Abstract
Schools were largely closed in 2020-2021 to counter COVID-19 spread, impacting students’ education and well-being. With highly contagious variants expanding in Europe while vaccine hesitancy persists, safe options to maintain schools open are urgently needed. We developed an agent-based model of SARS-CoV-2 transmission in school. We used empirical contact data measured in a primary and a secondary school in France, and field estimates for adherence to screening from 683 schools during the spring 2021 wave. Examining different screening protocols, we performed a cost-benefit analysis for varying epidemic conditions and vaccination scenarios. In a partially immunized school population, weekly screening would reduce the number of cases on average by 24% in the primary and 53% in the secondary school compared to symptom-based testing alone, if R=1.3 and 50% adhered to screening. This adherence was met in primary schools (53% (95% confidence interval 21-85%)), but insufficient participation was recorded in secondary schools (10% (1-38%) in middle schools, 6% (2-12%) in high schools). Regular screening would also reduce by 90% the number of student-days lost compared to reactive class closure. No difference was predicted when fully vaccinating teachers, due to their limited number and mixing. Partially vaccinating adolescents would still require regular screening for additional control (20% case reduction with 50% vaccinated students). In the upcoming fall, COVID-19 epidemic will likely continue to pose a risk to the safe opening of schools. Increasing vaccination coverage in adolescents and implementing regular testing while largely incentivizing adherence are essential steps to keep schools open.
INTRODUCTION
School closure has been largely used worldwide against the COVID-19 pandemic. The first wave witnessed many countries go into strict lockdowns closing schools for long periods of time [1]. Their reopening has been continuously challenged by successive waves and the need for social distancing restrictions, especially with the emergence of new variants driving rapid case resurgences [2], [3].
A rather heterogeneous landscape emerges from how countries worldwide relied on school closure for epidemic control, mainly due to limited understanding of viral circulation in children and their contribution to transmission [4]. In Europe, depending on the country, students lost from 10 to almost 40 weeks of school from March 2020 to March 2021 due to partial or total school closure (Figure 1). Outbreaks in schools [5], [6] are difficult to document, as infections in children are mostly asymptomatic [7] or present mild non-specific symptoms [7]. Despite the lower susceptibility to infections in children compared to adults [8], viral circulation can occur in school settings [6], especially in secondary schools [9]. Accumulating evidence is consistent with increased transmission in the community if schools are in session [9]–[11], and model-based findings suggest that school closure may be used as an additional brake against the COVID-19 pandemic if other social distancing options are exhausted or undesired [12], [13].
(a) Average number of in-presence school weeks lost by students in Europe because of school closures due to the pandemic. Source: Unesco [1]. (b), (c) Visualization of the empirical temporal contact data aggregated over two days, for the primary (panel b) and the secondary (panel c) school. Nodes represent teachers and students, each circle represents a class (each of a different color), and links represent contacts, with the thickness coding the contact duration. (d) Daily average number of distinct contacts per individual within the same class or in different classes, in the primary and secondary school. Horizontal dashed lines represent the average class size. (e) Daily average time that an individual spends in interaction within the same class or in different classes, in the primary and secondary school. (f) Daily average time that an individual spends in interaction for teachers and students in the primary school. In the three bottom panels, histogram bars refer to the empirical networks. Points and error bars refer to the synthetic networks (errors represent 95% confidence intervals). In panels d and e for the secondary school, there is an increase of average number of contacts and their duration in the synthetic networks compared to their empirical counterparts, due to the ad hoc addition of contacts between school years. In panel f, only values from the synthetic networks are shown, as teachers did not participate to the data collection, and their contact behavior was inferred from another dataset (see Methods).
Keeping schools safely open remains a primary objective that goes beyond educational reasons, and pertains to the social and mental development of children [14], as well as the reduction of inequalities. Several countries implemented safety protocols at schools, including the use of masks, hand hygiene, staggered arrival and breaks. Regular testing [15]–[17] was more recently introduced in educational environments as an additional control measure. Vaccination has been extended to the 12+ population in Europe, and recently boosted by the introduction of sanitary passes in some countries. Yet it is unlikely that schools will be largely vaccinated at the start of the 2021-2022 calendar year, while also awaiting the vaccine for children younger than 12. With the highly contagious Delta variant [3] rapidly circulating among young age groups, expected scenarios for the upcoming fall may threaten the safety of classrooms. Assessing the conditions of vaccination and regular testing in schools is therefore key to anticipate the safe opening of schools under a variety of possible epidemic contexts. Here, through an agent-based transmission model parameterized on empirical contacts at schools and using field data on adherence to screening in schools, we evaluate different testing protocols under varying immunity profiles of the school population, accounting for age-specific differences in susceptibility to infection, contagiousness, and contact patterns.
