Skip to content

Revisiting masks: Here’s how different face coverings perform against COVID-19

A few days ago I  wondered aloud about how good various kinds of masks are. Since then several people have sent me versions of something you may have seen: a matrix that provides a set of numbers—sometimes filter efficiency, sometimes estimates of how long they protect you—that depend on the type of mask you're wearing and the type of mask other people are wearing.

As it happens, that matrix is what started me thinking about this in the first place. And after a bit of investigation I've come to a couple of conclusions.

First, the old conventional wisdom is that masks reduce the number of virus particles you exhale, therefore helping keep the air clean for others, but they don't do much to protect you. That, however, turns out to be the old hotness. The new hotness is that masks do indeed protect both you and other people. So I stand corrected on that.

Second, I began to wonder where that matrix came from. Long story short, the earliest version I can find comes from Simon Smith, a Canadian expert in respirator filters. However, his matrix is labeled as a "sample calculation," which suggests it's something of a guess—though certainly an educated one.

So I started to look around for other estimates. In the end, I found four: Duncan, Smith, the EPA, and Koh. Their results are not always easy to summarize, but here's my best shot:

These numbers represent the percentage of virus-carrying aerosol droplets that the mask blocks, with the average shown in the first column. All of them are based on indoor settings with average ventilation. For example, if everyone else is unmasked, a typical cloth mask reduces your exposure by 31% compared to what you'd get if you wander around barefaced. That's not much. Conversely, a well-fitted N95 mask blocks 95% of the particles.

It works in the other direction too. If you do nothing but everyone else is wearing cloth masks, it reduces your exposure by 31% compared to what it would be if everyone else was also maskless.

The widest disagreement comes with the N95 masks. On the one hand, Duncan provides a spectacular estimate that a fitted N95 mask blocks 99.4% of all virus particles. Conversely, Koh is a skeptic because he believes the public is hopeless about wearing them properly. In real life, he thinks a surgical mask is probably better than an N95.

(And what does "well fitted" mean? Technically it has two parts. The first part is an annual test that takes about 20 minutes. Once you find a mask that passes, you are expected to wear exactly that kind of mask forever. The second part is a user fit test, which you do every time you put on the mask. We ordinary folks never do the first part, and it's unclear how much we do the second part. That's where the disagreement over effectiveness comes from.)

Anyway, I can't adjudicate the different estimates, so I just averaged them all. If we then assume that masks are equally effective at both inward and outward protection,¹ you get the famous matrix. Here's my version based on the averages above:

Take this for what it's worth. What it tells you, in theory, is that if you and everyone else are wearing surgical masks, it reduces particle volume by 79% compared to wearing nothing. That's about a 5x improvement. If you're wearing a non-fitted N95 and everyone else is wearing surgical masks, particle volume is reduced by 87%, an 8x improvement.

Here's another way of looking at the same data:

If you're wearing a surgical mask and everyone else is wearing cloth masks, it provides about 3x the protection of doing nothing.

Just generally, I'd beware of the N95 numbers. You should probably assume that even if you're being careful, your spiffy N95 mask isn't all that well fitted. The "N95 non-fitted" line is probably the most accurate for real life.

This is the best I could come up with doing a non-expert review of the literature. Don't take it as gospel, but as a rough estimate. If better data comes my way, I'll let you know.

¹I don't know if this is a reasonable assumption. Simon Smith uses it in his example, and he's an expert. Koh actually tests in both directions and produces a wide range of results. However, at a rough eyeball guess, his results suggest that inward and outward protection are roughly in the same ballpark.

67 thoughts on “Revisiting masks: Here’s how different face coverings perform against COVID-19

  1. akapneogy

    "It works in the other direction too. If you do nothing but everyone else is wearing cloth masks, it reduces your exposure to 70% of what it would be if everyone else was also maskless."

    So, in a universally masked environment, the virus load transmitted is reduced by ~ 50% even with the least effective masks. I don't know what that means in terms of getting sick with the virus - It probably depends on the strain and the immune system of the recipient. But a substantial reduction in risk, nonetheless.

    1. cmayo

      Exactly. "Just wear something" - unless you're really worried about getting sick (you have additional factors or you are going to be seeing people with additional factors), don't worry too much about what you're wearing. Just wear something.

