A year ago the CDC published a study of mask-wearing in elementary and secondary schools. They collected data from 565 counties and produced the following results:
With a couple of small exceptions at either end of the chart, schools that required masks were correlated with lower rates of COVID-19 cases in children.
A pair of researchers recently went back to duplicate the CDC study—which they did—and then to extend it to 1832 counties using newer data. Here's what they found:
With a broader set of cases, they mostly found just the opposite of the CDC study: schools that required masks were correlated with higher rates of COVID-19 cases in children.
There are all sorts of methodological issues with these studies, primarily the fact that it's a correlation between schools and counties, which might or might not mean anything. It's likely, for example, that schools which require masking differ in other non-random ways that might produce the differences seen here. It's also important to keep in mind that these studies are solely about masking in school, not masking in general.
That said, what conclusions can we draw? There are probably two:
- When you get beyond week 0 (the start of school) the differences between masking and non-masking schools are pretty trivial in the bigger study.
- Expanding the study changed the results fairly dramatically.
This makes it likely—though not proven—that school masking produces a very small difference that could easily be modestly negative or modestly positive. If I were a school board member using these studies to help my decision making, I'd probably vote to make masking optional. The benefit, if any, is probably not worth the likely degradation in the teaching environment.
But I would require vaccination.
As you point out, these are observational studies and not controlled studies, so we can't distinguish correlation from causation.
The mistake is in the last graf, where you claim that the evidence is still worth something, even if it ought to be discounted slightly. It's not. These charts have essentially zero value. The principle here seems to be "well, it's better than nothing" and it's not. Garbage data is not better than no data at all.
A quick ELI5 for anyone who doesn't get the weakness of observational studies, suppose you want to find out if consuming alcohol is good for your health.
In a controlled study, you'd round up a bunch of volunteers and then assign them randomly to the "zero alcohol", "moderate drinking" and "heavy drinking" groups. Then study the results. This is a controlled study.
In an observational study, you'd round up a bunch of volunteers, ask them how much they drink, and study the results. This is an "observational" study.
Your controlled study will show you that more alcohol = more health damage.
Your observational study will show something odd: That people who drink a small bit are healthier than those who drink none at all, and then reporters will write stories about how a little bit of booze is good for you. But the reason this is happening is that there's a bunch of people too sick to drink.
Anyway, the point is: Your experiment isn't random. And so the thing you're trying to study as an output--the health damage--is sneaking into your experiment as an input because you're letting people choose which group they're in, and some of them will make that decision (or have it made for them) on the basis of their health.
+1
Following on Laertes about variables being confounded, don't the newer set of data show that school districts that required masks were the ones that had higher disease burden in the weeks before school? Ie, if masks are helpful at reducing disease, and we know it, then places experiencing high disease burden would be more likely to be the ones who impose the mask mandate.
Then, in the first weeks of school, those places with already-high covid case counts remained higher than the non-masking schools. But those new data sure do look like the covid cases in non-masking counties nearly caught up to the masking counties.
So, yes, an observational study, but it might also be consistent with real effectiveness of mask mandates at reducing pediatric spread. Or am I misunderstanding the data?
This data is meaningless. Just some confounding issues off the top of my head:
- Counties overlap school districts in some states (especially in the South), but mostly they don't.
- Just because a school district has a policy doesn't mean a particular school or classroom is following it.
- School districts might be adopting mask policies in response to rising local covid numbers, so causation is backwards here.
- The data is covid numbers for the whole county population versus school district masking, not covid numbers for children enrolled and their families. So a county could have a spike at nursing homes, and this study would link that to school masking.
It's a garbage study. Put a mask on.
"Just because a school district has a policy doesn't mean a particular school or classroom is following it."
Okay, that one occurred to me too. These data are about the effects of "mask mandates" and not the effects of "masks."
But!
That seems like it's not a fatal flaw if the question you're asking is "should I do a mask mandate?" In that case, you don't really care if masks work. You care if mask mandates work.
yeah but if you're looking at the level of a disease spreading, it has to be at the school building level. district means nothing, because students (and 99% of personnel) are per building, and that's how diseases spread. My district has 24,000 students in 81 schools, examining a mask mandate across the district would make no sense.
Notice that the graph begins at minus 3 (as in weeks before school starts) and could probably be interpreted as follows: Districts that already had a high level of Covid-19 were more likely to respond by requiring masks. As the school year went on, the number of cases increased over the whole population, but the masked population grew less quickly and therefore equalled the unmasked population. Conclusion: Masks did something but were not completely protective of the whole population
the linked study makes the point that both studies are worthless and cannot be used to show casual relationships. Hopefully there are zero school districts considering a study that tells the reader not to use it draw conclusions.
If I'm reading this right, the paper shows a strong correlation between adult and pediatric infection rates.
Masks in schools won't stop infections in the community, only minimize transmission at school (presuming masks are worn on buses too).