METHODS
Empirical patterns of contacts
We used empirical data describing time-resolved face-to-face proximity contacts between individuals in two educational settings, collected using wearable RFID sensors in a pre-pandemic period. The Primary school dataset describes the contacts among 232 students and 10 teachers during two days in October 2009 in a primary school in Lyon, France [18]. The school is composed of 5 grades, each of them comprising two classes. The Secondary school dataset describes the contacts between 325 students of 9 different classes during one week in December 2013 in a secondary school in Marseilles, France [19]. Data collection concerned high school students of the second year of classes specific to the French schooling system, called “classes préparatoires”. These classes gather students for 2-year studies at the end of the standard curriculum to prepare for entry exams at specific Universities. Classes are divided in three groups, each focusing on a specialization (mathematics and physics; physics, chemistry, engineering studies; biology).
Datasets are available as lists of contacts over time between anonymized individuals, with a classification by class and in terms of students/teachers. From the raw data, we built the corresponding temporal contact networks, composed of nodes representing individuals, and links representing empirically measured proximity contacts occurring at a given time (Figure 1). As each dataset covers only a few days, we developed an approach to temporally extend the datasets by generating synthetic networks of contacts that reproduce the main features of the empirical data (class structures, within-vs. between-classes links, density of links, heterogeneity of contact durations, and similarity of contacts between consecutive days). The secondary school synthetic network was further extended to generate a synthetic first year (to consider the full curriculum of the “classes préparatoires”) including teachers whose contacts were inferred from an additional dataset for the same school. The resulting network for the secondary school was composed of 650 students and 18 teachers. Details are provided in the Appendix.
Field data on adherence to screening in schools in France
In response to a rising third wave in France in the spring 2021, authorities started offering the screening at schools on a voluntary basis to limit cases and avoid closures. We used data on adherence to screening collected in 683 schools between March 8 and June 7, 2021, in the Auvergne-Rhône-Alpes region of France. Participating schools included different levels: 94 pre-schools, 427 primary schools, 158 middle schools, and 4 high schools, for a total of 209,564 students and 18,019 personnel. 91 schools (13.3% of the total) participated more than once to the testing, with a maximum of 3 screenings in 5 schools. PCR tests from saliva samples were proposed in pre-schools and primary schools, and self-tests in middle and high schools. This initiative was interrupted in April due to school closures against the third wave, and resumed in early May at reopening. More details on the data are provided in the Appendix.
Ethics statement
Contact studies were approved by the Commission Nationale de l’Informatique et des Libertés (CNIL, the French national body responsible for ethics and privacy) and school authorities. Informed consent was obtained from participants or their parents if minors. No personal information of participants was associated with the RFID identifier. Testing at school was part of surveillance activities approved by school authorities and proposed with parental consent. Data on adherence to screening were aggregated and anonymized at school level.
Transmission model in primary school and secondary school
We developed an agent-based stochastic transmission model for SARS-CoV-2 infection spreading on the network of contacts. Time progression of the infection in the host is specific to COVID-19, including prodromic transmission followed by clinical or subclinical disease stages. We considered individuals in the prodromic and subclinical compartments to be less infectious and to remain undocumented unless tested [20].
The model is parameterized with age-specific estimates of susceptibility, transmissibility, probability of developing symptoms, and probability to detect a case based on symptoms. A systematic review indicates that minors have lower susceptibility to SARS-CoV-2 compared to adults [8], but building evidence suggests that high school aged students may be as susceptible as adults [6], [21]. Here we considered a relative susceptibility of 50% in children and 75% in adolescents compared to adults, and tested homogeneous susceptibility across age for sensitivity. In presence of symptoms, the probability to recognize a suspect COVID-19 infection was set to 30% for students and 50% for teachers, based on studies indicating that about two thirds of symptomatic children [7] and half of symptomatic adults [22] have unrecognized symptoms before diagnosis. These detection rates were varied for sensitivity. We considered a lower transmissibility in children, based on multiple studies suggesting that transmission in children may be less efficient [23], [24]. We also explored increased values of relative transmissibility to explore possible changes induced, for example, by the circulation of a novel variant with a higher infectivity on children. Full details on the model are reported in the Appendix.
