Impacts of K-12 school reopening on the COVID-19 epidemic in Indiana, USA

1 Introduction

The United States has been the country most severely impacted by the COVID-19 pandemic in terms of total reported cases and deaths, with over 5.2 million cases and 167,000 deaths through August 14 [1]. This severity led to social interventions on an unprecedented scale, including restrictions on mass gatherings, bans on non-essential travel, and school closures [2, 3, 4, 5]. While such restrictions were initially successful in reducing transmission, the subsequent relaxation of restrictions on mass gatherings and movement were followed by large increases in notified cases and, more recently, deaths [1, 3, 6, 7]. The state of Indiana reported its highest daily number of cases to date (1,249) on August 6 [8].

It is within this context of intense community transmission that attention has turned to the reopening of schools for the start of the academic year in August [9, 10, 11]. During influenza epidemics, school closures have been estimated to reduce transmission community-wide [12, 13, 14]. In general, schools are seen as key drivers of the transmission of respiratory pathogens due to close contact among children at school [15, 16, 17]. However, several factors complicate the effect of school reopenings on SARS-CoV-2 transmission. In particular, children and adolescents appear less susceptible to infection and are much less likely to experience severe outcomes following infection [18, 19, 20, 21, 22, 23]. It is also still unclear what their contribution to transmission is, but several studies suggest they can play an important role [18, 24, 25, 26]. There are additional economic and social factors to consider, too, such as the economic costs of school closures for families that must then stay home from work, and the nutritional benefits of school reopening for children who rely on free and subsidized school meals [27, 28, 29].

Our objective in this study was to explore the impacts of strategies for school reopening on the burden of COVID-19 in the state of Indiana, USA. We synthesized evidence around features of COVID-19 epidemiology in children [19] in an agent-based model originally developed for pandemic influenza [30] that is equipped to translate such evidence into projections of community-wide transmission under alternative scenarios about school reopening. We focused on the role of reducing the capacity of classrooms and adherence to wearing face masks in schools, given that both physical distancing and face masks have been shown to reduce transmission of SARS-CoV-2 in community settings [31]. Our main outcome measures were changes in statewide totals of SARS-CoV-2 infections, symptomatic infections, and deaths, across different scenarios of school capacity and face-mask adherence, compared with a baseline scenario in which schools operate remotely.

2 Methods

3 Results

Our model was generally consistent with the data to which it was calibrated, capturing trends over time in daily deaths, hospitalizations, and test positivity (Fig. 1A-C), as well as greater proportions of deaths among older age groups (Fig. 1D). Some trade-offs in the model’s ability to recreate different data types were apparent, such as a recent increase in hospitalizations that the model failed to capture (Fig. 1C), likely due to the predominance of data on deaths in the likelihood. Even so, the model’s predictions reproduced the range of variability in the data, as assessed by coverage probability of its 95% posterior predictive intervals (daily deaths: 0.85; daily hospitalizations: 0.93; daily test positivity: 0.95; cumulative deaths by age: 1.0). The model was also consistent with data withheld from fitting. Across all ages, the model’s 95% posterior predictive intervals of the cumulative proportion infected through late April (median: 0.017; 95% CrI: 0.0045-0.051) and early June (median: 0.022; 95% CrI: 0.0058-0.069) spanned estimates from two state-wide serological surveys [53] (Fig. 2A). Our model’s predictions also overlapped with age-stratified estimates from those surveys (Fig. 2B), although it underpredicted infections among individuals aged 40-60 years.

• Download figure
• Open in new tab

Figure 1.

Model calibration to data: A) daily incidence of death; B) proportion of deaths through July 13 in decadal age bins; C) daily incidence of hospitalization; and D) daily proportion of tests administered that are positive for SARS-CoV-2. In all panels, blue diamonds represent data. In A, C, and D, the gray line is the median, the dark shaded region the 50% posterior predictive interval, and the light shaded region the 95% posterior predictive interval.

• Download figure
• Open in new tab

Figure 2.

Model comparison with data withheld from fitting. We validated the model’s predictions against data withheld from fitting on A) the cumulative proportion of the population of Indiana infected through late April and early June, and B) the cumulative proportion infected among individuals aged 12-40, 40-60, and 60+. Data are shown in navy and come from a random, statewide serological survey [53]. Model predictions are shown in gray. In A, the line and band indicate the median and 95% posterior predictive interval. In B, lines, boxes, and error bars indicate median, interquartile range, and 95% posterior predictive interval.