I agree with the commentators here that is hard to conclude anything from the source material. The second chart has masked school systems with a higher rate Covid rate before school starts and then it evens out. So if anything, masks have a small but measurable impact in the reducing the spread of the disease. A Covid shot would probably more effective, where is a charts for that? So, if you are student you either take the shot or wear a mask.
On one hand, this is clearly reverse causation, where places with higher community infection rates are more likely to have mask mandates in schools. On the other hand, if there was an *enormous* positive effect from masking it should still show.
I wonder what a lagged study would look like -- a month after a mask mandate goes into effect, what does the case rate look like?
Pretty anecdotal here, but I will say that school districts near me varied their preventive measures based on disease prevalence. Big outbreaks resulted in distance learning for periods of time, for example. That could affect data.
Also, in the elementary school that I was in every day, it was very clear that COVID wasn't being transmitted between masked kids. You could tell based on the lack of pattern of transmission. Kids would go home with COVID, but their tablemates would stay well. My kid didn't even catch it from her best friend, who sat next to her all day. When I did hear directly about transmissions, it was mostly family (especially family get-togethers) followed by sports.
Relevant discussion here: https://statmodeling.stat.columbia.edu/2022/12/14/lack-of-correlation-between-school-mask-mandates-and-paediatric-covid-19-cases-in-a-large-cohort/
I am all for vaccine mandates, where the jab actually stops or substantially slows transmission, which they do not. They are great at protection against serious illness (triple jabbed and bolstered here) but by not preventing infection, you kind of lose the argument of unvaccinated being of greater risk to others, not to mention immunity from actual infection. I get the flu shots every year, but I don’t think they should be mandated
Vaccines may not stop the transmission but they do slow it down. Protection against infection was 90% in the early going (before the variants spoiled the effect); they are still in the 50 - 60% range. Meaning a fully vaccinated population could slow down the spread significantly, making it easier for doctors, nurses and hospital to care properly for the unlucky ones who do get sick. Not to mention that fewer serious cases also helps in the same way.
This alone makes vaccine mandates worth having; it could go along way to minimize economic damages from a pandemic, apart from the benefit for the individual.
Not to mention, this study clearly defies the common sense test.
Outside of fever swamp propaganda, there is no biological mechanism by which wearing a mask would increase the chance of contracting Covid. The possibilities are "reduces the chance of Covid" or "no difference".
So the fact that the results point in a nonsensical direction says they are clearly measuring confounding issues of some sort.
Bingo. My first thought too.
That's not remotely true. If masking led people to undertake substantially riskier behaviors then it might well cause a counterintuitive effect.
cf anti-lock brakes.
What sort of behavior for example? What could a masked person do that would raise the risk (short of taking the mask off)?
Attend large and crowded gatherings with a mask on that they wouldn't with a mask off. Go to things when they're feeling sick with a mask on that they wouldn't with a mask off. Think that any mask is fine and thus wear an ill-fitting or unprotective one and go places that they wouldn't without the mask.
"Garbage data is not better than no data at all."
I must thank Laertes for this...One of the great and very true sentences of all time.
I will be using it often.
Best Wishes, Traveller
both studies do appear to suggest that the beginning of school boosts covid transmission.
this isn't surprising and it would defy all logic to say that gathering a bunch of people inside for 8 hours would not increase covid spread.....but I do remember there being many loud voices insisting that putting kids in school did not lead to an increase in covid transmission because reasons....
Oh yes, who could forget one of Kevin’s hobby horses at the time…
I would really like these discussions to stop lumping together elementary schools with middle and high schools. I just don't think it's apples-to-apples for drawing any meaningful conclusions.
* Elementary kids are cohorted, spending the majority of their day in the same room with the same people, which differs from the levels of exposure someone bounces around all day between different classrooms with a different mix of people
* Elementary kids are younger so their maturity levels may have an impact on how compliant they are with hygiene measures and how effectively they wear masks
* Younger kids have different immune systems and susceptibility to Covid than teenagers
studies like this are useless by now.
You cannot find any kind of uniform groups to work with. For example, children who do not man in school may have lower COVID rates going forward because their parents do not bother to test them for COVID when they get sick - what is the benefit and these are the parents who no longer care about COVID.
Comparing COVID mortality would have more value, but COVID mortality among children without severe disabilities or comorbidities is so rate that most of your numbers would be zero unless your dataset was larger than the US school age population.
That is also why Vax mandates are not justified. Young boys have a high enough side effect rate with vaccines to make the cost benefit a toss up.
Here is a single data point. My family all got COVID in the same week, presumably same strain spreading in our home. I have two shots, my significantly younger wife has three, and our preteen sons are unvaxxed.
I was seriously sick for just 12 hours. My wife was worse and it lasted 24 hours. Our unvaccinated sons were moderately sick for two days each. Both felt unwell, neither had fever. All of us tested positive with a home rapid test. We did not report our cases to anyone or see a doctor.
Vax mandates make sense for people who work with vulnerable people - hospital and nursing home workers for example. Maybe even visitors to such places. They do not make sense for school children or teachers.