Masks Revisited


Science-Based Medicine

A recent Cochrane review, limited in scope and problematic in methodology, does not show that masks do not work, despite common misreporting.

Steven Novella on February 15, 2023

This is an issue that will not go away – do masks and mask mandates work to reduce the spread of COVID? It’s actually a good thing that the question continues to be researched. This is actually a difficult question to pin down definitively, so more research is always better. What is unfortunate is the degree the question has been politicized. One’s answer to this question is less likely to be determined by the science than your political affiliation.

A recent review of studies has rekindled the debate, and is making the rounds within the political right media. Anti-maskers are treating the study as if it is the final word in the debate – but it isn’t. The study is a Cochrane review, which does give it some gravitas, but it has important limitations – specifically in the kinds of evidence that is reviewed.

There are different ways to address the question – do masks work to reduce the spread of respiratory viruses in general and COVID-19 specifically? I reviewed these types of evidence here, but will summarize quickly. We can test whether or not different kinds of masks reduce the spread of respiratory droplets, which are the known primary mechanism by which COVID is spread. The answer is yes, for surgical masks and N95 masks, and equivocal for cloth masks. So face masks do what they are supposed to do, and better masks do it better.

Do they work out in the community when people wear them? This question has a lot of potential confounding variables, and this is why there are so many different results from research. One factor is – how much is the infection being studied spreading in the community. Studies looking at mask wearing during low risk situations do not find any statistical benefit (likely because the baseline risk is too low), but during high risk situations they are more likely to find a benefit and the magnitude and statistical significance of that benefit will be greater.

Also, how are we measuring infections? Do we simply allow people to self-report, do we check public health records, or do we only count laboratory-confirmed infections? Studies can also focus on individual or on communities. They can be controlled or observational. They can also follow an intention-to-treat model, looking not specifically at whether masks work but whether mask mandates or other public interventions work. These are actually different questions. Masks may work when worn properly, but mask interventions fail due to low compliance (people wearing the wrong masks, or wearing them incorrectly, or not wearing them when they are supposed to). Further, if the people in the population being studied are already taking preventive precautions, this will dilute out the effect of the intervention. You can’t force people to not wear a mask or to expose themselves to infection. You can only encourage them to do so or make it easier vs doing nothing.

It’s worth pointing out that all of these potential confounding factors decrease the observed effect of masks and mask interventions, with one exception. As one systematic review concludes, controlled studies can only create a false negative for the efficacy of mask-wearing, not a false positive, and they find that such studies actually underestimate the effect of mask wearing. However, observational studies may overestimate the efficacy of mask wearing in one way – mask wearers may engage in other protective activities, such as avoiding indoor crowds. However, again this depends on your question. If the question is, “does wearing a specific type of mask in a specific way reduce the risk of infection?” observational studies may overestimate that effect. But if your question is, “do mask mandates work?” then it doesn’t matter. If mandating mask wearing makes people engage in other good hygiene and protective behavior, and it total they reduce spread, that’s a good thing. And in fact the evidence is strongest for this effect.

With this background it is also important to note that there is no one perfect study that addresses this question. There are only studies with different strengths and weaknesses and different trade-offs.

For example, one of the bigger trials was the DANMASK trial, which was a randomized controlled trial with statistically negative results (although a trend of benefit). But this study has multiple limitations – the study was only powered to see a 50% reduction in infection. The rate in the community was already very low. Compliance with mask wearing in that group was <50%. Infections were self-reported. The dropout rate was high – 16%. But anti-maskers, or uninformed journalists, can point to this study and say it shows masks don’t work.

If we look at all the evidence we do see conflicting results, but on the whole there is a good signal that proper mask wearing is effective at reducing the risk of infections in relatively high risk environments. As I wrote previously:

Epidemiological studies provide perhaps the best evidence for the efficacy of masks, or more specifically, mask policies. One study comparing mask-wearing policies in different states in the US found that after states initiated mask-wearing policies, the spread of the virus decreased. Another study found that the risk of a spreading event was far greater in “mask-off” social settings than “mask-on” settings. Comparing countries with different mask-wearing policies also finds a good correlation with reduced spread.

Further, during social distancing and mask wearing policies, the rate of all respiratory infections plummeted. We basically had no flu season for the last two years, and now that policies have eased, the flu is coming back. Of course, this effect is not specific to masks and includes all precautions. But this data unequivocally shows that together these precautions work.

Let’s take a look at the recent Cochrane review. This is a review only of controlled studies – the kind of study that is most likely to underestimate the effect of mask-wearing. They did not included other kinds of studies. Not all of the studies reviewed were of COVID – they included studies of other respiratory viruses (which therefore may be different) and they included studies that were not taking place in the middle of a pandemic. This is a huge red flag, as it is pretty clear from existing data that masks only work in high-risk situations.

The one study they included of health professionals in a high risk setting only compared surgical masks with N95 masks, which showed no statistical difference, but did not compare either to no mask. Also, health care professionals are the most likely to engage in the full range of protective behavior (distancing, hand washing, protective face shields, gloves and gowns). Given this, it is not surprise that the difference between a surgical mask and an N95 mask was not statistically significant.

There are other criticisms of this review as well. The studies that took place in a hospital setting included part-time mask wearing, only when with patients. But the data shows that mask wearing only works when you where them continuously when in the hospital, not just in the patient room. They combined data from different settings and with different interventions. They also did not consider the effect of the infected person wearing a mask, only the effect on the wearer. But we know that masks are most effective when everyone wears them – both the giver and the receiver.

But the biggest error I see in reporting this study is the conclusion that it shows that “mask mandates don’t work”, when that was not even studied. Studies looking specifically at mask mandates show that they do work. The most recent and largest study looking at this concludes:

Our estimates imply that the mean observed level of mask wearing corresponds to a 19% decrease in the reproduction number R. We also assess the robustness of our results in 60 tests spanning 20 sensitivity analyses. In light of these results, policy makers can effectively reduce transmission by intervening to increase mask wearing.

This recent Cochrane review is very limited in scope and is highly problematic in its methods. The most we can conclude from it is that we need better and more relevant controlled trials of mask wearing to more precisely determine its effect on the spread of COVID. But it does not show that mask wearing does not work or that mask policies don’t work. Further, if we look at the totality of the evidence (not just these trials) the best current conclusions are:

Properly wearing face masks when in public during high risk of spread reduces the risk of spread of respiratory viruses in general and COVID specifically.

During a pandemic of a respiratory virus, mask mandates are an effective public health measure.

N95 masks likely offer the best protection, but need to be worn over the mouth and nose to be effective, and need to be worn continuously when in public (not just in targeted situations).

Founder and currently Executive Editor of Science-Based Medicine Steven Novella, MD is an academic clinical neurologist at the Yale University School of Medicine. He is also the host and producer of the popular weekly science podcast, The Skeptics’ Guide to the Universe, and the author of the NeuroLogicaBlog, a daily blog that covers news and issues in neuroscience, but also general science, scientific skepticism, philosophy of science, critical thinking, and the intersection of science with the media and society. Dr. Novella also has produced two courses with The Great Courses, and published a book on critical thinking – also called The Skeptics Guide to the Universe.