Out in PNAS: "The effects of school closures on SARS-CoV-2 among parents and teachers". On March 18, 2020, upper secondary schools in Sweden moved online while schools for younger students remained open. 1/n https://www.pnas.org/content/118/9/e2020834118

This allows for a comparison of groups differently exposed to open/closed schools but otherwise facing similar conditions. Using register data, we link the entire population to all PCR-confirmed cases as well as Covid diagnoses from healthcare visits and hospitalizations. 2/n

We analyze the period from school closure until the summer break in mid-June. During the period, infections rates were very high, testing capacity low and testing highly restricted. Biases due to differences in the propensity to get tested is hence less of a concern. 3/n

We compare parents with the youngest child in school years 9 (open) and 10 (online) and find that exposure to open schools resulted in a 17% higher risk of infection. Around 450000 parents (4.5% of the population) are exposed to children in lower secondary schools 4/n

According to the estimates, moving lower secondary schools online would have resulted in would have resulted in a 17% decrease in infections among 4.5% of the Swedish population. Keeping lower secondary schools open thus had a minor impact on the overall transmission. 5/n

We control for a host of factors but the estimates are very similar with and without these controls, which is expected since essentially all students attend both lower and secondary school in Sweden. We do not extrapolate to parents of younger children. 6/n

We then compare lower and upper secondary teachers, arguing that upper secondary teachers are a good counterfactual for lower secondary teachers, had their schools moved online. We find that infection rates and hospitalizations doubled among teachers exposed to open schools. 7/n

In numbers, we find 79 hospitalizations among 39500 lower secondary teachers until the end of June, 2020. Estimates suggest this number had been down 46 if lower secondary schools had moved online. 8/n

Partners of lower secondary teachers had a 30% higher rate of PCR-confirmed infections that their upper secondary counterparts. Estimates for hospitalizations are noisy, however. 9/n

Comparing 124 occupations, PCR-detected infections among upper secondary teachers (online) were at the median, while lower secondary teachers (open) were the 7th most affected. (Healthcare occupations are not part of this comparison as they were targeted for testing.) 10/n

The figure indictates teachers at the lower primary (years 1-3) and upper primary (4-6) level. Differences could reflect increasing risk in student age, but also that primary school teachers meet fewer students or different modes of interaction among teachers. 11/n

The patterns among different groups of teachers suggests that the impact of, for example, opening upper secondary schools may be different from opening schools at the lower secondary level. 12/n

The results are from a setting in which precautionary measures in schools were mild; no quarantine of those exposed without symptoms of infection, no imposed class-size reductions, and no use of face masks. Also, schools were not targeted for testing. 12/n

Even if keeping schools open appears to have had a minor impact on the overall rate of infections in Sweden during the spring of 2020, the results for teachers suggests that precuationary measures could be considered. 13/13