      TBH, I'd also call a 30% reduction in risk a "substantial" reduction, especially for the vaccinated.

  2. Jimm

    "If we then assume that masks are equally effective at both inward and outward protection"

    To my knowledge, we cannot make that assumption, as I've seen expert analyses that describes single-layer cloth masks as useless relative to protecting the wearer, but somewhat useful for protecting others.

    I'll try to dig it up, may have been the Mask Nerd guy, I know it was sometime in past month or two.

      1. Jimm

        https://www.theguardian.com/commentisfree/2021/dec/27/best-masks-covid-tests-cloth-surgical-respirators

        "With the Omicron variant spreading around the world, the most common question I get is still the most basic: what kind of mask should I be wearing? If possible, not a cloth or surgical one. These masks are moderately effective at capturing particles coming out of your own mouth – what we engineering nerds call “source control”. But they offer little protection against incoming particles. We call this part – how well the mask protects the wearer – respiratory protection."

            1. KenSchulz

              Because I followed the first couple of links cited by Aaron Collins (in the sentence ‘Analysis and studies suggest that with cloth or surgical masks, the risk of infection may increase within minutes of being around others with either no mask or low-performing masks’), and both papers assume inward and outward leakage to be equal; nor does either make any statement to imply that filter penetration is directional*. So if particles pass through the filter equally in both directions, and leak around it equally in both directions, how does the protective asymmetry arise?
              *if it were, they would have to test filter efficiency in two directions. And state how one distinguishes the ‘inside’ and ‘outside’ of a piece of cloth.

              1. Jimm

                https://health.clevelandclinic.org/heres-how-wearing-a-cloth-mask-helps-fight-the-spread-of-coronavirus/

                "Dr. Dweik confirms that those homemade masks, often made of material like cotton, won’t do much to protect you from inhaling particles that carry the virus. But they do protect wearers from possibly spreading the virus to others."

                “There is evidence that such masks reduce the exhaled aerosols from infectious, but asymptomatic, individuals.” he says. By blocking the exhale of virus particles in to the air around you, your mask is keeping the virus from spreading.

                "Additionally, the masks serve as a really helpful physical barrier when you cough or sneeze."

        1. KenSchulz

          I’m only being partly facetious. I have seen this assertion in multiple places, but I have not seen either data, or a mechanism.
          So what could be the mechanism underlying the asymmetry of protection? Certainly the material isn’t directional (otherwise turning the mask inside would actually work!)

          1. Jimm

            I'm not an expert in masking, and don't pretend to be, and neither are you, but off the top of my head can think of several ways exhalation differs from inhalation, most obviously that we cough and sneeze, which has no equivalent in inhalation.

            So sure, if you want to just think of the easiest thing you can think of, regular breathing, and then posit how masks could vary there in either direction, be my guest, have fun with that while the rest of this examine the whole range of things you would test a mask against.

          2. dausuul

            Here is an article on the subject: https://iopscience.iop.org/article/10.1088/1752-7163/ab8a55

            The theory seems to be, when you exhale the virus, it comes out in tiny droplets of liquid. Those droplets eventually evaporate in the air, leaving the virus floating around "dry." If you're wearing a mask, what you exhale hits the mask while still "wet," and the droplets stick to the mask and form a layer that attracts other droplets.

            With "dry" virus, you don't get this effect. You do still get the benefits of the mask physically filtering particles, however.

            I haven't found much in the way of studies confirming that this is actually how it works. But there is at least a theoretical basis for it.

            1. KenSchulz

              Thanks for the link. This is a plausible mechanism. The Nature article I referenced as https://www.nature.com/articles/s41598-020-78110-x found that evaporation only reduces droplet diameter by about half, though, so virions are not floating free, but in smaller clumps of proteins and small molecules. So one has to assume that the smaller size and absence of water account for the difference in protection factor. As reasonable as the assumption is, it would be nice to see data establishing the size of the difference.

  3. Jimm

    I would add that even a 30% reduction in viral particles should be considered "effective", especially if (fairly) universally worn. We don't know all the dynamics of the disease process, but we do suspect (and some research supports) that higher viral loads to translate to worse outcomes (beyond any threshold where your body may just outright smother the infection before it turns to disease).