Calibration of the parameter that scaled the magnitude of SARS-CoV-2 importations [56, 57] in our model resulted in a median of 1.299 (95% PPI: 0.502-1.461) imported infections per day from February 1 to August 10. To ensure that the model reliably reproduced the high occurrence of deaths observed in long-term care facilities, we seeded infections into those facilities at a daily rate proportional to the prevalence of infection on that day; this calibrated proportion was 0.037 (95% PPI: 0.022-0.092). On the opposite end of the age spectrum, our calibration resulted in a median estimate of susceptibility among children of 0.346 (95% CrI: 0.311-0.506), compared to 0.834 (95% CrI: 0.652-0.946) in adults (Fig. S1). Our calibration resulted in an estimate of transmissibility (median: 0.593; 95% CrI: 0.501-0.788) that corresponded to values of R(t) during the initial phase of the epidemic in Indiana of 1.46 (95% CrI: 0.31-3.48), which represents an average of daily values across the month of February (Fig. 1A). Driven by a calibrated estimate 32.1% (95% CrI: 28.8-66.9%) on April 7 (Fig. S2A), our estimates of R(t) dropped to a low of 0.57 (95% CrI: 0.42-0.71) on April 7 and have remained below 1 since (Fig. 1A). Also impacting our estimates of R(t) is the increasing use of face masks in the community, which we estimated at 0.534 (95% CrI: 0.461-0.540) as of July 19 (Fig. S2B). Note that this estimated distribution of community face-mask adherence does not differ between scenarios and is not the same as the level of adherence in schools, which we imposed at different levels depending on the scenario.

In the event that schools reopen at full capacity and without any use of face masks, our model projects that R(t) would increase to 1.72 (95% CrI: 1.43-2.17) by mid-September (Fig. 3A). This assumes that levels of sheltering in place and face-mask adherence in the community remain at their estimated levels as of August 13 (Fig. S2B). Consistent with our model’s prediction that schools contributed appreciably to transmission early in the epidemic (median: 26.2%; 95% CrI: 21.0-27.5%), this increase in transmission is associated with an increase in the proportion of infections arising in schools upon their reopening (Fig. 3B). The sensitivity of the proportion of infections arising in schools to model parameters was highest for the inflection point of the age-susceptibility relationship, the transmissibility parameter, and the rate at which infections were imported into long-term care facilities (Fig. S6). This increase in infections arising in schools is projected to give rise to additional transmission statewide (Fig. 3C). An example of transmission chains arising from schools and spilling out into school-affiliated families and the broader community is shown in Fig. 4.

• Download figure
• Open in new tab

Figure 3.

The impact of school reopening in Indiana on: A) the reproduction number, R(t), over time; B) the proportion of infections acquired in different location types over time; and C) the daily incidence of infection over time. In all panels, schools reopened on August 24. In A and C, the line represents the median, and the shaded region represents the 50% posterior predictive interval.

• Download figure
• Open in new tab

Figure 4.

Example transmission tree of SARS-CoV-2 from one simulation of the model for Saint Joseph County, Indiana. Green shapes at the top represent agents in the model infected at school, and arrows descending from them and other shapes show transmission events in school-affiliated households (orange) and the community at large (purple) that trace back to infections originally acquired in schools. Different shapes distinguish children (circles) from adults (squares).

In the event that schools remain open throughout the semester and no other policies or behavioral responses occur, the increase in R(t) driven by reopening schools at full capacity and without the use of face masks would be projected to result in 2.57 million (95% CrI: 2.36-2.88 million) infections (Fig. 5A) and 10,246 (95% CrI: 7,862-13,794) deaths (Fig. 5B) from Indiana’s population as a whole between August 24 and December 31. At the other extreme, if schools were to go to remote instruction and all children remained at home, our model projects that R(t) would remain near current levels through the remainder of 2020 (Fig. S3A). Again, this assumes that levels of sheltering in place and face-mask adherence in the community as a whole remain at their estimated levels as of August 13. Under this scenario, transmission would continue through contacts at workplaces, within homes, and elsewhere in the community (Fig. S3B), with 45,579 (95% CrI: 14,651-132,546) infections (Fig. S3C) and 790 (95% CrI: 176-1680) deaths between August 24 and December 31.

• Download figure
• Open in new tab

Figure 5.

The impact of different school reopening strategies on A) cumulative infections and B) cumulative deaths in Indiana between August 24 and December 31. Scenarios are defined by the capacity at which schools operate (x-axis) and face-mask adherence (shading). Orange lines represent projections under a scenario of school reopening at full capacity without masks (solid: median; dotted: 95% posterior predictive interval). Blue lines represent a scenario where schools operate remotely. Error bars indicate inter-quartile ranges.

