Take-home message
- Wearing masks has a positive benefit for preventing disease for both the wearer and people around them (very high confidence)
- It is essential to have a mask that fits well, i.e. is airtight (high confidence)
- For controlling disease, the strength of masks seems to be, in order: 1) well-fitting N95 masks, 2) well-fitting surgical masks, 3) well-fitting cloth masks, 4) poorly fitting masks of any type (medium confidence)
- I’ll probably wear N95 masks if I can find them cheaply, and a high-quality cloth masks if I cannot. People who dislike wearing masks might reasonably decide not to wear masks.
- The precise size of that benefit depends on the context, existing rates of infection in the community, etc (very high confidence - thanks captain obvious). That said, some ballpark numbers might be that somewhere between 100 and 1,000 people wearing a mask would be needed to prevent 1 infection of COVID (though that infection would obviously cause additional, subsequent infections to other people that are not included in this number).
Some quick maths (wrong but useful):
- We’re probably somewhere in that 100 to 1,000 range number for COVID in Australia in 2024, based on current infection rates (see bottom of this page)
- The 100 to 1,000 range is based on the results of the meta-analysis by Ollia et al 2022, for which the follow-up period ranged from 4 days to 19 months, though mostly on the shorter end
- Though if you also consider other infectious respiratory disease, not just COVID, then that number-needed-to-treat might decrease significantly. I’m not sure how much, but reducing it by 50% or even 65% seems reasonable (i.e. if COVID is half or one-third, respectively, of current the respiratory infections in Australia)
- So it might be more like 50 to 500 people wearing a mask to prevent one infection over a period of a few weeks
- If you assume that applies to, say, 6 weeks, then it’s closer to between 6 to 60 people wearing a mask to prevent one infection over 12 months
- That would suggest that wearing a mask would have somewhere between a ~1% and ~17% chance of preventing an infection over 12 months
- That seems like excellent value to me
- Update 26 April 2024: the rate of COVID is probably closer to 16% (rather than one-third or one-half), see my post here. I haven’t included this updated number in the above calculation. If I did, it would mean that the chance of a mask stopping an infection increases by a factor of 2 or 3. That is, rather than the ~1 to 17% as chance that I conclude below, the answer would be something like ~2.5% to ~45%. The error bars are still very wide, resulting from the other parameters in this calculation that have wide error bars.
Notes from studies
- We approximate the effect of masks on population health by exploring the number needed to treat (NNT), that is, how many individuals need to wear a mask to prevent one person from contracting a respiratory infection.
- Based on the results from our meta-analysis and assuming a low baseline risk of 0.01 and OR = 0.88 in the community setting, the NNT is 841. If the baseline risk is higher, 0.1, the NNT is 92. For adults, our results indicate RR = 0.89, which corresponds to the NNT of 910 if the baseline risk is 0.01 and that of 91 for the baseline risk of 0.1
- only one study measured wearer protection only , while most of the studies included in the meta-analysis included source control (Table S1)
Márquez-Alvarez et al 2022, Relative assessment of cloth mask protection against ballistic droplets: A frugal approach
- Beyond the discussion about the dominant transmission mechanisms, it is safe to say that the generalized use of face masks provides protection in two ways: by limiting the emission of droplets from an infected subject into the environment (source control), and by reducing the inhalation and deposition of droplets by the wearer (wearer protection).
- the perception that a mask wearer protects others better than her/himself is true for the case of ballistic droplets.
Enright et al 2024, Efficacy of facemasks in preventing transmission of COVID-19 in non-healthcare settings: A scoping review
- Wearer protection (PPE) refers to the reduction in exposure to infectious droplets when an individual is wearing a face mask. Within this review, this theme was applied, when both the recipient or target simulators were masked. According to Howard et al., there are two considerations for the wearer in relation to the efficacy of the face masks which include filtration ability and the fit.
- Three simulated experimental studies investigated a number of different wearer protection options with all concluding that when both the source and target were wearing a well fitted mask there was a reduction in exposure to respiratory aerosols by 96%–99.5% (Lindsley et al., 2021a; Wendling et al., 2021; Tomshine et al., 2021). Sugimara et al. (Sugimura et al., 2021) conducted a survey of close contacts and revealed that those who wore face masks were 60% less likely to be infected with COVID-19 than non-mask wearers.
- Two studies comparatively evaluated the performance of face shields and surgical masks as both source control and wearer protection devices (Lindsley et al., 2021a; Wendling et al., 2021) with conflicting conclusions. When both the emitter (source control) and receiver (wearer protection) manikin heads wore a face shield, Wendling et al., (Wendling et al., 2021) results indicated a 98% reduction for the face shields versus 97.3% for the masks. The authors suggested that face shields offered a better barrier effect than the mask against small inhaled particles (<0.3 µm–0.3 to 0.5 µm–0.5 to 1 µm) in all configurations. This, however, is not consistent with results from Lindsley et al., (Lindsley et al., 2021a) who observed face shields were less effective as source control devices, blocking only 2% of the cough aerosol.
- Articles in this review, that examined wearer protection concluded that the fit of the facemask was the most important element to consider when discussing facemask efficacy and was more important that the type of mask worn (Farthing and Lanzas, 2021; Goyal et al., 2021; Lindsley et al., 2021a; Pan et al., 2021).
