https://nymag.com/intelligencer/2021/08/the-science-of-masking-kids-at-school-remains-uncertain.html
The author treats the Georgia study as if the researchers found evidence that student masking is not effective. On the contrary, the researchers report evidence that student masking is effective: risk of COVID was estimated to be 21% lower for students in schools with required masking, all else being equal. Because of uncertainty in this estimate, values ranging from 50% lower risk with masks to 8% higher risk with masks would also be consistent with the data (i.e., the 95% confidence interval for the risk ratio is 0.50-1.08). But our best estimate from these data is that masks reduce risk by 21%. The author of the article makes the ubiquitous mistake of confusing lack of a “statistically significant” effect with evidence that there’s no effect. The statistics community has written extensively about this issue (e.g., in a special issue of The American Statistician; see https://www.tandfonline.com/doi/full/10.1080/00031305.2019.1583913)
There’s a claim in the article that “risk to kids from COVID is vanishingly low.” As a percentage of total cases, 470 COVID-related deaths in children looks small. Any number can be made to seem small by dividing it by a large enough denominator; this is a common ploy in efforts to downplay the seriousness of COVID. And death isn’t the only concern; pediatric hospitalizations have soared to all-time highs in recent weeks, we’re still learning about MIS-C and long COVID, and infected kids can infect vulnerable adults. (Note: the author cites NCHS data on CDC’s site for 361 COVID deaths in kids. There are reporting lags in these data. The number on CDC’s more up-to-date Data Tracker is 470.)
The author states that staff-to-staff transmission is more common than transmission involving students, as if the greater threat posed by the former is somehow relevant to the threat of the latter. Relativizing the threat of COVID by comparing it to some other cause of death is another common tactic.
The author cites British data suggesting Delta doesn't lead to worse outcomes; studies from Scotland and Canada suggesting otherwise aren’t mentioned. Cherry-picking studies to support a particular position can be misleading. Good researchers evaluate multiple studies and compare their findings to get a complete picture of what we know, and don’t know.
The author argues that cloth masks have limited effectiveness and then jumps to the idea of forcing children to wear N95s tightly all day. These aren’t the only two alternatives.
I’ll stop there and turn to limitations in the Georgia study:
https://www.cdc.gov/mmwr/volumes/70/wr/mm7021e1.htm
Surveys were mailed to 1,461 schools in 159 counties; only 169 (11.6%) schools in 51 (32.1%) counties chose to respond. This raises questions about selection bias (due to schools self-selecting for participation) and representativeness of the sample.
Schools were not randomized to different conditions; local schools/districts decided whether they would mask or not mask, etc. This makes it difficult to infer causation: Are observed differences in COVID rates between schools who did vs. did not adopt a particular prevention measure attributable to use of the prevention measure, or due to potentially unobservable, pre-existing differences between schools that determined whether they chose the prevention measure, or both? This is why we do randomized controlled trials; randomly assigning each participant to condition gives us groups that are comparable, which allows us to attribute any difference in outcomes to the study condition.
Presumably mask use in schools without required masking varied widely. Apparently this wasn’t accounted for in the analysis. We might not expect much difference between a school with 80% masking and one with required masking. Ideally, to get at the question of whether masks work, we would compare a school with 0% masking to one with 100% masking; but it would be unethical to ask a school to ban masks altogether for research purposes.
As always, I’m going with the consensus of the medical and scientific community. There may be differences of opinion around the edges (e.g., on universal masking for children ages 2-5; CDC recommends this, WHO doesn’t). But on the basic facts and recommendations there’s overwhelming agreement.