day in one study, and five hours per day in the other study. This isn’t great, but I guess it’s hard to get people to wear masks more than that. The reason I think these studies are “best” is because they are the ones most relevant to the covid pandemic, where healthy people in a non-hospital setting are being told to wear masks as much as possible.
So, what were the results? Overall 18,8% in the mask group became sick, compared with 24,7% in the control group. This is a 5,9% absolute difference in favour of wearing masks. After correcting for the fact that these trials were using cluster randomization (a method in which an entire group, for example an entire dorm room, is randomized to a treatment instead of randomizing individuals), the reviewers determined that the actual absolute difference was more like 4,2%, which gives a relative risk reduction of around 17% (24,7/4,2). This would mean that for every 24 people who wear a surgical mask, you prevent one infection (100/4,2), or to put it another way, you prevent one in six infections. The result was statistically significant, just.
Ok, so what conclusions can we draw from all these studies?
First of all, when it comes to preventing the spread of respiratory infections, N-95 masks might be better than surgical face masks, and surgical face masks are probably better than cloth masks. In fact, cloth masks may not provide any protection at all! So if you’re going to wear a mask, wear a surgical mask or an N-95.
Secondly, if you or someone in your household is sick, you probably don’t need to bother wearing a mask at home. The infection will spread at the same rate within the household regardless. If there is a member of a high risk group living in the household, i.e. someone over the age of 70 with serious co-morbidities (and that individual isn’t the one who is sick), then it might make sense for either that person to spend the next week somewhere else, or for the sick individual to do so.
Thirdly, face masks do seem to slightly decrease the risk of spreading respiratory infections outside the household setting. However, it is questionable whether an intervention that only impacts one in 24 people, and that only decreases the relative risk of infection by 17%, is having a big enough effect to noticeably slow the speed at which a highly infectious disease like covid-19 spreads through a population.
Rather than require that everyone wear a mask at all times when out in public, it might make more sense to restrict mask use to specific situations, for example when interacting with high risk groups in nursing homes (and in those situations to combine the masks with face shields to create a maximally impervious barrier), especially considering that for people below the age of 70 who are otherwise healthy, the risks connected with covid are tiny.
You might also like my article about whether hydroxychloroquine has a role in treating covid-19, or my article about whether vitamin D supplements protect against respiratory infections.