Epidemic contexts and vaccination scenarios
We considered an epidemic scenario for the next fall characterized by an effective reproductive number R=1.3, 25% natural immunity in the population, 50% to 100% of vaccinated teachers, and 0 to 70% vaccination coverage in adolescents. R=1.3 corresponds to the value estimated during the second pandemic wave in France. Here, it includes the effectiveness of sanitary protocols (e.g., physical distancing, hand hygiene, use of masks) applied in the school settings to limit viral circulation. We also explored additional values including R=1.1, R=1.2 (as estimated during the third wave in France), R=1.5 (as estimated in early-June 2021 in the United Kingdom due to the circulation of the Delta variant), and R=2 (based on early-July 2021 estimates for the Delta circulation in France).
Introductions in the school settings are estimated from the community prevalence [25]. Here we considered one introduction every two weeks (and explored a range for sensitivity), approximately corresponding to the value estimated in January-February 2021 in France.
Vaccination is modeled with 95% effectiveness against infection, 97% effectiveness against symptomatic disease [26], and 50% effectiveness against transmission [27] estimated one week after a full cycle of two doses of Pfizer vaccine, the most common one used in France (>79% as of July 21, 2021).
Testing and isolation protocols
Symptom-based testing and case isolation (ST) is considered as the basic strategy, present in all protocols, and against which interventions are evaluated. It considers that clinical infections are detected with the estimated probability and tested; if the result is positive, the case is isolated for 7 days. We tested the following intervention protocols:
Reactive quarantine of the class (ST+Qc): once a case is identified through ST, their class is put in quarantine for 7 days. If quarantined individuals develop symptoms, they remain in isolation for an additional period of 7 days, before returning to school. This protocol is close to the one in place in France before regular testing was deployed.
Reactive quarantine of the class level or specialization (ST+Ql): as the previous protocol, but quarantine is applied to the classes of the same level (2 classes in the primary school) or specialization (3 classes in the secondary school) of the detected case. This option is considered as empirical contact data show a larger mixing between students of the same level (i.e. of approximately the same age), than across different levels, in the primary school; and analogously for the specialization in the secondary school.
Regular testing with α adherence (ST+RTα%): in addition to symptom-based testing, regular testing is offered to the school and performed at a certain frequency (once every two weeks, once or twice per week). We considered that only a certain percentage α of the non-vaccinated school population will participate to regular testing. Detected cases are isolated without triggering any class closure.
Regular testing with α adherence, and reactive quarantine of the class (ST+RTα%+Qc): in addition to the protocol above, the reactive closure of the class is triggered at every detected case, whether from recognizable symptoms or regular testing.
Following protocols adopted in France, we considered PCR testing from saliva samples in the primary school (test sensitivity of 70% in the prodromal phase, 80% in the subclinical phase, and 90% in the clinical phase [28], with a result available after 24h) and nasal self-testing in the secondary school (test sensitivity of 50%, 70%, and 80%, respectively, with a result available after 15’). We varied test sensitivity after the prodromal phase between 70% and 100% to assess the impact of sampling accuracy (e.g. samples collected by non-professionals). We considered that teachers are required to show proof of a negative PCR test when returning to school after infection.
Adherence to regular testing was informed from field data on the testing initiatives in France in the spring 2021, and then further explored in a range between 10% and 75%.
Simulation details and analysis
Results for each scenario were obtained from 1,000 (primary school) to 2,000 (secondary school) stochastic simulations over the course of a school trimester (90 days). We computed medians and 95% confidence intervals (CI) for the quantities of interest.
RESULTS
Contact networks measured through wearable sensors display a strong community structure around the classes, common to both the primary and secondary schools (Figure 1). The patterns of interaction, however, vary substantially between the two educational settings. Children have on average a larger number of distinct contacts during a day, interacting with almost their entire class, compared to adolescents establishing links with about one third of their class. Approximately 50% more links occur between classes than within classes in the primary school, contrary to what observed for adolescents (75% fewer links). But accounting for duration, students in both settings spend on average more time interacting within the class than with individuals from other classes. This indicates that the majority of between-classes interactions observed in the primary school are volatile (e.g. short-duration mixing during breaks). Students are also responsible for longer interacting times compared to teachers (about 64% longer).
