Implications of the school-household network structure on SARS-CoV-2 transmission under school reopening strategies in England

The COVID-19 pandemic forced many countries to close schools in early 2020 to curb the spread of SARS-CoV-2. As governments worldwide sought to ease these closures, understanding the associated risks became crucial. This research directly addresses this need by investigating the potential for SARS-CoV-2 transmission between schools in England under various reopening scenarios. The study's importance lies in its potential to inform public health policy and strategies for safely reopening schools while minimizing the risk of widespread outbreaks and subsequent disruptions to education and the wider community. The consequences of uncontrolled outbreaks in schools extend beyond the immediate educational setting, impacting families, healthcare systems, and the overall economy. Therefore, a thorough understanding of transmission dynamics within and between schools is vital for developing effective mitigation strategies. This paper focuses on utilizing a novel approach to model school-to-school transmission to assess the risks associated with different school reopening strategies.

Existing literature on school closures and infectious disease transmission provided a foundation for this study. Studies examining the impact of school closures on influenza transmission highlighted both the effectiveness of closures in reducing spread and the associated economic and social costs (Wu et al., 2010; Cauchemez et al., 2008; Sadique et al., 2008; Berkman, 2008; Wong et al., 2016; Bayham & Fenichel, 2020). However, the unique characteristics of SARS-CoV-2, including its transmissibility and the potential for asymptomatic spread, required a specific investigation into school reopening strategies in the context of this novel virus. The literature also revealed varying findings on the role of schools in SARS-CoV-2 transmission, with some studies suggesting a limited contribution to overall transmission (Macartney et al., 2020; Covid-19 in schoolchildren-A comparison between Finland and Sweden-Folkhälsomyndigheten, 2020), while others indicated a potentially significant role (Ismail et al., 2021; Flasche & Edmunds, 2020). These discrepancies in findings emphasized the need for further research using refined methodologies to better understand the complexities of school-based transmission.

This study utilized a novel approach to model SARS-CoV-2 transmission between schools in England by constructing a network of schools connected through households. Data on pupils attending state-funded schools in England, including school URN, postcode, pupil's postcode, and address, were obtained from the UK Department for Education under a formal data-sharing agreement. The data, collected between September and December 2019, allowed the researchers to identify households with children attending multiple schools. A household code was assigned to each group of pupils living at the same address, with a validation process comparing assigned codes to official unique address codes. This process allowed for the creation of a network where each node represented a school and edges represented connections through shared households. The weight of each edge was determined by the number of unique contact pairs between schools within households (Equation 1). Subsequently, a transmission probability network was created, estimating the probability of transmission between schools (Equations 2-5). The probability of transmission was calculated by considering the probability of an outbreak in one school seeding an outbreak in an adjacent school. The model incorporated the within-school reproduction number (R), representing the average number of secondary infections within a school, and the per-contact probability of transmission between children within households (q), which was set to 0.15, consistent with estimates of household secondary attack rates for SARS-CoV-2. The analysis was repeated for a range of R values (1.1-1.5) to assess the impact of within-school transmission rates on the overall network. To summarize the potential spread of the virus, the researchers sampled binary outbreak networks where transmission between schools either occurred (edge weight 1) or did not occur (edge weight 0). The distribution of connected components within these networks provided insights into the potential outbreak sizes (measured in terms of number of schools and households affected). The analysis evaluated six different school reopening scenarios, varying which year groups returned to school (Fig. 2 and Table 1), allowing for a comparison of risks associated with different reopening strategies. The weighted degree distribution (Equation 6) was calculated to illustrate the expected number of schools infected by each school.

The study's key findings highlight the significant difference in transmission risk between primary and secondary school reopening scenarios. Reopening a limited number of year groups (Reception, Year 1, Year 6) presented a low risk of large-scale transmission between schools, with simulated outbreak clusters typically affecting fewer than 10 schools or 1000 households (Fig. 5). However, including even a small subset of secondary school years (Years 10 and 12) considerably increased the connectivity of the network and the potential for large-scale outbreaks. Opening secondary schools alone resulted in the highest connectivity and potentially impacted up to 2.5 million households at an R value of 1.5 (Fig 4 and Table 2). The number of households impacted increased significantly with the within-school reproduction number (R), indicating the importance of implementing strong infection control measures within schools to keep R low (Fig. 3 and Fig 4). While most schools remained in smaller outbreak clusters even at higher R values, certain parts of the network demonstrated significantly higher connectivity, potentially leading to geographically concentrated outbreaks. The findings emphasized that the risk of larger outbreak clusters was considerably higher for scenarios involving secondary school reopening compared to primary school reopening (Table 2 and Fig. 5). The analysis considered the size of schools, with secondary schools generally larger than primary schools, impacting the number of households potentially affected by an outbreak. The study demonstrated how the network structure significantly influenced transmission risk, irrespective of minor variations in the within-school reproduction number (R).

The findings of this study directly address the research question by quantifying the risk of SARS-CoV-2 transmission between schools in England under different reopening scenarios. The results strongly suggest that reopening strategies prioritizing primary schools over secondary schools, or focusing on select year groups within secondary schools, significantly reduces the risk of large-scale outbreaks. The methodology employed, constructing a network based on household connections between schools, provides a novel and insightful approach to analyzing transmission dynamics. The study's significance lies in its contribution to evidence-based decision-making in public health. By highlighting the differential risks associated with different reopening strategies, the research informs policymakers on the importance of implementing robust infection control measures and targeted interventions to minimize the potential disruption caused by school-based outbreaks. The relative risks observed align with some existing literature suggesting that older children might pose a greater risk of onward transmission than younger children. The study’s findings, although specific to the English school system, have implications for other settings with comparable educational structures. The need for careful monitoring and reactive school closure strategies is also highlighted.

This study demonstrates the crucial role of network structure in SARS-CoV-2 transmission between schools. The findings emphasize the significantly higher risk associated with reopening secondary schools compared to primary schools. A key contribution is the development and application of a novel network-based methodology for assessing school reopening strategies. The results strongly support implementing effective infection control measures within schools and informed, targeted interventions to minimize potential outbreaks. Further research could focus on refining the model to incorporate more detailed within-school contact structures and explore the impact of different reactive closure strategies. The principles highlighted are not specific to SARS-CoV-2 and may be valuable for managing other infectious disease outbreaks affecting school-aged children.

The study's limitations include the use of data only from state-funded schools in England, potentially underestimating the overall transmission risk if private schools were included. The model assumes a simplified within-household transmission probability and homogeneous mixing within schools, potentially overestimating outbreak sizes compared to real-world scenarios. The impact of immunity was not fully accounted for, given the variability and limited data available. Finally, the model assumed that the risk of an infectious pupil seeding a school outbreak is proportional to the prevalence of infection in the community, a complex relationship that may require further investigation.