Some warm-up exercises, then we review the Bangladeshi paper.
But first a reminder: I have no burden, no burden whatsoever, not even in the least degree, to prove mask mandates don’t work. Mask mandate supporters, however, must show conclusive evidence that their mandates provide value. This they cannot do, and have not done.
Dr. Anthony Fauci wrote in February 2020 that store-bought face masks would not be very effective at protecting against the COVID-19 pandemic and advised a traveler not to wear one.
I am loath to agree with anything this foolish man says, but when he’s right, he’s right. And he’s right.
The commonly worn cloth and surgical masks are roughly 10 percent efficient at blocking exhaled aerosols, a University of Waterloo study found.
The study, examining the effects of masks and ventilation, ultimately found that commonly used cloth and surgical masks do little to filter exhaled aerosols.
“The results show that a standard surgical and three-ply cloth masks, which see current widespread use, filter at apparent efficiencies of only 12.4% and 9.8%, respectively,” the study concluded, noting that KN95 and N95 masks were far more effective at filtering out aerosols.
“Apparent efficiencies of 46.3% and 60.2% are found for KN95 and R95 masks, respectively, which are still notably lower than the verified 95% rated ideal efficiencies,” researchers continued in the data published last month prior to the Centers for Disease Control and Prevention (CDC) reversing course, advising fully vaccinated individuals to wear masks if they are in high-risk areas.
The study’s conclusion continued:
Furthermore, the efficiencies of a loose-fitting KN95 and a KN95 mask equipped with a one-way valve were evaluated, showing that a one-way valve reduces the mask’s apparent efficiency by more than half (down to 20.3%), while a loose-fitting KN95 provides a negligible apparent filtration efficiency (3.4%). The present results provide an important practical contrast to many other previous experimental and numerical investigations, which do not consider the effect of mask fit when locally evaluating mask efficiency or incorporating mask usage in a numerical model. Nevertheless, if worn correctly, high-efficiency masks still offer significantly improved filtration efficiencies (apparent and ideal) over the more commonly used surgical and cloth masks, and hence are the recommended choice in mitigating the transmission risks of COVID-19.
Golly. The peer-reviewed paper is “Experimental investigation of indoor aerosol dispersion and accumulation in the context of COVID-19: Effects of masks and ventilation” in Physics of Fluids by Shah et al. Abstract abstract:
However, leakages are observed to result in notable decreases in mask efficiency relative to the ideal filtration efficiency of the mask material, even in the case of high-efficiency masks, such as the R95 or KN95. Tests conducted in the far field (2?m distance from the subject) capture significant aerosol build-up in the indoor space over a long duration (10?h).
This is another paper testing masks under “ideal” conditions. Which are, of course, scarcely realized in practice.
The study, which analyzed some 90,000 elementary students in 169 Georgia schools from November 16 to December 11, found that there was no statistically significant difference in schools that required students to wear masks compared to schools where masks were optional.
“The 21% lower incidence in schools that required mask use among students was not statistically significant compared with schools where mask use was optional,” the CDC said. “This finding might be attributed to higher effectiveness of masks among adults, who are at higher risk for SARS-CoV-2 infection but might also result from differences in mask-wearing behavior among students in schools with optional requirements.”
Golly times two.
From June 2, 2020 through August 12, 2020, there were 40,771 reported cases of COVID-19 within Bexar County, with 470 total deaths. The average number of new cases per day within the county was 565.4 (95% confidence interval [CI] 394.6–736.2). The average number of positive hospitalized patients was 754.1 (95% CI 657.2–851.0), in the ICU was 273.1 (95% CI 238.2–308.0), and on a ventilator was 170.5 (95% CI 146.4–194.6). The average deaths per day was 6.5 (95% CI 4.4–8.6). All of the measured outcomes were higher on average in the postmask period as were covariables included in the adjusted model. When adjusting for traffic activity, total statewide caseload, public health complaints, and mean temperature, the daily caseload, hospital bed occupancy, ICU bed occupancy, ventilator occupancy, and daily mortality remained higher in the postmask period.
There was no reduction in per-population daily mortality, hospital bed, ICU bed, or ventilator occupancy of COVID-19-positive patients attributable to the implementation of a mask-wearing mandate.
These pre- post- intervention studies are common enough. And yes, it’s true, it could very well be that something besides mask wearing caused deaths to increase.
Yet there is zero evidence here masks work. As is zero.
(Never, ever put confidence intervals around an observation unless that observation is measured with error, which is here not zero. You saw 565.4 average patients a day. You didn’t see a confidence interval.)
The paper is “The Impact of Community Masking on COVID-19: A Cluster-Randomized Trial in Bangladesh” by Abaluck and a slew of others. This appears to be an NBER pre-print.