In a partially immunized school population with 50% vaccinated teachers, regular testing constitutes an efficient protocol for preventing infections (Figure 2). It would substantially outperform protocols based on simply identifying cases given recognizable symptoms and additionally closing the class of the detected case, if adherence is large enough. However, the introduction of testing at schools on a voluntary basis during the third wave in France was met with rather heterogeneous participation rates. Adherence of the school population was higher in lower school levels (39% (95% CI 9-72%) in pre-school, 53% (21-85%) in primary school) compared to secondary schools (10% (1-38%) in middle school, 6% (2-12%) in high schools), and subject to large variations within each school level. We found that with 50% of non-vaccinated individuals participating, i.e. approximately the value recorded in the French primary schools, weekly screening would reduce the number of cases by 24% in the primary school and by 53% in the secondary school compared to symptom-based testing alone. The case reduction would rise to about 30% and 70% in the two schools, respectively, by either increasing adherence to 75%, or by keeping the same participation (50%) but increasing the frequency of tests (twice a week). This shows how infection prevention improves with both adherence and frequency of tests, and higher frequency is needed to compensate for lower adherence. However, if the adherence to testing is too low (10%), as recorded in the French secondary schools, weekly testing would have a very limited impact, estimated to be <10% case reduction. While trends are similar across settings, age-specific contact patterns and epidemiological properties make regular testing more efficient in secondary school. Isolating a case in secondary school indeed prevents a larger number of transmissions compared to what would occur in primary school, because of the larger transmissibility and susceptibility of adolescents. Considering a higher transmissibility of children, possibly due to a newly emerging variant, or a higher susceptibility would lead to similar results in the primary school (Appendix). Findings were robust against changes in detection rates and test sensitivity (Appendix). Larger reproductive numbers would require a higher screening frequency (Appendix).
(a) Number of schools participating to the screening initiative during the third wave in France, according to school level (pre-school, primary school, middle school, and high school). (b) Predicted percentage of reduction in the number of cases achieved by each intervention protocol with respect to the basic strategy of the symptom-based testing (ST) in the primary school. Intervention protocols are: symptom-based testing and case isolation, with reactive quarantine of the class (ST+Qc); symptom-based testing and case isolation, with reactive quarantine of the class level (ST+Ql); symptom-based testing and case isolation, coupled with regular testing with a percentage of adherence (ST+RTα%), with α=10%, 50%, and 75%. For regular testing, different frequencies are shown: one test every two weeks, a weekly test, two tests per week. Error bars correspond to 95% confidence intervals. The empty marker corresponds to the adherence estimated from empirical data recorded in schools of that level (from panel d). (c) As in panel b, for the secondary school. (d) Estimated adherence to screening recorded in the different school levels participating to the screening initiative in the third wave in France. Error bars correspond to 95% confidence intervals. (e) Probability distribution of the simulated final epidemic size in the primary school for selected protocols at the end of the trimester. Regular testing is done with weekly frequency. (f) As in panel e, for the secondary school.
Next to reducing the number of infections, regular testing is able to strongly limit the number of days of absence of students. The quarantine of the class of a detected case implies about 30 to 50 times more student-days lost compared to symptom-based testing alone (Figure 3). These numbers inevitably increase when additional classes (of the same grade or specialization) are closed after detection. Not being sufficiently targeted, the reactive class closure quarantines individuals while their risk of infection remains low. On the other hand, despite detecting more cases, regular testing leads to a small increase in student-days lost,≤6 times the number of days lost with the basic strategy and about 90% less than the class closure intervention, as isolation is only applied to detected cases. The cost-benefit analysis of Figure 3 shows that for all regular testing strategies, the cost in terms of person-days lost remains low, even when the benefit becomes very high, for a range of different epidemic contexts. Strategies based on class quarantines do not manage to reach a very high benefit, even at large cost. Closing the class at each case detected through regular testing would provide a limited additional benefit in reducing the number of cases, except for very small adherence values (from 10% to 50% case reduction if participation is equal to 10% as measured in secondary schools in France, Figure 4): as testing is largely underperforming given the low participation, closing the class ensures the isolation of undetected infections. However, this gain is offset by the large increase of absence from school (about 40 times more days lost per student).
(a) Predicted increase in student-days lost with respect to symptom-based testing, ST for different protocols in the primary (solid bars) and the secondary (lighter color bars) schools. The empty marker corresponds to the adherence estimated from empirical data recorded in schools of that level. (b) Predicted percentage of reduction in the number of cases vs. predicted increase in student-days lost in the primary school. Both quantities are computed relatively to the basic strategy (symptom-based testing, ST). Each point in the plot corresponds to a protocol. Additional values of R in the range 1.1-2.0 are also shown. (c) As panel b, for the secondary school.