  4. Crissa

    It presumes a particle size - which is a droplet size and ignores what portion of the viral load is in what size particles. And then it presumes t-shirt material for the cloth and 0% baseline for the n95?

    The n95 filters less than cloth when not fitted because of pressure. I don't understand what they think.

    1. akapneogy

      "The n95 filters less than cloth when not fitted because of pressure. I don't understand what they think."

      Bypass fraction higher because of higher pressure inside the N95?

  5. Solarpup

    "In the field" tests, at least according to the medical school experts following it here, don't seem to be showing the same efficacies as shown in these kind of matrices. Probably because so often I see folks wearing masks pulled down below there nose, or even chin. Or they're wearing an N95 on top of a bushy beard.

    That being said, none of this is going to stop me from wearing my KN95s. Just going by risk/reward, I honestly don't feel any significant "burden" wearing it, so even if the efficacy isn't what I might hope, I truly don't get why folks would object to wearing a mask for even just a 50% reduction. Other than "freedom" of course.

  6. golack

    The nature of the virus and how it spreads has changed a bit. All masks help. But it's possible that the good cloth masks that could drop the infection rate, R0, from 1.2 to 0.9 may not help as much with the new variants with starting infection rates of 1.5 or higher.
    (yes, it's a lot more complicated...)

      1. Rattus Norvegicus

        The R0 number (reproduction rate given no countermeasures) for omicron is estimated, because that is all we can do, is in the range of 12 to 14. About as infectious as measles.

        However, human populations react to an infectious disease. People get infected and recover, people wear masks, people get vaccinated, people do like I do and take all these measures and also hole up in their homes.

        What you are referring to is the R(t), the effective transmission rate. In the US it is right around 1 right now.

        1. illilillili

          San Mateo county publishes an estimate of R(t). Last week it was around 1.6, this week it's down to 0.76. Just as an example for one locale...

  7. fredtopeka

    A few of us pointed to Aaron Collins, he has been testing masks for almost 2 years using an aerosol in his home (and actually wears the mask, he just puts them on so no professional fitting). I link to his latest video below, it's just a short but includes a link to his latest testing data

    https://youtu.be/0me15TZdgGw

  8. Rattus Norvegicus

    The county health department where I live has been tracking spread in school districts which had mask mandates vs. those w/o mandates. Guess what? Districts w/o mandates had about 3x the spread of districts with mandates.

    I guess one of the telling signs is that often when a district started off with a "mask options" policy they always pretty quickly imposed a mandate as COVID-19 quickly spread and caused staffing and class attendance issues within a week or less of reopening. In particular, the local high school district announced over Christmas break that they would reopen in January with masks options (done while omicron was spreading rapidly in the county, don't ask me why, I live in a conservative state). It took less than a week for them to realize the mistake they made.

    Oh well, but if everyone is wearing masks, they do work.

  9. KJK

    People should be required to wear N95 / KN 95 / KF94, ect, for the simple reason that with those types of masks, it is much harder for blithering idiots to wear them below their noses (or below their chins).

      1. Justin

        I guess I should have been more specific.

        So... in my local area... in the retail stores and restaurants I have visited... few (less than 10%) wear masks. It's been like this for quite a while now... 10 months maybe.

        At my workplace, masks are required except while eating or in an area alone.

  10. kk

    This matrix only claims to measure what % of virus particles go through the mask. Unless the mask is real tight all around, lots of air bypasses the mask - airflow tends to take the path of least resistance.

    A few months ago there was a (large) study from Bangladesh. It showed that surgical masks are 12% effective in terms of protecting against covid infection and cloth masks are ~0% effective.

      1. golack

        The effect was real and reasonable, especially since the effect on the elderly was a 30% reduction. The authors were also seeing mask usage was not perfect nor universal--so real world effects.
        Cloth masks may also help--but error bars were too large to be conclusive. The surgical masks were washed, and filtration dropped a little with washing.
        https://www.science.org/doi/10.1126/science.abi9069

    1. Crissa

      Their qualifications for studies fail for having too small a sample, and too specific a testing regime. Their idea of a proper study could not be conducted.

  11. kahner

    "the old conventional wisdom is that masks reduce the number of virus particles you exhale, therefore helping keep the air clean for others, but they don't do much to protect you. That, however, turns out to be the old hotness. The new hotness is that masks do indeed protect both you and other people."