Differing policies on the capacity at which schools operate in person and enforce the use of face masks have a strong influence on the projected statewide burden of COVID-19. If schools operate at 50% capacity and achieve high face-mask adherence, the number of infections and deaths that we project is similar to what we project under the scenario in which schools operate remotely (Fig. 5, Table S1). Of these two policies, projections of infections and deaths were more sensitive to the capacity at which schools operate, with the worst outcomes projected to occur when schools operate at full capacity and with low face-mask adherence. Under this scenario, cumulative infections statewide were projected to be 42.8 (95% CrI: 41.3-44.3) times greater than if schools operated remotely, and cumulative deaths statewide were projected to be 9.2 (95% CrI: 8.9-9.5) times greater (Table S1). The sensitivity of these results to model parameters was relatively high for parameters that define the age-susceptibility relationship and, in some cases, for the transmissibility and shelter-in-place compliance parameters (Fig. S5).

We also explored the sensitivity of these results to alternative assumptions about the contribution of children to SARS-CoV-2 transmission and protection afforded by face masks. Under an alternative assumption that susceptibility of children aged 0-10 years was only 0.1, we observed reductions in cumulative infections and deaths that were consistent but relatively modest compared to differences due to operating capacity and face-mask adherence (Table S10, S11). Under an alternative assumption that the relative infectiousness of asymptomatic infections was 50% that of symptomatic infections, we observed reductions in cumulative infections and deaths under high operating capacity but increases under low and medium capacity (Table S10, S11). Again though, these differences were relatively modest compared to differences due to operating capacity and face-mask adherence. Alternative assumptions about protection afforded by face masks resulted in consistent differences in cumulative infections and deaths, with lower face-mask efficacy associated with more infections and deaths (Table S12, S13). For example, under a scenario with 50% operating capacity and 100% face-mask adherence in schools, the increase in cumulative infections relative to operating remotely ranged from 7% (95% CrI: 3-11%) to 307% (95% CrI: 294-322%). Even so, the relative impacts of varying levels of operating capacity and face-mask adherence were generally similar under these alternative assumptions.

The burden of COVID-19 associated with reopening schools differed for students, teachers, and their families. Relative to a scenario with remote instruction, risk of infection and symptomatic infection was greatest for students (Fig. 6, left column), with a hundred-fold or greater increase in the risk of infection if schools operate at full capacity with moderate or low face-mask adherence (Table S2). Due to their older ages, teachers and families experienced a much higher risk of death under scenarios with high capacity and moderate or low face-mask adherence, as compared with a scenario with remote instruction (Fig. 6, center & right columns). Under a scenario with high capacity and low face-mask adherence, there was a 166-fold higher risk of death for teachers (Table S3) and a 223-fold higher risk of death for family members of students and teachers (Table S4). At the same time, the risk of death under a scenario with high capacity and low face-mask adherence was around 72% the risk of death if schools were to operate at full capacity with no masks (Tables S7 & S8).

• Download figure
• Open in new tab

Figure 6.

The impact of different school reopening strategies on cumulative infections (top row), cumulative symptomatic infections (middle row), and cumulative deaths (bottom row) in Indiana between August 24 and December 31. These outcomes are presented separately for students (left column), teachers (middle column), and school-affiliated families (right column). Scenarios are defined by the capacity at which schools operate (x-axis) and face-mask adherence (shading). Orange lines represent projections under a scenario of school reopening at full capacity without masks (solid: median; dotted: 95% posterior predictive interval). Blue lines represent a scenario where schools operate remotely. Error bars indicate inter-quartile ranges.

4 Discussion

Our model provides a detailed, demographically realistic representation of SARS-CoV-2 transmission in Indiana that is consistent both with data to which it was calibrated and to data that was withheld from calibration. In contrast to models that rely on assumptions about intervention impacts or estimate them statistically [2, 58], our model makes predictions about intervention impacts based on first-principles assumptions about individual-level behavior and contact patterns. Consistent with results from other analyses [2, 58, 20], the inputs and assumptions in our model led to a prediction that schools made a considerable contribution to SARS-CoV-2 transmission in February and early March, prior to large-scale changes in behavior. Extending that, a primary result of our analysis is that schools could, once again, make a considerable contribution to SARS-CoV-2 transmission as they resume normal activity in the fall semester.

