COVID-19, Health, Science / Tech, Top Stories

Reassessing the Guidance on Face Masks

The efficacy of face masks for limiting the spread of SARS-CoV-2 remains uncertain and hotly contested. Recommendations vary between countries as do the reasons given. In Norway, where I live, masks are not considered necessary because very few people are infected, and efforts to contain the spread of the virus have been quite successful without mandating their use by the general public. But the debate about whether or not masks “work” is complicated and requires attention to numerous variables and contingencies. Even if we can agree that masks do help limit the spread of the virus to some degree, the conditions under which they are most effective can differ enormously, and recommendations to wear masks “in public” can actually mislead more than they illuminate and even cause some harm.

We now know more than we did even a few weeks ago about how the virus spreads. In a blog post inspired by a Quillette essay analysing superspreader events, immunologist Erin Bromage delineates the salient principle as “Successful Infection = Exposure to Virus x Time.” Exposure time is important because infection requires a minimum number of virus particles (the exact threshold remains unknown, but Bromage estimates 1,000). So, repeated or continued exposure for many minutes or hours progressively increases the risk. Oddly, the discussions I read online, even in forums dedicated to SARS-CoV-2, often ignore exposure duration entirely, and some experts have neglected this important variable too.

Bromage points out that infection is rare outdoors, and that the risk is much higher if you are in the same room as an infected individual for a period of time, especially if that person is shouting or singing, never mind coughing or sneezing:

The reason to highlight these different outbreaks is to show you the commonality of outbreaks of COVID-19. All these infection events were indoors, with people closely-spaced, with lots of talking, singing, or yelling. The main sources for infection are home, workplace, public transport, social gatherings, and restaurants. This accounts for 90 percent of all transmission events. In contrast, outbreaks spread from shopping appear to be responsible for a small percentage of traced infections.

The environments Bromage lists highlight three problems with recommendations that enjoin mask use “in public.” First, much of the spread occurs in settings that are private, or that people would not normally think of as public, such as workplaces. Second, some public spaces such as restaurants are hardly optimal for mask wearing. Restaurant staff can wear masks, but it is impractical for patrons to wear them while eating, and innovations like the Israeli Pac-Man design are unlikely to catch on. Third, open air public settings are not especially risky.

Public transport is the public setting in which guidance recommending or mandating masks makes most sense. The risk is high due to exposure time, it’s feasible to wear one, and it is often difficult to avoid crowding that forces people into close proximity with one another. In the United States, there is a clear correlation between reliance on public transport and the COVID-19 fatality rate. New York City tops the list of cities with high transit ridership, and a recent analysis of COVID-19 trends in China, New York, and Italy suggests that masking requirements have contributed to reducing the spread. A recent Danish simulation study of superspreading found that “transmission can be controlled simply by limiting contacts such as public transportation and large events.” A New York City subway car can carry up to 250 passengers. Although the duration of a ride may be shorter than, say, a concert, the number of “events” is much higher because travellers sometimes make multiple journeys a day.

Shops are less hazardous because the risk of spending much time close to an infected person is low unless an unusually high proportion of people present are infected. (The risk is obviously higher for store personnel who spend all day there.) Workplaces may be public in a legal sense, but most employees probably don’t think of themselves as being “in public” at work, and workplaces are typically too varied for a blanket recommendation to be useful. On the other hand, masks worn at home can help limit spread, but only before symptoms appear. This is unlikely to be a general recommendation, but it is worth considering in high-risk cases (such as when an elderly relative lives in the family home). This further highlights the mismatch between guidance recommending the wearing of face masks “in public,” and a more accurate understanding of relative risks posed by various environments.

recent article in Science supports the assessments of relative risk outlined above:

Thus, it is particularly important to wear masks in locations with conditions that can accumulate high concentrations of viruses, such as healthcare settings, airplanes, restaurants, and other crowded places with reduced ventilation.

The authors conclude by recommending:

For society to resume, measures designed to reduce aerosol transmission must be implemented, including universal masking and regular, widespread testing to identify and isolate infected asymptomatic individuals.

The World Health Organisation’s guidance from June 5th is similar in its particulars. Advice differs depending on whether or not there is “widespread transmission and limited or no capacity to implement other containment measures.” Even when there is no such capacity, the WHO recommends masking in “settings where a physical distancing cannot be achieved,” specifically public transport and high-risk occupational settings. Only where there is widespread transmission does it call for the wearing of masks in a much wider range of settings such as “grocery stores, at work, social gatherings, mass gatherings, closed settings, including schools, churches, and mosques.” All the news coverage I have seen on the WHO guidance appears to have missed this distinction.

So a scientific consensus appears to be emerging consistent with the available data. But it is conveyed to the public in confusing ways that unintentionally leave the impression that public settings are dangerous and private settings are safe. Studies that attempt to model the effects of masking without considering the differences in risk between settings therefore produce misleading findings. A much-publicised study from April 21st (two days before the article in Quillette appeared) claimed that “if 80 percent of Americans wore masks, COVID-19 infections would plummet.” Another recent study found that “When 100 percent of the public wear face masks, disease spread is greatly diminished.” This is an attractive idea. Wearing masks is not much of an inconvenience compared to some of the other restrictions imposed upon societies to limit spread. Wouldn’t it be wonderful if we could all just wear masks and otherwise just get on with our lives as normal without worrying too much?