    Does anyone know what the basis of the "masks protect others but not the wearer" theory was? That always sounded illogical to me and tbh i just didn't believe it, but since either way the thing to do was wear a mask, I never spend any time looking into it. But now that it's apparently no longer the consensus I'm curious why it ever was.

    1. Jimm

      https://www.ft.com/content/7739d918-f56d-11e9-b018-3ef8794b17c6 (from 2015):

      "The use of surgical facemasks is ubiquitous in surgical practice. Facemasks have long been thought to confer protection to the patient from wound infection and contamination from the operating surgeon and other members of the surgical staff. More recently, protection of the theatre staff from patient-derived blood/bodily fluid splashes has also been offered as a reason for their continued use."

      "A contemporary questionnaire-based study, which attempted to assess the attitudes of surgeons, revealed that 96% of responders wore facemasks. About equal numbers did so with the primary aim of protecting the patients compared to protecting themselves. However, it was also found that 20% of responding surgeons wore the mask for the sole purpose of respecting tradition."

      1. kahner

        Ok. But the use case of surgical staff in an operating room is completely different from what we're talking about with covid. Of course when you're cutting someone open and there's not particular expectation that anyone has a highly communicable airborne spreading respiratory disease then masks are mostly there for patient protection, not wearer protection. But in the case where we have a global pandemic that we know is spread via aerosols transmission, the surgical situation is pretty irrelevant in my opinion. I would have hoped there was some better data driving the theory than that.

        1. kahner

          ETA: also, even if the surgical case were generalizable to covid spread a "questionnaire-based study, which attempted to assess the attitudes of surgeons" is also pretty meaningless to me. I don't care what some surgeons happen to think about mask use, I can what actual data from a high-quality study says.

            1. kahner

              I haven't had time to read through it all yet, but even this intro bit, "overall there is a lack of substantial evidence to support claims that facemasks protect either patient or surgeon from infectious contamination.....even today, it remains unclear as to whether they confer any tangible benefits to surgical outcomes. ", is pretty surprising. But I supposed it is a tough thing to test under ethics rules.

      1. Crissa

        It's because viruses hitch a ride upon the vapor droplets, the particles of water exhaled. These then float into the air on the hot breath and evaporate, distributing the virus into the air.

        The masks provide a surface for the vapor to adhere to, and absorb the viral particles befor they get launched into the air.

        It's about the absorbency, not the particle size.

        1. KenSchulz

          That is a plausible mechanism for protection asymmetry. But I haven’t seen the data to support it yet. What range of droplet sizes remain airborne long enough for the water to evaporate and aerosolize the contained virions? How many droplets in this range are emitted during normal breathing? During speaking? Or coughing, sneezing, singing? Are aerosolized virions emitted directly in exhalations? What modes pose the greatest risk for transmission?

  12. Vog46

    OK enough about masks
    I worked in NC"s OSHA program for 20 years
    Respiratory protection has got to have 2 things going for it
    1 - filtration. The more it filters the better the protection
    But this may come at a "cost" and that is "breathability"
    2 - its got to be comfortable to don, and use

    EVERYONE including Kevin looks respirator use as a ONE item thing. It is not.
    The USER in industrial settings has to be MEDICALLY evaluated to be able to wear the mask. THEY take a Pulmonary function test) Certain people with impaired lung function cannot wear filtrating masks because they can't breathe through the filter media. Next thing you know the particle mask is below their nose. I have COPD and could only wear a respirator for 1 hour - which in most cases was long enough for me to get into the area, see what was going on and determine which employees I needed to interview (outside the work zone). If I needed help I could always call for back up. In 99% of my inspection activities I never wore a respirator - never had to. When I would respond to a spill, we KNEW what chemical we were responding to (Ammonia, being the refrigerant of choice for food processing was the most common. It only stayed at ground level for a brief few minutes. In some cases it took HOURS for us to respond.)

