Science’s COVID-19 reporting is supported by the Heising-Simons Foundation.
A careful analysis of health data from Sweden suggests keeping schools open with only minimal precautions in the spring roughly doubled teachers’ risk of being diagnosed with the pandemic coronavirus. Their partners faced a 29% higher risk of becoming infected than partners of teachers who shifted to teaching online. Parents of children in school were 17% more likely to be diagnosed with COVID-19 than those whose children were in remote learning.
Whether the harms of school closures outweigh the risks of virus transmission in classrooms and hallways has been the subject of intense debate around the world. Outbreaks have demonstrated that the virus can spread via schools to the wider community at least occasionally, and some data suggest teachers have higher than average risk of infection. However, it has been difficult to separate school-based transmission from other confounding factors, especially because schools have tended to open or close in concert with other restrictions lifting or tightening.
Coming out the same week as new guidelines for opening schools from the U.S. Centers for Disease Control and Prevention, the new study will help policymakers better understand and weigh the risks and benefits. “It’s just great to see such a carefully done study,” says Anita Cicero, an expert in pandemic response policy at the Johns Hopkins University Bloomberg School of Public Health. “We’ve been starved for studies” that quantify the impact of open or closed schools on wider community transmission.
In March 2020, schools around the world closed as governments tried to keep SARS-CoV-2 in check. But children in Sweden through ninth grade continued to attend class, while 10th through 12th graders shifted to remote learning. Economists Jonas Vlachos, Helena Svaleryd, and Edvin Hertegård at Uppsala University took advantage of this natural experiment and Sweden’s detailed health care data.
They compared infection rates of parents whose youngest child was in ninth grade with those whose youngest was in 10th grade. They also compared infection rates in teachers who continued to teach in person at lower secondary schools (grades seven to nine) with those of teachers at upper secondary schools (grades 10 to 12), who taught remotely. Finally, they compared infection rates in the spouses of teachers in the two types of schools. They describe their results in a paper posted on 12 February in the Proceedings of the National Academy of Sciences.
The authors took steps to make sure their groups were as comparable as possible. For example, they excluded families with health care workers from the study because they had more exposure to the virus and were tested more frequently. Sweden’s coronavirus testing was very limited in the spring, swabbing only people with moderate to severe symptoms. Although this missed many cases, Vlachos says, it was actually an advantage for their analysis. As testing increased in the summer and fall, testing rates started to correlate more with income, which would have skewed the findings. (So few children and teens were tested that the researchers couldn’t draw conclusions about their infection rates.)
Swedish schools instituted only relatively mild precautions against infection in the spring. Health authorities encouraged pupils and teachers to wash or disinfect their hands regularly, keep their distance when possible, and stay home when ill. But neither teachers nor students wore masks, and close contacts of confirmed cases were not quarantined.
The impact on teachers was significant, the authors say, and the results underscore the need to prioritize educators in COVID-19 vaccination schedules. Whereas teachers at upper secondary schools had an average infection risk among 124 occupations in Sweden, the researchers found, lower secondary teachers ranked seventh. (Primary school teachers had a somewhat lower, but still above average, risk.)
Among the country’s 39,000 teachers in lower secondary schools, 79 were hospitalized with COVID-19 between March and June 2020, and one died. Shifting these schools to online learning would have prevented perhaps 33 of those severe cases, the authors estimate.
Adding masks would likely have reduced the risks to both teachers and families, says Danny Benjamin, a pediatrician at Duke University who has studied the spread of the pandemic coronavirus in North Carolina schools. But the Swedish study shows that “even if schools do not require masking, risk to families of in-person schooling is low,” he says.
Vlachos agrees that more interventions would reduce risk further. “Our estimates are likely an upper bound,” he says.
The authors calculated that keeping lower secondary schools open likely led to 500 additional detected cases in the spring among the 450,000 parents with kids in lower secondary school and 38 additional cases among teachers’ partners. (Because testing was so limited, the real number of additional infections was likely much higher, the authors note.)
“The results for parents provide perhaps the best evidence of how school closure impacts virus transmission in society,” says Douglas Almond, an economist at Columbia University. By comparing families with ninth graders and 10th graders, the team was able to compare families with teenagers whose social behavior and viral risk were similar, he says. “This is where their natural experiment really shines.” The ability to link teachers to their spouses through the health registries “is also quite elegant,” says Jonas Björk, an epidemiologist at Lund University.
“It is to be expected that opening schools can increase COVID-19 infections, but knowing that does not really inform policy,” Almond says. “One needs to know how much infections increase due to school reopening. This is the best paper I know of that quantifies this effect.”
More comparisons of schools with different policies regarding masks, distancing, and quarantines would be helpful, Cicero says. Using the Swedish health registry, the researchers could even take the analysis a step further and look at risk to parents of teachers, Björk says, which would help estimate the impact on a more vulnerable age group.
The emergence of more transmissible variants of SARS-CoV-2 means mask wearing and other interventions to prevent school transmission are even more important, Benjamin says. Cicero agrees. “That is the next step” for research, too, she says: funding studies on the impact of the variants, and which interventions can keep risks at schools as low as possible.
*Correction, 16 February, 2:35 a.m.: This story has been corrected to say the study could be extended to parents of teachers, not grandparents of students.