These papers are modelling studies, not empirical ones using actual people, and as far as I can tell, they pay no attention to length of exposure or the difference in risk between settings. In other words, these studies do not seem to adequately reflect the real-world situations where the virus spreads. The earlier of the two papers advocates universal masking, and the recommendations it proposes echo the idea that masks should be worn “in public.” It does recommend that masking be made mandatory or strongly recommended for the general public when using public transport or in public spaces, for the duration of the pandemic.

The researchers claim to have modelled the effect of mask wearing by a given percentage of the population. But they do not seem to have considered that 80 or 100 percent would not use masks in possibly infectious settings such as restaurants or private gatherings. This might be the equivalent of halving the amount of mask wearing, and we could easily end up in a situation where their own models predict that the effect of masking basically evaporates:

“This is the goal,” De Kai maintained. “For 80 or 90 percent of the population to be wearing masks.” Anything less, he added, doesn’t work as well. “If you get down to 30 or 40 percent, you get almost no [beneficial] effect at all.”

Overly broad masking requirements are at best useless, and possibly harmful, since they can cause confusion and prompt at least some to rebel against masking if the practice is too onerous or impractical. People will think for themselves, so if recommendations are to be effective, they need to make sense. The public needs to understand why wearing a mask in certain settings is important and less so in others so they can make informed judgments about risk and safety on their own, instead of being asked to robotically follow a generalised guideline about behaviour “in public.” Clear, comprehensible, and reasonable policies will hopefully provide less room for confusion and for polarised discussions. This is particularly critical at a time when societies are trying to reopen their economies and return their populations to free and normal lives as safely as possible.


Dagfinn Reiersøl is a software developer and author of PHP in Action. You can follow him on Twitter @dagfinnr.

Photo by engin akyurt on Unsplash.


  1. I’ve literally seen people say that because many of the people are wearing masks, packed thousand strong protests with people right on top of each other screaming are not dangerous.

    I’ve taken a moderately pro-mask stance since the beginning of the outbreak, but I do sort of feel like its taken on a symbolism that scares me.

  2. Nassim Taleb has said some intelligent things on other issues, but he has been the most extreme COVID alarmist in the world. His latest:

    I would love to see him challenged on the economic ramifications of shutdowns, vis a vis the cost of COVID deaths. He should at least address the question.

  3. A sensible article. I like the stress on the high-risk circumstances and the factor of duration of exposure, not just exposure per se. Good reasons to wear masks but not be a fanatic about it.

    Airplane cabin air is filtered. The real risk with air travel is airports, especially public restrooms. It’s not the fault of the cleaning staff. Even the cleanest restrooms have minimal ventilation.

  4. Good article but I couldn’t focus well because I’m too busy wowing over the name Dagfinn. Dagfinn! That has to be the coolest name ever! What does it mean?

    Dagman! Dawgfinn! The Finnster Dude!

  5. I also appreciate the author’s reasonable approach to mask wearing and emphasizing that usefulness depends on the situation. I’ve become frustrated with the polarization with this issue. I know some who get upset at the mere mention of having to wear a mask in an establishment (I find that this is not difficult and I don’t mind doing it), but seeing people wearing it walking in the neighborhood where they don’t get close to anyone seems to be pure symbolism. As my city is discussing making face masks mandatory, I am afraid that I will be forced to put on a face cover just to go for a walk outdoors.

  6. The standard for whether you can be legally compelled to do something should not be “It is not that difficult and I don’t mind doing it.”

  7. I’m sorry if it sounded like I support legal compulsion to wear face masks - I do not - I merely meant to say that I don’t agree with the blanket refusal to wear a mask in any situation that I encounter sometimes. I think voluntarily wearing face-masks in some situations is probably a good thing. But the fact that mask wearing seems to now demonstrate which “side” you are on, in many cases, makes the act of mask wearing (or not) no longer about its objective usefulness but a matter of virtue signaling.

  8. “These papers are modelling studies, not empirical ones using actual people, and as far as I can tell, they pay no attention to length of exposure or the difference in risk between settings. In other words, these studies do not seem to adequately reflect the real-world situations where the virus spreads. The earlier of the two papers advocates universal masking”

    Thank you for this article and for pointing out the quoted portion. I’ve read the well-publicized studies and modeling exercises coming out in the last couple of months, and likely 100 more that received only transitory attention, or which were older studies cited in the newer studies and modeling exercises. As Mr. Reiersøl and Professor Bromage point out, modeling may be useful but often deviates from what happens in the environment.

    Most of the modeling exercises I’ve read seem to take assumptions for which the evidence is still controversial, apply the assumptions to their model, and then announce “Voila! Our Study Shows 80% If Americans Wore Masks, COVID-19 Infections Would Plummet!” This is the title Vanity Fair put on it, for example (except the sarcastic “Voila!” is mine) - the actual title is “Universal Masking is Urgent in the COVID-19 Pandemic…” [“Urgent!”]. Part of the problem is also the unwillingness or inability of most mass media writers to dig into the actual study and take note of any potential circularity issues.