    But that is WORK. Now take that same person, me, with COPD, and put me in a grocery store, If it's crowded I do NOT enter. (I shop either very early or just before closing). I am fully vaxx'd, have had COVID, and am boosted. I social distance, but I still wear a surgical mask with an cloth mask overlay. I can breath and even exert myself wearing this set up. I can't run a marathon due to my lungs and 2 artificial knees but I can stride my way through a grocery store, in a hurry if I have to

    But think about this for a moment
    The boomers smoked - a LOT
    The next 2 generations tempered cigarettes by using pot, now we have "the vapors". Add to that the obesity, lack of exercise and other factors and you can see why MANY of us don't want to wear masks and use the "anti-mandate" argument to JUSTIFY our out of shape-ness, therefore "I won't wear a mask" attitude.

    Cut the shit Kevin.
    No matter how good a mask is if the user struggles to breath, he/she will take the easy route and pull it down below their nose, use the cheapest easiest to breathe in masks or not wear one at all
    As I have stated - many men wear masks over their beards. As Ron White said: "You can't fix stupid".
    No one will spend the money to use a powered filtering respirator. For the Uninitated that's a motorcycle helmet that seals around the head or neck and pumps air through a filter an into your breathing zone. (Very simplified)
    ANY protection is better than no protection so long as those that need to, wear it.
    There are far too many of us that believe we are "above" all that. THAT is the problem

    1. Crissa

      You need to choose a respirator with a higher amount of surface area and a lower resistance to breathing.

      Cut the shit indeed. I switched to a better mask, and now my biggest problem is the headband.

    2. fredtopeka

      If you go to the Aaron Collins video I linked to above, he also measures pressure drop for the masks which is a measure of breathability. He found some of the KF94s have a pressure drop similar to cloth masks, so they do a much better job of filtering and are about as breathable.

  13. Vog46

    I am exempted from wearing a mask Crissa. Medically I have the reasons NOT to wear one and the documentation I need if I decided to go into a place that requires masking and I was Un maksed.
    So I choose to wear the one (or 2 in my case) that gives me the ability to breath somewhat normally and along with some common sense allows me to reduce the risk to myself and to others around me.,
    Respiratory protection is hard. In an industrial setting you know what you can be exposed to. In public it's a whole new ball game and anyone who has medical conditions has to be cognizant of their surroundings.
    Far too many people are wearing masks the wrong way.
    Over beards? Over their nasal cannula's attached to their Oxygen tanks?
    Good grief.
    A good PFT will tell a user IF they can wear a mask and in our society there are many MANY people who simply just don't know just how badly their lungs are performing

    1. Spadesofgrey

      Jr better watch it, he has skeletons that can be brought out, rather than whining boring dialectics. Try harder bub. Everything you said is played out boring.

  14. Eric Pedersen

    I like this analysis; it's clearly mechanism based, and gives a lot of useful information. Have you thought about translating the grid into a matrix of exposure times though? It might be easier for most people to interpret if this was given as time it would take to get an equivalent dose of the virus" compared to no masking. It would help with the risk calculation, and convey the fact that, regardless of masking, longer exposures are worse than shorter exposures.

    It would also make it easier to translate into cases where the amount of virus being emitted, like the gym or in a classroom or lunchroom with people talking) is higher relative to environments where people are sitting and talking.

  15. Pingback: Mike’s Blog Round Up | Crooks and Liars – Starwide

  16. illilillili

    Your model seems to assume that each virus particle has an equal chance of causing an infection. An alternative model is that some threshold number of virus particles must find a home where they can reproduce and thence overwhelm the immune system. In which case, once you reach a certain level of paired-mask effectiveness, filtering more particles wouldn't help much.

  17. shamhatdeleon

    The "fit test" is 95% reminding you how to put it on correctly. Most people wear the "regular" size.

    Highlights:

    The bottom strap below your ears and behind your neck. The top strap goes across the top.

    Don't pinch the nose strip because you might create a pocket for air to bypass the mask. Instead, press it against your nose after it's on.

    If you have a beard, it isn't going to seal no matter what you do.

  18. doughelo

    I'm not seeing ANYTHING here that equates to actual risk in the real world. What's the viral load of an asymptomatic vaxxed individual? What're the odds of a vaxxed person being exposed to enough virus to actually get infected, especially where vax rates are over 70%? We're seeing a lot of positive tests, but not a lot of vaxxed people getting sick, and almost none requiring hospitalization. If it wasn't for the unvaxxed clogging up the ERs, we could all burn our masks tomorrow risking no more than the equivalent of a bad cold.

Comments are closed.