(a) Predicted percentage of reduction in the number of cases achieved by weekly testing protocols (ST+RT) and by weekly testing protocols coupled with the reactive quarantine of the class after the detection of a case (ST+RT+Qc). Quantities are computed relatively to the basic strategy (symptom-based testing, ST). (b) As panel a, for the secondary school. (c) Predicted increase in student-days lost obtained with weekly testing protocols (ST+RT) and weekly testing protocols coupled with the reactive quarantine of the class after the detection of a case (ST+RT+Qc). Quantities are computed relatively to the basic strategy (symptom-based testing, ST). (d) As panel c, for the secondary school. In all panels, the empty markers and bars correspond to the adherence estimated from empirical data recorded in schools of that level.
Benefits and costs of regular testing remain stable when vaccination coverage of teachers increases from 50% to 100%, with no observed difference across the two schools (Figure 5 and Appendix). Increasing vaccination coverage in adolescents is a strong protective factor against school outbreaks, so that regular testing becomes less performant in preventing cases as vaccination largely increases. Nonetheless, it would provide an important supplementary control: even with 50% vaccinated students, weekly self-testing would additionally reduce by almost 20% the number of cases if half of the non-vaccinated students would self-test, without impacting on school closure. The vaccination level needed to reduce the performance of regular testing to 15% or below increases with R (from about 55% for R=1.3 to 75% for R=2, Appendix).
(a) Predicted percentage of reduction in the number of cases achieved by selected protocols as a function of the vaccination coverage in teachers in the secondary school. The case reduction is computed relatively to the basic strategy (symptom-based testing, ST). (b) Predicted percentage of reduction in the number of cases achieved by intervention protocols as a function of the median epidemic size in the secondary school. The case reduction is computed relatively to the basic strategy (symptom-based testing, ST). Dots of the same protocol for different vaccination coverage are clustered together. (c) Predicted increase in student-days lost for selected protocols as a function of the vaccination coverage in teachers in the secondary school. The increase in days lost is computed relatively to the basic strategy (symptom-based testing, ST). (d) Predicted percentage of reduction in the number of cases achieved by selected protocols as a function of the vaccination coverage in adolescents in the secondary school. The case reduction is computed relatively to the basic strategy (symptom-based testing, ST). (e) Predicted total epidemic size at the end of the trimester vs. the vaccination coverage in adolescents in the secondary school in selected protocols. (f) Predicted increase in student-days lost for selected protocols as a function of the vaccination coverage in adolescents in the secondary school. The increase in days lost is computed relatively to the basic strategy (symptom-based testing, ST). The 50% and 70% data-points are not shown as the median student-days lost for ST is equal to zero (the likelihood of generating a clinical infection among the pupils - recognizable on the basis of the symptoms - becomes low with increasing vaccination coverage). In all panels, the empty markers and bars correspond to the adherence estimated from empirical data recorded in schools of that level.
DISCUSSION
Safely maintaining schools open during the COVID-19 pandemic is a matter of controversial debate and relatively limited knowledge from the field. As countries in Europe prepare for the start of the next school calendar, school remains a central issue in the midst of vaccine hesitancy, and the rapidly increasing circulation of the Delta variant [3]. Using empirical data on contacts established in a primary and a secondary school in France, we developed an agent-based transmission model to evaluate different testing strategies in the school setting. For realistic epidemic contexts and immunity profiles of the population, we found that weekly self-testing with large enough adherence provides an optimal balance: it largely improves epidemic control in the school population while avoiding disruptions in the school calendar due to class closures. It also remains essential for infection prevention in situations with zero to moderate vaccination coverage of the student population.
Adherence to self-testing is crucial for the strategy to be efficient, suggesting that at least half of non-vaccinated individuals should participate to weekly testing to achieve an important case reduction. This requirement was met by the testing initiatives conducted in primary schools in France in the spring 2021, though many schools only participated once. For testing to work, especially in those settings where the student population is not yet eligible to vaccination, these levels of adherence should be confirmed and possibly improved in the fall 2021, but most importantly maintained over time with a weekly frequency. We also estimated that the very low adherence recorded in secondary schools in France would be less performant than the previously implemented protocol (symptom-based testing with isolation of the case and quarantine of the class). In the current context of increasing viral circulation due to the Delta variant, preparation for the next school calendar should consider improving strategies for the communication, implementation, and engagement of the school community to considerably improve participation. So far in France, screening at school remained voluntary. As required vaccination for school entry is unlikely to be considered, adaptations of the recently introduced sanitary pass (requesting either a complete vaccination or a recent negative test) to the school setting may eventually be envisioned, similarly to other countries where school attendance is constrained to regular screening.
