Most people agree that in-person education is best for kids, but we lack information on risks, especially in the US.
Economist Emily Oster has launched a platform looking at enrollment, reopening information, and virus mitigation in 550 schools (and counting) across the US.
Among in-person students, the confirmed COVID-19 infection rate is 0.078%, or a bit less than 1 in 1000 students. For staff, it's 0.15%.
Should schools open? Which schools? Where? With masks? What about ventilation? Outdoor classes? Does it matter what the infection rate is in the area? Are kids superspreaders?
It is difficult to overstate the level of interest in schools during the pandemic, most notably among parents and policymakers, but even going beyond that. We worry about public health — spread in schools, and resulting spread out of schools — but also about the negative effects on children of not being in school, to say nothing of the issues facing parents with children in remote school.
There is, I think, widespread agreement that in-person education is best for most children, and if we can open schools safely, we should do so. Disagreements arise most vividly on the question of whether that can be done, and how. Part of what has made these disagreements difficult to resolve is a lack of information on risks.
Our best data on schools, so far, comes from Europe. But the situation is different there — lower infection rates, different school structure. We have a bit of data on child care settings in the US, although child care and school are not the same. The bottom line is that there is no substitute for actual data from US schools.
The US data we have so far has focused on case counts. That is: the number of cases at the school or district level. The New York Times probably has the most comprehensive look at this.
There is some value in such counts. But from the standpoint of learning about risks in school, or appropriate mitigation policies, they fall short.
Understanding COVID risks requires not just counts, but denominators. If there are 10 cases, it matters a lot whether there are 20 people or 1000 people in the school.
At a minimum, we should know if the school was open for in-person instruction (case counts are often reported even for remote learning). Are staff in the building without students? What kind of hybrid model? Are people masking?
We need context.
A new platform is collecting coronavirus-related data from schools across the US
That's what I am trying to provide, with a team from Qualtrics and various school associations. Our goal is to start the data collection at the school level, looking at enrollment information, reopening information, and mitigation protocols. Then we layer case reports on top of that.
We want to give people the opportunity to really dig inside the data and get into the context, rather than just seeing a topline report number.
To do this, we've built a platform — a data entry system for schools which starts with the baseline data and then adds on cases. These data feed into a dashboard, which shows some topline numbers on infection and reopening plans, but also lets users filter the data, and even run their own analyses.
These data will update over time. As new schools opt in and provide their data, this will happen automatically. Later waves of the data will allow for analysis of over time variation. I'll be writing more here on what we are learning.
So far, the in-person student infection rate is less than 1 in 1000
The dashboard as it stands now covers 550 schools across the US, with 340,000 students and 40,000 staff. About 100,000 of these students are in person, along with a much larger share of the staff (there turn out to be a lot of schools which are fully remote but with in person staff).
Among these in-person students, the confirmed infection rate is 0.078%, or a bit less than 1 in 1000 students. For staff, this is 0.15%. If we include suspected infections, these numbers are 0.23% and 0.49%. Suspected infections will include some cases which are not COVID-19, but also capture cases where testing isn't available. I think of these as two bounds.
These numbers are a bit higher for high school than the other age groups. They're a bit lower in private and parochial schools. Eventually, it will make sense to look by states and even counties as the sample size grows.
I know we will be asked: are these infection numbers small or big? I don't think there is one answer to that, since there is not a benchmark for an "expected" number. One possible benchmark is the testing positivity rate, which other analyses have used to predict the number of cases that would arrive in schools as they open. The numbers here are lower than those testing positivity rates would suggest.
There is more to be done! I hope to use this space to visit more detailed analyses of the data, provide context, and answer questions. So send thoughts, send ideas, send suggestions to email@example.com.
Emily Oster is a Professor of Economics at Brown University and the author of Expecting Better and Cribsheet.
Read the original article on Insider