My conclusion (for those short of time): even if masks work, the effect is below even trivial. Am not changing my mind. Mask mandates are useless. Now for the details.
Abstract Method (my emphasis):
We conducted a cluster-randomized trial of community-level mask promotion in rural Bangladesh from November 2020 to April 2021 (N=600 villages, N=342,126 adults). We cross-randomized mask promotion strategies at the village and household level, including cloth vs. surgical masks. All intervention arms received free masks, information on the importance of masking, role modeling by community leaders, and in-person reminders for 8 weeks. The control group did not receive any interventions. Neither participants nor field staff were blinded to intervention assignment. Outcomes included symptomatic SARS-CoV-2 seroprevalence (primary) and prevalence of proper mask-wearing, physical distancing, and symptoms consistent with COVID-19 (secondary)…
Right away, we see reporting should be done by village, hierarchically. Summing across villages is dangerous because of the very real possibility of Simpson’s paradox, especially if those villages differ widely in population. In other words, we shouldn’t see the authors summing across all villages, but averaging effects across “randomized” (and not blinded) villages.
Let’s recall their goal: “Outcomes included symptomatic SARS-CoV-2 seroprevalence (primary)…”
Let’s see their reporting (my emphasis):
Blood samples were collected from N=10,952 consenting, symptomatic individuals. Adjusting for baseline covariates, the intervention reduced symptomatic seroprevalence by 9.3% (adjusted prevalence ratio (aPR) = 0.91 [0.82, 1.00]; control prevalence 0.76%; treatment prevalence 0.68%). In villages randomized to surgical masks (n = 200), the relative reduction was 11.2% overall (aPR = 0.89 [0.78, 1.00]) and 34.7% among individuals 60+ (aPR = 0.65 [0.46, 0.85]).
Red flag #1: They summed across villages, with the real possibility of Simpson’s paradox. Red flag #2: they collected data only on (some) “symptomatic individuals”, and not all individuals. Since many “symptomatic” did not have Covid (as you will see), “symptoms” were badly defined, even though the definitions were from the WHO. Just what were the symptoms symptoms of? Why report on symptom differences when only a fraction of those with symptoms had Covid? Reducing symptoms is therefore almost meaningless.
Red flag #3: They only collected blood on 10,952/342,126 = 3% of the population. This isn’t necessarily a red flag, because that sample size could be sufficient. But if they grabbed all “symptomatic” people, then at worst only 3% of the population developed symptoms, which is small. Meaning Covid did not spread much, because only a fraction of that 3% had Covid at the time of testing. They didn’t test for antibodies, so people infected previously were missed.
Red flag #4: They don’t present the raw numbers, only the “adjusted” numbers. Meaning some model was used. This is odd because they do show the raw numbers for the other outcomes, like mask compliance rates (obviously higher in those who were reminded to wear them).
Pay attention now. The adjusted rates in those with symptoms only were this: control prevalence = 0.76% and treatment prevalence = 0.68%. Thus, at best, and only for people with “symptoms”, free masks and mask education reduced prevalence by 0.76 – 0.68 = 0.08%. Real percent, not relative percent.
I’ll repeat that: the reduction, at best, and assuming no Simpson’s paradox, and assuming nothing screwy with the strange “symptoms” definition and sampling, and assuming nothing screwy with the adjusting model (huge assumption!), was only 0.08%. That is zero-point-zero-eight percent. And only in those with “symptoms.”
Important. It appears from the text they grabbed all people with symptoms: “We collected capillary blood samples from participants who reported
COVID-like symptoms during the study period.” This is 3.2% of the population.
That means the population reduction of masks, given all those other caveats hold, is 0.032 x 0.0008 = 0.000026, or 0.0026%.
Masks reduce seroprevalence in the population by 0.0026%. At best. And assuming zero has gone wrong in all the experiment and modeling steps, and assuming the only cause active was masks, and that no other behavior changed in the mask group.
Since we can’t be sure of all that, there is some margin of error in that 0.0026%. Meaning the real reduction could be, and likely is. even smaller.
Since seroprevalence is not a direct measure of disease severity, and we know only a fraction of those with the bug become seriously ill, even fewer than that “at best” 0.0026% are protected against illness. No better than, say, 1%, on average across all ages become seriously ill or die.
That means masks protection against serious illness is 0.01 x 0.000026 = 0.00000026, or 0.000026%.
Conclusion: Skip mask mandates. They do nothing.
Bonus rant Stop looking at relative rate ratios! Look at comparisons of actual rates. It’s too damned easy to fool yourself with proportions. As people have done here.
A rate change from 0.0000000002 to 0.0000000001 is 50%. Wow! 50%! Halved! Alert the media! Yet the real reduction is 0.0000000001.
Stop using relative numbers in cases like this.
Subscribe or donate to support this site and its wholly independent host using credit card or PayPal click here