Our results indicate that operating at reduced capacity and achieving high face-mask adherence would reduce the burden of COVID-19 in schools and across the state. In the event that both interventions are pursued fully, our model projects that infections and deaths statewide would be around 10% greater than under a scenario with fully remote instruction. In the event that schools operate at full capacity, our model projects that infections and deaths statewide could be one to two orders of magnitude greater than under a scenario with fully remote instruction, especially if there is poor face-mask adherence in schools. The impacts associated with reducing capacity result from reductions in both the number of contacts within the school and the probability that an infected student would be in attendance in the first place, similar to the logic behind why smaller gatherings are associated with reduced risk of transmission [3, 59, 60]. The magnitude of our results was most sensitive to the degree of protection afforded by face masks, which remains uncertain in school and other community settings for SARS-CoV-2 [31].

Although the scenarios we considered resulted in projected impacts spanning nearly the full range between fully remote instruction and fully in-person instruction with no face masks, they are a simplification of the full range of scenarios of how schools could operate this fall. Scenarios that we did not explore include different groups of students attending in person or remotely [61], varying degrees of modularization within schools [62], and the implementation of testing-based control strategies in schools [63]. A related simplification of our statewide analysis is that the state, in reality, consists of a patchwork of policies across districts. In light of this complexity that our model does not capture, our results should be interpreted with caution in setting specific, quantitative targets for capacity or face-mask adherence. For any scenario though, our results illustrate the importance of reducing capacity and maximizing face-mask adherence to the extent possible, as do other modeling studies [61, 62, 63, 64, 65].

The burden of COVID-19 associated with reopening schools is not expected to fall evenly across the state’s population. Under scenarios with schools operating at full capacity, our model projects that hundreds of thousands of children could be infected during the fall semester. Whereas very few deaths are expected among infected children, the numbers of deaths among teachers and school-affiliated families could number in the hundreds. In comparison, the total number of deaths projected across the state is projected to be in the low thousands, meaning that adults with close ties to schools could represent a sizable fraction of deaths across the state in coming months, if schools operate at full capacity and with low face-mask adherence. Importantly, the magnitude of these projections depend further on factors outside the control of schools [2]. In the absence of school reopening, our projections assume steady, controlled levels of transmission with R(t) < 1. Changes in government policies, individual behavior, and face-mask adherence in the community all have the potential to alter this trajectory for the state as a whole, as well as its schools [2, 31, 66].

A critical assumption of our analysis is that children are capable of being infected with SARS-CoV-2 and transmitting it to others at meaningful levels. Although the burden of severe disease skews strongly towards older ages [22, 67, 8], there are other lines of evidence that support our assumption. These include a contact-tracing study that found no distinguishable difference between infectivity of children and adults [26], several studies that found no distinguishable difference in viral load between children and adults [68, 37, 36, 69], a study that observed a greater secondary attack rate among children in homes [26], and a modeling study that found no evidence that children were less infectious [70]. More direct evidence comes from COVID-19 outbreaks that have already been observed in schools, including one in a high school in Israel in which 13.2% of students and 16.6% of staff were infected in just 10 days [71]. Even more pertinent, some schools in Indiana recently reopened and have already reported cases of COVID-19 among students [72]. Our analysis offers perspective on what those initial cases could give rise to in coming months, depending on the degree to which schools choose to operate at reduced capacity and enforce face-mask adherence. Based on two alternative scenarios we explored about the potential for SARS-CoV-2 transmission among children, these lessons about the importance of operating capacity and face-mask adherence appear robust to outstanding biological uncertainties about transmission of SARS-CoV-2 by children.

There is now a growing body of evidence that school closures contributed to mitigating the first wave of the epidemic and could lead to rising case numbers if relaxed [6, 63]. Our study adds to this evidence, and suggests an even greater impact of school reopening than several other studies [65, 63, 62, 73, 61]. This is due in part to our assumption that asymptomatic and symptomatic infections contribute similarly to transmission [26, 68, 37, 36, 69], and in part to our model’s ability to capture chains of transmission within schools and extending out into the community. Our study echoes several modeling studies in emphasizing the importance of reducing school capacity to impede transmission [62, 63, 64, 65, 73]. As schools grapple with COVID-19 throughout fall 2020 and beyond, results such as these provide an important basis for motivating the adoption and sustainment of reductions in operating capacity and adherence to face-mask requirements. As we demonstrated, these actions are highly consequential for those directly linked to schools and for the communities in which they are embedded.