Confused about masks? It isn’t your fault. So are the experts — they just don’t want you to know it. Weak claims are referenced as though they were airtight assertions, and poor research is misrepresented as though it were slam-dunk evidence. One example of this in practice is the UCSF (University of California, San Francisco) article Still Confused About Masks? Here’s the Science Behind How Face Masks Prevent Coronavirus. Let’s evaluate it.
The article is written by Nina Bai. Ms. Bai is a science writer, not a doctor, epidemiologist, or researcher. That in itself doesn’t necessarily make her writing invalid (I’m not a doctor or epidemiologist, either!), but it’s worth noting that this is an article, not a study, and is heavily reliant on the interpretations of the author and those she’s interviewed. There’s little reason to consider it any more credible a source than this post you’re reading right now. Its value — or lack thereof — will be found in the quality of the content.
The article opens by pointing out what most of us already know — that certain government and/or public health agencies have begun recommending (or mandating) masks, and that this is a reversal from previous recommendations, which is potentially confusing. She goes on to say that “health experts say the evidence is clear that masks can help prevent the spread of COVID-19.”
Regardless of what the “experts” say (which depends in large part on which “experts” you talk to!), this thesis that “the evidence is clear” about the ability of masks to prevent spread is what we want to look at. Does the article solidly make this case?
Why did the CDC change its guidance on wearing masks?
According to the two medical doctors Bai interviewed (professors of medicine, and of epidemiology & biostatistics, respectively, at UCSF):
What may have finally convinced the CDC to change its guidance in favor of masks were rising disease prevalence and a clearer understanding that both pre-symptomatic and asymptomatic transmission are possible – even common. Studies have found that viral load peaks in the days before symptoms begin and that speaking is enough to expel virus-carrying droplets.
It seems to be true that the CDC changed its recommendations based on a belief that pre- or asymptomatic transmission were possible. Notably, not due to increased evidence that masks work. Whether or not asymptomatic individuals can transmit the virus is completely irrelevant if we can’t demonstrate that masks work.
So what evidence does the article provide here? For the claim that pre- and asymptomatic transmission are possible, another non-study article. This article tells us outright that “public health experts don’t know exactly how much spread is caused by asymptomatic or pre-symptomatic patients.” It then goes on to address assumptions and modeling, with little reference to verifiable empirical data.
There is a single study referenced which evaluates real-world data, but it isn’t clear how this Chinese study determined the sources of each individual’s infection in order to compile the data in the first place. If these data are reliable, the claim seems valid, but I would be more comfortable with replicated results in a context where the underlying data collection can be validated. It also cites a study of a skilled nursing facility that relies on assumptions to conclude that “asymptomatic residents…most likely contributed” to the spread.
No reference is provided in the main Still Confused article to support the claim that “viral load peaks in the days before symptoms begin.” The linked article provides some evidence of heavy viral load in the early stages of illness, for purposes of testing, but offers no clear evidence of what the viral load is like in normally-expelled bodily fluids. (In my opinion, this study actually raises more questions about the testing than anything else, but that’s another discussion for another day.)
The linked article, in fact, links, in turn, to a heavily-cited WHO document which, while acknowledging that viral load is often detectable “1-3 days before their symptom onset,” also points out that “detection of viral RNA does not necessarily mean that a person is infectious and able to transmit the virus to another person.” In short, we just don’t know.
They also tell us that “SARS-CoV-2 transmission appears to mainly be spread via droplets and close contact with infected symptomatic cases. In an analysis of 75,465 COVID-19 cases in China, 78-85% of clusters occurred within household settings, suggesting that transmission occurs during close and prolonged contact,” (emphasis mine) and that “respiratory droplet transmission can occur when a person is in close contact (within 1 metre) with an infected person…” These are not the same passing-contact settings where we are being pushed to wear masks. (The WHO recommends masks in public settings where physical distancing is not possible.)
So is asymptomatic or presymptomatic transmission possible? Maybe.
What evidence do we have that wearing a mask is effective in preventing COVID-19?
There are seven total citations in this section, and this is where I want to primarily concentrate because, as previously noted, whether pre- or asymptomatic transmission is possible is irrelevant to the question of whether to wear masks unless we can prove that masks work.
So let’s start from the top of the section.
There are several strands of evidence supporting the efficacy of masks.