- [that is, airtightness]
- The meta-analysis of RCTs found a significant protective effect of facemask intervention (OR = 0.84; 95% CI = 0.71–0.99; I 2 = 0%). This protective effect was even more pronounced when the intervention duration was more than two weeks (OR = 0.76; 95% CI = 0.66–0.88; I 2 = 0%)
- Meta-analysis of eight studies showed a significant protective effect (Fig. 2. ≤ 2 weeks, N = 5242; OR = 0.84; 95% CI: 0.71–0.99; I 2 = 0%). In the university residence halls, this protective effect was more pronounced if the intervention duration was more than two weeks (Fig. 2. > 2 weeks, N = 2261; OR = 0.76; 95% CI: 0.66–0.88; I 2 = 0%). The subgroup analysis of intervention settings (households, resident halls or tents) and population (by index, contacts or both contacts and index) did not show any significant difference (Fig. S1, S2).
- [i.e. masks are still good for wearer protection (= “contacts”)]
Gurbaxani et al 2023, Unpacking Cochrane’s Update on Masks and COVID-19
- From the standpoint of workplace safety and materials engineers, the debate on masks is something of an enigma: the utility of wearing masks should be obvious. Viruses like SARS-CoV-2 populate the respiratory tract.9 During talking, singing, coughing, and sneezing, viruses are expelled into the ambient air in small droplets and aerosols.10,11 Tight-fitting masks of various weaves and fiber content filter the droplets and aerosols from the air we breathe with various efficiencies.12 Susceptible uninfected people are protected when the infectious, potentially asymptomatic shedder wears a mask (source control) or when wearing a mask themselves (wearer protection13). Every step in this causal chain of reasoning has been researched and documented and has been verified in studies of household transmission of SARS-CoV-2.14,15
- The exact efficiency of transmission and filtration in each of the stages described can be measured, analyzed, and debated, but it is certainly not zero. From an engineering and materials standpoint, then, the question is not “Do masks work?”…
Arora et al 2022, Face masks to fight against COVID-19 pandemics: A comprehensive review of materials, design, technology and product development
- Different types of protective masks are available in the market which may be broadly categorised into cloth masks, surgical masks and respirator masks.
- In general, cloth masks are more comfortable in terms of breathability and thermo-physiological comfort in comparison to non-textile products.
- Surgical masks […] fit loosely over the face and may not have a face shield
- Some research studies24,52,53 have shown that surgical masks are not inferior to N95 in preventing workplace-acquired influenza. However, one major issue with surgical masks is that they allow airflow through the sides when the user inhales or exhales, causing potential risks of infection.
- On the other hand, respirator masks shown in Figure 2(b), snuggly fit to the wearer’s face and thus seal the inhalation of any fluid, particulate matter, hazardous airborne microorganisms and bio-aerosols to maximum efficiency.
- The most popular respirator mask designed to protect the users, primarily doctors, medical staffs and industrial workers is N95 which blocks at least 95% of very small particles (0.3 μm).
- Despite being more efficient [than surgical masks], respirator masks are uncomfortable due to their high breathing resistance, excessive humidity and heat generation, and for causing itching or irritability. Besides, they also obstruct vocal and auditory capacities, limit dexterity during vocational use, and are also more expensive.
- Though the randomised wear trials to compare the efficacy of these masks show conflicting results, in general, respirator masks have been found to be much superior to surgical masks, whereas, the latter is superior to cloth masks.
- Cloth face masks were found to have significantly poorer filtering performance than medical/surgical masks and N95 masks, but only if non-cloth masks were properly fitted.
- We observed substantial performance differences favouring medical/surgical as well as N95-typed masks over cotton cloth masks
- Examinations of effects accounting for mask fit indicated considerable performance increases of medical/surgical masks and even larger ones for N95-typed masks when they were fitted appropriately
- However, when non-cloth masks were fitted inappropriately, the observed sign changed direction, indicating substantial performance advantages of cloth masks over medical/surgical and N95-typed masks
- This means that although medical/surgical and N95-typed masks generally outperform cotton cloth masks, they are considerably less effective than cloth if not worn properly
- Consequently, it seems reasonable that mandating wearing cloth masks may be justified when this can be expected to yield higher compliance rates than requirements of wearing medical/surgical or N95-typed masks. This could lead to lower infection rates due to (i) higher compliance with the mask mandate in general, (ii) better mask fit due to individually customized designs and (iii) more correctly worn masks because of better breathability and consequently better comfort of cloth masks.
Rates of COVID in Australia in 2024
https://www.usc.edu.au/about/unisc-news/news-archive/2024/february/covid-hasn-t-gone-away-here-s-where-australia-needs-to-do-better
- In Australia, more than 50,000 infections have been reported so far in 2024. And this is likely to be a significant underestimate, as we are testing much less than we used to.
- This was written mid-Feb
- So if that’s a representative 6-week period, it would equate to ~450,000 infections over a 12-month period as an underestimate
- That’s 1-2% of the Australian population, though plausibly that number could be more like 5% or even 10% if you assume that rates of non-reporting are high (which they are)