The potential impact of School Closure Relative to Community-based Non-pharmaceutical Interventions on COVID-19 Cases in Ontario, Canada

Naimark, D. M.; Mishra, S.; Barrett, K.; Khan, Y.; Mac, S.; Ximenes, R.; Sander, B.

Importance: Resurgent COVID-19 cases have resulted in the re-institution non-pharmaceutical interventions, including school closure, which can have adverse effects on families. Understanding the impact of schools on the number of incident and cumulative COVID-19 cases is critical for decision-making. Objective: To determine the quantitative effect of schools being open relative to community-based non-pharmaceutical interventions on the number of COVID-19 cases. Design: An agent-based transmission model. Setting: A synthetic population of one million individuals based on the characteristics of the population of Ontario, Canada. Participants: Members of the synthetic population clustered into households, neighborhoods or rural districts, cities or a rural region, day care facilities, classrooms (primary, elementary or high school), colleges or universities and workplaces. Exposure: School reopening on September 15, 2020, versus schools remaining closed under different scenarios for non-pharmaceutical interventions. Main Outcome and Measures: Incident and cumulative COVID-19 cases between September 1, 2020 and October 31, 2020. Results: The percentage of infections among students and teachers acquired within schools was less than 5% across modelled scenarios. Incident case numbers on October 31, 2020, were 4,414 (95% credible interval, CrI: 3,491, 5,382) and 4,740 (95% CrI 3,863, 5,691), for schools remaining closed versus reopening, respectively, with no other community-based non-pharmaceutical intervention; 714 (95%, CrI: 568, 908) and 780 (95% CrI 580, 993) for schools remaining closed versus reopening, respectively, with community-based non-pharmaceutical interventions implemented; 777 (95% credible CrI: 621, 993) and 803 (95% CrI 617, 990) for schools remaining closed versus reopening, respectively, applied to the observed case numbers in Ontario in early October 2020. Contrasting the scenarios with implementation of community-based interventions versus not doing so yielded a mean difference of 39,355 cumulative COVID-19 cases by October 31, 2020, while keeping schools closed versus reopening them yielded a mean difference of 2,040 cases. Conclusions and relevance: Our simulations suggest that the majority of COVID-19 infections in schools were due to acquisition in the community rather than transmission within schools and that the effect of school reopening on COVID-19 case numbers is relatively small compared to the effects of community-based non-pharmaceutical interventions.