This links to a literature review from the PNAS (Proceedings of the National Academy of Sciences) Journal. Given that this is a review and I don’t want this post to end up a rabbit hole of links within links within links, I’m not going to make any attempt to address every single sentence and link within that paper. But I would point out that:
- None of these are randomized controlled trials (RCT) on the use of masks as source control for SARS-CoV-2.
- The studies that addressed the efficacy of masks with other respiratory illnesses widely addressed the use of masks and hand washing. If you’re studying these in combination, it’s impossible to determine how much the mask did or did not contribute.
- The authors gave lip service to “balancing potential harm of cloth masks with additional benefits for concurrent epidemic,” but the only potential harm they addressed to any degree was whether the masks increase risk of infection with COVID-19.
- The ultimate conclusion was that masks should be worn on the basis of the precautionary principle — a conclusion that is inappropriate when there are unaddressed concerns about the safety of masks.
One category of evidence comes from laboratory studies of respiratory droplets and the ability of various masks to block them. An experiment using high-speed video found that hundreds of droplets ranging from 20 to 500 micrometers were generated when saying a simple phrase, but that nearly all these droplets were blocked when the mouth was covered by a damp washcloth. Another study of people who had influenza or the common cold found that wearing a surgical mask significantly reduced the amount of these respiratory viruses emitted in droplets and aerosols.
The first link here is to the video experiment. I would point out, first of all, that a face mask is not a damp washcloth. Second, the experimental images were taken at a distance of only 50-75 mm (less than 3 inches), so this tells us very little about the realistic transmission of droplets at normal distance. (I don’t know about you, but unless I’m kissing someone, I’m not within three inches, even in “close” contact!) Moreover, it tells us nothing about what droplets do or do not get around a mask.
The second link is to the study about the surgical mask. It is technically true that the study founnd that surgical masks reduced the amount of the respiratory viruses emitted. What this article doesn’t report (about the symptomatic patients in the study) is how little virus they detected even without masks.
When describing the limitations of their study, the authors indicate that “among the samples collected without a face mask, we found that the majority of participants with influenza virus and coronavirus infection did not shed detectable virus in respiratory droplets or aerosols….For those who did shed virus in respiratory droplets and aerosols, viral load in both tended to be low. Given the high collection efficiency of the G-II and given that each exhaled breath collection was conducted for 30 min, this might imply that prolonged close contact would be required for transmission to occur, even if transmission was primarily via aerosols…The major limitation of our study was the large proportion of participants with undetectable viral shedding in exhaled breath for each of the viruses studied.” (emphasis mine)
A recent study published in Health Affairs, for example, compared the COVID-19 growth rate before and after mask mandates in 15 states and the District of Columbia. It found that mask mandates led to a slowdown in daily COVID-19 growth rate, which became more apparent over time. The first five days after a mandate, the daily growth rate slowed by 0.9 percentage-points compared to the five days prior to the mandate; at three weeks, the daily growth rate had slowed by 2 percentage-points.
The study this references might be one of the worst I’ve ever seen. It has a number of methodological problems, but the most blatant is the timing. The authors of the study indicate that the date of a mandate was recorded as the date it was first issued (verbalized, essentially), not the date it went into effect.
Now, consider that COVID-19 has an incubation period. People are not being exposed and instantly becoming sick. The CDC estimates that the incubation period is 2-14 days. (According to the next-cited study, the average is 5.1 days.) We should expect, then, that there is a delay between any mitigation measure and a decline in cases. This makes it improbable that a decline within the first five days following a mask mandate is a consequence of the mandate and not merely a pre-existing decline in the curve.
Moreover, given the way the dates were recorded, some mask mandates did not even go into effect until five or more days after they were issued!
Another study looked at coronavirus deaths across 198 countries and found that those with cultural norms or government policies favoring mask-wearing had lower death rates.
There are inherent difficulties with comparing data like this between countries, where reporting methods may vary, and where cultures and policies are widely variant that it’s impossible to control for every variable. Assuming the overall data are fairly accurate, some things are relatively stable traits or behaviors for comparison. The percentage of the population over 60, for instance, or what percentage are smokers, are traits that are likely to remain fairly constant over the course of a year (or an outbreak).
There are greater difficulties, though, when examining responses to an outbreak — like the wearing of masks or a lockdown. These are behaviors that are very likely to be variable over the period of time reported, and which may, to varying degrees, be associated with other factors. For instance, did countries that pushed mask wearing early also social distance early and/or encourage frequent hand washing? Were they the same as the countries that locked down? What percentage of the population was wearing masks and/or locked down for how many days at what point of the outbreak? Were countries that wore masks more or less likely to be nations with heavy use of mass transit? Where were they wearing masks?
