Modelling safe protocols for reopening schools during the COVID-19 pandemic in France

The COVID-19 pandemic prompted various strategies across Europe for lifting lockdowns, with significant debate surrounding school closures and reopenings. Different countries adopted diverse approaches, ranging from keeping schools closed until September (Italy, Spain) to phased reopenings with limited attendance or alternating classes (Denmark, Norway, Austria, Greece). France implemented a progressive reopening plan starting with pre-schools and primary schools, followed by middle and secondary schools, contingent on regional infection rates. The challenge in assessing the risk of school reopening lies in the uncertainty regarding children's role in COVID-19 transmission. Early evidence suggested children were less susceptible to infection and more likely to be asymptomatic or paucisymptomatic, although their role as infection sources remained unclear. Some studies highlighted a lower COVID-19 incidence in younger children (under 10) compared to adolescents and adults, suggesting younger children may have a weaker role in transmission. This study aimed to model the impact of different school reopening protocols on epidemic control in Île-de-France, considering the uncertainty around children's transmission role. The research was conducted during the lockdown, relying on scenario analysis and later supplemented by an ex-post assessment using data available after the initial submission.

The literature review section of the paper examines existing studies on children's role in COVID-19 transmission. It cites several studies indicating children's lower susceptibility and higher likelihood of asymptomatic or paucisymptomatic infections. However, the role of children as sources of infection remained unclear, partly due to asymptomatic infections, limited testing during the early pandemic phase, and early school closures. The review also highlights studies showing that adolescents might play a different role in transmission compared to younger children. Findings from Iceland, Italy, and South Korea show lower COVID-19 incidence in younger children than adolescents and adults. A retrospective analysis of a cluster in northern France showed evidence of significant asymptomatic viral circulation in a high school. The review provides a basis for the study's focus on school contacts and the impact of different reopening protocols, considering the uncertainty surrounding children's role in COVID-19 transmission dynamics.

The study employed a stochastic discrete age-structured epidemic model, using demographic and age profile data for Île-de-France, divided into four age classes (0–11, 11–19, 19–65, and 65+). Social contact matrices from 2012 were used to model mixing between age groups in various settings (household, school, workplace, etc.). The model incorporated COVID-19 specific compartments (susceptible, exposed, infectious, hospitalized, ICU, recovered, deceased), with the infectious phase divided into prodromic and asymptomatic/symptomatic phases. The model assumed children were half as susceptible as adults, with younger children less likely to transmit than adolescents, exploring four transmissibility reduction values (r0-11 = 0.1, 0.25, 0.33, 0.55). Intervention measures (lockdown, moderate interventions) were modeled by modifying contact matrices. The model was calibrated to hospital and ICU admission data up to April 26, 2020, estimating the reproduction number before and during lockdown. Different school reopening scenarios were simulated: (1) reopening only pre-schools and primary schools on May 11 using progressive (25%, 50%, 75%, 100% attendance) or prompt (50%, 100% attendance) protocols; (2) reopening pre-schools/primary schools on May 11 (Progressive 100%) followed by middle/high schools on June 8 using progressive or prompt protocols; (3) reopening all schools on May 11 using various protocols. The scenarios were compared to a school closure scenario with moderate interventions and case isolation. Each scenario was assessed based on new clinical cases at the start of summer holidays and ICU bed demand on August 1. 500 stochastic runs were performed for each scenario, and a sensitivity analysis was conducted examining the impact of varying case isolation rates, reproduction numbers, and contact reduction in adolescents. An ex-post assessment using data after lockdown compared ICU admission data to different exit scenarios, fitting the model to observed data to estimate the avoidance of physical contacts.

The model projected that reopening only pre-schools and primary schools on May 11 would increase new clinical cases by 2 to 2.4 times compared to the school closure scenario, with ICU occupancy reaching a maximum of 62% [54, 68]%. Reopening middle and high schools on June 8, even with limited attendance, would result in similar epidemic activity. Full attendance in middle and high schools on June 8 could lead to ICU occupancy reaching 76% [67, 84]%. Reopening all schools on May 11 with full attendance would likely saturate ICU capacity. Progressive reopening protocols showed only limited benefits in reducing the epidemic peak. The study showed the importance of testing and isolating cases. A 50% case isolation rate would lead to exceeding ICU capacity. The ex-post assessment indicated that a high adherence to physical distancing measures (estimated at 90% avoidance of physical contacts), coupled with limited school attendance (primarily in younger grades), contributed to the observed decrease in the epidemic after lockdown. The study highlights that large-scale testing (at least 100,000 tests per day in Île-de-France) and contact tracing are critical for effective epidemic control in conjunction with even moderate social distancing measures.

The findings suggest that while school reopening increases COVID-19 cases, it's possible to manage the epidemic without overwhelming the healthcare system by implementing specific protocols. Prioritizing pre-schools and primary schools allows younger children to resume learning while minimizing ICU strain. Full attendance in middle and high schools is not recommended unless the community transmission is low. The study stresses the importance of large-scale testing, contact tracing, and adherence to preventive measures. The significant adherence to physical distancing measures observed in the ex-post analysis highlights the crucial role of public health communication and preventive measures in epidemic control. The results emphasize the need for a flexible approach to school reopening based on the current epidemic situation (reproduction number, community transmission levels). While the model showed some differences in the effects of progressive vs prompt reopening, the overall effect is dependent on the extent of the physical distancing and the speed and scale of case finding and isolation. The model's assumptions regarding the transmissibility of younger children show to have minimal impact on the overall findings. The significant needs for testing suggest major investments to support the scaling up of diagnostic and testing activities and surveillance.

This study provides valuable insights into safe school reopening protocols during a pandemic. Prioritizing younger children, implementing limited attendance in older grades, and aggressive testing and contact tracing are critical. Future research should focus on refining the understanding of children's roles in transmission across different age groups and the efficacy of different mitigation strategies, including the use of masks and ventilation improvements. Further investigation into the long-term effects of school closures and the effectiveness of various testing and tracing strategies in different contexts is also warranted.

The study's projections relied on data up to April 28, 2020, and assumed a constant reproduction number during lockdown, which might not reflect actual dynamics. The model did not explicitly incorporate mask usage (not yet widespread during the studied period) or seasonal variations in viral transmission. Contact data from 2012 might not fully reflect the actual contact patterns in 2020. The summer period's parameterization had uncertainties due to data limitations. The analysis focused on Île-de-France and did not consider spatial targeting within the region or the emergence of new variants. The model does not include reactive school closure as a mitigating factor.