In France, criteria for class closures have evolved over time, with the intervention triggered after 1 or 3 detected cases in the same class, depending on the time period. These strategies are, however, very expensive in terms of student-days lost, despite the probability of detecting a case being rather small in younger individuals [7]. They also have a limited value in epidemic control, as other classes may be already affected due to unobserved introductions from the community or silent spreading within the school. This second effect becomes particularly important when between-classes mixing is higher, as observed in the primary school under study. If efficient regular testing is in place, additionally closing the class after each detected case would provide a small additional advantage to epidemic control if adherence to testing is >50%, but it would largely increase absence from school. While regular testing is able to detect more cases than detection based on symptoms, and thus a priori leads to more time spent in isolation per student, it keeps days lost low for two main reasons. First, isolation is only applied to detected cases during their infectious period, thus representing a targeted intervention compared to class quarantine, which is not specific enough. Second, detecting cases that otherwise go unnoticed helps control the epidemic, preventing further diffusion and reducing the overall number of cases generated during a school outbreak. As a consequence, the time spent in isolation is also reduced. This evidence therefore supports the use of weekly self-testing without class closures.
Increasing vaccination in teachers will largely protect them from infection and symptomatic disease [26], but it yields limited protection for the school population, even under full coverage. This results from the small number of teachers in the school population and the lower rates of contacts they have with students. Despite being specific to the measured contact patterns, these findings were obtained for both school settings and suggest that extending vaccination to eligible students is needed to achieve a collective benefit. Numerical evidence confirms that if coverage is large, outbreak size is already reduced thanks to the vaccination in adolescents. In these conditions, self-testing would bring a supplementary control whose application should be evaluated in light of resources, logistics, adherence, and epidemic conditions. Regular testing remains however critical in low to moderate coverage situations, as it would prevent a substantial portion of undetected infections. This becomes particularly important as incidence rapidly increases in the younger age groups due to the Delta variant, and epidemic scenarios for the upcoming fall highlight the important role of non-vaccinated minors in sustaining SARS-CoV-2 circulation in the community [29]. Moreover, preliminary data indicates an imperfect effectiveness of the vaccine in preventing infection and transmission due to the Delta variant [28], reinforcing the relevance of regular screening. Self-testing at school could detect silent spreading events and break the chains of transmission, with a direct impact to the school environment, reducing the number of infections and long COVID in children [30], and an indirect impact on the community, protecting students’ contacts.
This study has a set of limitations. First, it focuses on two school settings for which empirical contact data were available, but contacts in other schools may be different, depending on the structure of curricula and the organization of activities. While precise estimates reported here remain specific to the schools under study, findings on the efficiency of self-testing and vaccination are robust across a range of epidemic conditions and synthetic contact patterns, and can thus inform on the choice of strategies to safely keep schools open. Second, we considered only contacts occurring within the school premises during school hours, but we did not consider socializing activities outside the school, for example at the exit. These contacts may further increase viral circulation in the student population, making vaccination and regular testing at school even more important. Third, the study focuses on school outbreaks and it does not assess the impact that these strategies will have on the viral circulation in the community.
In the fall, COVID-19 epidemic will likely continue to pose a risk to the safe opening of schools. Regular testing remains a key strategy to epidemic control in primary schools, and an important additional support in secondary schools with low or moderate vaccination coverage, all the while minimizing days lost.
Data Availability
The data that support the findings of this study are openly available at the references cited.
ACKNOWLEDGMENTS
We thank the Assistance Publique - Hôpitaux de Paris and Santé publique France for useful discussions; Philippe Vanhems, Elisabeth Bothello-Nevers, Olivier Epaulard, Jean Beytout, Annabelle Ravni and the Academie of the Auvergne-Rhône-Alpes region for the school screening initiatives. This study was partially funded by: ANR projects COSCREEN (ANR-21-CO16-0005) and DATAREDUX (ANR-19-CE46-0008-03); EU H2020 grants MOOD (H2020-874850) and RECOVER (H2020-101003589); REACTing COVID-19 grant.