I think you get the idea. Another notable fact is that several of the countries with the lowest mortality were those the researchers observed began using masks either before the first known case appeared in their country or within days of the first known case. This raises the question of whether those countries also began physical distancing at this stage (something the paper doesn’t address). This would be significant, because containment of an outbreak is radically more realistic if caught very early than after there’s been significant community spread.
The researchers definitely found correlation here. Whether or not they found causation is considerably less clear.
Two compelling case reports also suggest that masks can prevent transmission in high-risk scenarios, said Chin-Hong and Rutherford. In one case, a man flew from China to Toronto and subsequently tested positive for COVID-19. He had a dry cough and wore a mask on the flight, and all 25 people closest to him on the flight tested negative for COVID-19. In another case, in late May, two hair stylists in Missouri had close contact with 140 clients while sick with COVID-19. Everyone wore a mask and none of the clients tested positive.
People who are knowingly ill should stay home. If they absolutely must go out (catching a flight, for instance, may be unavoidable), wearing a mask certainly makes sense. Clearly-sick people should not be part of the discussion over whether healthy people should wear masks, in the first place.
However, even at that, these case studies tell us nothing about what would have happened if the individuals in question had not worn masks. Perhaps they would have spread the illness. Perhaps not.
An asymptomatic woman was reported as having come into contact with 455 people — unmasked, as far as I’m aware — with no one falling ill as a result. Yet this case study is often brushed off by the pro-mask crowd as “unscientific” because it’s “just a case study.” Either this case study serves as evidence that asymptomatic spread is unlikely, or case studies are invalid for consideration — including the two cited above.
How many people need to wear masks to reduce community transmission?
“What you want is 100 percent of people to wear masks, but you’ll settle for 80 percent,” said Rutherford. In one simulation, researchers predicted that 80 percent of the population wearing masks would do more to reduce COVID-19 spread than a strict lockdown.
The latest forecast from the Institute of Health Metrics and Evaluation suggests that 33,000 deaths could be avoided by October 1 if 95 percent of people wore masks in public.
Both of these are simulations. Simulations are mathematical models calculated based on whatever assumptions you feed into them. They are not empirical (observed) data. The first paper also indicated in its modeling that the 80%+ masking made a significant difference if instituted within 50 days of the start of the outbreak. (This would have been no later than mid-February in the United States…and yet the authors’ calculations didn’t begin until late March.)
“We assumed an initial infected population of 1% and modelled the assumed effects of social distancing, lockdown, and universal masking over time on the rates of infection in the population” (emphasis mine). They attempted to validate the figures with empirical data, but using the COVID-19 data available by mid- to late-March is pretty premature given the scope of their project.
The IHME publication presents conclusions (and the bold statement from Director Dr. Christopher Murray that “people need to know that wearing masks can reduce transmission of the virus by as much as 50 percent, and those who refuse are putting their lives, their families, their friends, and their communities at risk”), but gives no indication* as to how these numbers were determined (or on what basis Dr. Murray made his claim about 50% transmission reduction).
If we’re practicing social distancing, do we still need to wear masks?
A mnemonic that Chin-Hong likes is the “Three W’s to ward off COVID-19:” wearing a mask, washing your hands, and watching your distance.
“But of the three, the most important thing is wearing a mask,” he said.
Rutherford, as well, is cited in this closing paragraph as saying that “mask wearing is more important” than social distancing. There is no support whatsoever provided for this claim. Simply their personal assertions.
So Are Masks Helpful or Not?
To this point, I’ve been examining the sections of, and the claims in, the Still Confused article. My aim has not been to prove that masks are helpful or are not helpful, but to demonstrate the way the public is being manipulated by misrepresented studies and by writers making claims that are much stronger than they are able to support with data.
When the medical community engages in this type of shoddy and/or dishonest communication, it makes them difficult to trust…and that increases confusion; it doesn’t clear things up.
Which means we’re still left with the question of whether masks work. My personal conclusion, based on the research I’ve seen so far, is: sometimes. By that I don’t mean that they’re unreliable, but that it depends on the context. That is an important discussion, but it’s a discussion for another day.
*Elsewhere on the IHME site there does seem to be information provided. However, given how much effort it takes to track it down, linking to this page does not seem to provide substantive support for the article we’re assessing.