    Another one mentioned here that I dug into last week was the one that modeled the transmission rates and trends thereof in Italy, New York, and Wuhan China from January through early May, and concluded the mask wearing in public was “the most effective means to prevent interhuman transmission” (“Identifying airborne transmission as the dominant route for the spread of COVID-19”). The 7-day moving average for daily new cases in New York dropped from 8400 per day to 2400 per day in the 4 weeks following the masking order on April 17th, which is quite impressive. Italy’s 7-day moving average for daily new cases dropped from 4400 to 1600 in the 4 weeks following the April 6 order in the hard-hit Northern regions to mask in public; similarly impressive.

    Fine so far as it goes. But one wonders why not include contemporaneous contra examples in their modeling exercise? Singapore’s in-public mask-wearing mandate was April 14, and by April 17 SG was reporting over 85% compliance with mask wear (more than enough to hit the desired 80% threshold), and since that time is consistently reported over 90%. The week before masks became mandatory, the 7-day moving average was 150 new cases per day but was rising quickly (hit 250 the day mandatory masking began). But in SG, 4 weeks later - which should have been enough time for universal mask wear to have become relevant, if 4 weeks was relevant for NY, IT - the SG 7-day moving average was 750 new cases per day. And 8 weeks post-order was still around 400 new cases per day. As of yesterday it’s down around 350.

    It may not be really worth mentioning NZ as an alternate contra example, due to its small size and less frequent international travel, but I’ll still toss it out as a country with a quite abbreviated disease progression curve despite no universally masking.

    Then there’s California, where the 7-day moving average of new daily cases has gone from 1760 up to 3440 the last 5 weeks, and if the news reports are correct, the majority of these increases are from LA County. Note Los Angeles and LA County made in-public mask wear mandatory 5 weeks ago, although unlike the example of SG, I don’t have data as to what extent Angelenos took this mandate to heart. But still, if public masking mandate should cause infections to plummet, it did not work in this case.

    None of the above proves masks don’t work. I just get irritated by what are to me exercises in cherry picking and/or over-enthusiastic assignations of efficacy.

    As an example, a friend sent me this one saying “See! He’s a real science guy - at NIST even! - and proves mask wearing reduces COVID spread.”.

    It’s the usual schlieren photography you’ve probably already seen demonstrating mask wear, except he only images the plume from mask-front. I told my friend that this scientist (a) should not have equated his physical findings with a conclusion relating to disease spread and (b) should have included imaging of the plumes expelled along the sides of his head and out toward the back of his head. That way his readers could understand that it’s better to be masked and yelling or coughing at each other face-to-face, than it is to be standing behind someone in a queue who is wearing a mask and panting or talking loudly or coughing. (abstract) (full paper)

    “… surgical and hand-made masks, and face shields, generate significant leakage jets that have the potential to disperse virus-laden fluid particles by several metres… all showed an intense backward jet for heavy breathing and coughing conditions.”

  9. It certainly seems this way from all the news reports I read. But I wonder how endemic this “picking sides, virtue signalling” by mask wear or not-wear really is. I have a large household and am the cook and grocery shopper. In my area, I would estimate perhaps 60% of the public I see are masked. No one who is not wearing a masks seems to care that these others are, and no one who is wearing a mask seems to care that the rest are not.

    For the few businesses in my area that request customers don a mask, I happily do so. Which is what I took you to mean in your first comment (“some who get upset at the mere mention of having to wear a mask in an establishment (I find that this is not difficult and I don’t mind doing it))” - you were talking about what individual businesses require, not gov’t mandates. Otherwise, I do not routinely wear a mask outside. If I had to be in a higher risk situation (crowded tube conveyance for an hour’s ride, e.g.), I would be sure to have gotten myself an N95 or higher level mask and wear good eye cover. To protect myself, that is.

  10. My first Dagfinn was:

    He’s a pretty cool guy, although it’s been many years since I’ve even thought of the name.

  11. Can’t quite get why I see some people driving with windows rolled up wearing a face mask! Are they thinking they might infect themselves?

  12. In many areas, businesses are required to require masks by their governments. This presents an opportunity for dishonesty - “It’s not the government forcing you to!” when actually that’s exactly what it is. There is also the threat of litigation against businesses if someone decides to believe that he caught the virus while there.

  13. I imagine they go home and then lay alone in their bed with a condom on.

  14. As a software developer, I can assure you that we are experts in everything.

  15. It’s a health precaution with varying effectiveness given context. If you wear it in a context where either the danger is almost nil or where the danger is so high a mask won’t protect you then its little more then a symbol.

    Wearing a mask isn’t going to make going to a thousand strong protest standing arm in arm safe or advisable.

    Wearing a mask when nobody is around or has been in that area is nonsensical.

Continue the discussion in Quillette Circle

17 more replies


Comments have moved to our forum