Leveraging School Infection Data to Address Community COVID-19 Data Gaps

Abstract

Background

At-home COVID-19 tests became available in the USA in April 2021 with widespread use by January 2022; however, the lack of infrastructure to report test results to public health agencies created a gap in public health data. Kindergarten through grade 12 (K-12) schools often tracked COVID-19 cases among students and staff; leveraging school data may have helped bridge data gaps.

Methods

We examined infection rates reported by school districts to ABC Science Collaborative with corresponding community rates from March 15, 2021 to June 3, 2022. We computed weekly ratios of community-to-district-reported rates (reporting ratios) across 3 study periods (spring 2021, fall 2021, and spring 2022) and estimated the difference and 95% confidence intervals (CIs) in the average reporting ratio between study periods.

Results

In spring 2021, before approval or widespread use of at-home testing, the community-reported infection rate was higher than the school-reported infection rate (reporting ratio: 1.40). In fall 2021 and spring 2022, as at-home testing rapidly increased, school-reported rates were higher than community-reported rates (reporting ratios: 0.82 and 0.66). Average reporting ratios decreased between spring 2021 and fall 2021 (−0.58, 95% CI −0.84, −0.32) and spring 2021 and spring 2022 (−0.73, 95% CI −0.96, −0.48); there was no significant change between fall 2021 and spring 2022 (−0.15, 95% CI −0.36, 0.06).

Conclusions

At-home COVID-19 testing resulted in significant data gaps; K-12 data could have supplemented community data. In future public health emergencies, reporting of school data could minimize data gaps, but requires additional resources including funding to track infections and standardized data reporting methods.

At-home COVID-19 testing provided critical access to rapid diagnostics; however, a lack of infrastructure for reporting these results resulted in large data gaps. We examined whether reporting in kindergarten through grade 12 schools could help fill community data gaps.

INTRODUCTION

On March 13, 2020, the USA declared a national emergency concerning the coronavirus disease 2019 (COVID-19) pandemic. Following this, substantial efforts by local, state, and national health agencies were deployed using epidemiologic methods to monitor the incidence of COVID-19 cases. Decisions around mitigation measures and mandates relied on local counts of cases, hospitalizations, and deaths; these metrics largely informed public health guidance to prevent the spread of COVID-19 in the community. As the pandemic progressed and waves of infections occurred, public health mitigation strategies changed, and the number of cases was tracked to inform the community. When kindergarten through grade 12 (K-12) schools reopened, districts often asked families and staff to report positive COVID-19 tests to track students’ transmission rate within the school populations for purposes of isolation and quarantine. School districts often relied on these data to determine when to ease or enforce COVID-19 mitigation practices such as physical distancing and masking.

In April 2021, a new phase of the pandemic began when at-home COVID-19 testing became publicly available in the USA [1]. As supplies became available, rates of at-home testing for COVID-19-like illness tripled from fall 2021 to spring 2022 [2]. Use of at-home tests rapidly increased in January 2022, at the peak of the Omicron variant, when the federal government launched mailing of at-home COVID-19 antigen tests [3]. One study reported that by early 2022 at least one-quarter of those who reported any COVID-19 testing conducted testing at home [4]. Despite the federal government’s investment of more than $12 billion to expand COVID-19 testing [5], at-home testing was deployed without a clear method for reporting test results, resulting in large gaps in community COVID-19 data [6]. However, many schools and school districts continued to track COVID-19 infections among their students and staff, including results of both at-home tests and tests obtained in a health care setting [7]. These school data represent a potential source of additional COVID-19 case data that was not uniformly captured by local public health agencies.

In this study, we aimed to evaluate whether school district-reported COVID-19 rates increased relative to community rates and might represent a means to capture cases that were missed by public health agencies due to a lack of means for reporting at-home test results. To do so, we leveraged data reported to the ABC Science Collaborative [8], an organization that created partnerships between scientists and school communities to share knowledge and use COVID-19 data to make informed decisions that impacted school operations. Here, we compared weekly school district-reported data to public health agency-reported community transmission rates [4], and examined trends in these rates before and after the widespread use of at-home testing.

METHODS

Study Design and Population

We conducted a retrospective observational case study. The study period was from March 15, 2021 to June 3, 2022, divided into 3 time periods: spring 2021 (March 15, 2021 to May 28, 2021), fall 2021 (August 16, 2021 to December 17, 2021), and spring 2022 (January 10, 2022 to June 3, 2022). During spring 2021, 101 North Carolina school districts and 20 Wisconsin school districts reported at least 1 week of data to the ABC Science Collaborative. Formal statewide data reporting to ABC Science Collaborative began on March 15, 2021 following (1) passage of legislation in North Carolina that required reporting of cases [9], and (2) partnerships with ABC collaborators in Wisconsin who collected data from districts in a uniform manner. At the start of the fall 2021 period, as ABC Science Collaborative grew to have national partners, more than 13 800 school districts nationwide were invited to participate in a study to report and share school transmission cases with ABC; 72 districts reported data at least once during either fall 2021 or spring 2022. Schools that opted into the study could voluntarily report weekly COVID-19 cases in an open cohort format.

For this analysis, we included any district that expressed interest in the study, completed an initial demographics survey that included the number of students and staff, and reported total infections on a weekly basis for at least 1 week. School districts that reported fewer than 4 weeks during a given study period were excluded from that period. However, the exclusion of a district from 1 study period due to lack of reporting did not preclude the district from being included in another study period. Charter schools were also excluded; occasionally a charter school was the sole school reporting data within a county, these rates were felt to be less representative of the surrounding community than a traditional K-12 district.

Definitions and Outcomes

An infected person (case) was any person who was reported by the school to ABC Science Collaborative as having a positive COVID-19 test result. Cases were collected by the schools primarily by passive reporting of cases by parents and staff. Some schools had in-school testing programs that were active during portions of the study period and cases found through at-school testing were also captured. The method of testing (polymerase chain reaction vs home antigen testing) was not captured. Total weekly school cases were defined as the total count of new infections. The school district population was defined as the total number of students and staff enrolled in-person provided by the district. The district-reported weekly infection rate was the total cases reported in a week divided by the district population, standardized to a rate per 100 000 population. To capture community infection rates for any week that a school district reported, we assigned each school district to their surrounding county and used the Centers for Disease Control and Prevention (CDC) USA COVID-19 Community Levels by County dashboard to extract the community weekly infection rate, calculated as a rate of new COVID-19 cases per 100 000 population [10]. In instances where case rates were suppressed by the CDC because of low case rates <10/100 000, we imputed a rate of 9.99. The reporting ratio was the ratio of the community-reported rate to school district-reported rate for a given week. A reporting ratio >1 was indicative of higher community rates than district-reported rates; a reporting ratio <1 indicated that district-reported rates were higher than community rates. The primary outcome was the average reporting ratio for each of the 3 study periods. The secondary outcome was the average weekly reporting ratio across all districts reporting for that week.

Data Source

Schools reported weekly counts of COVID-19 cases among students and staff to the ABC Science Collaborative using a secure database. Data were aggregated to the district level. Demographics captured included district population size (students enrolled and staff employed), county, weeks of reporting, and total weekly student and staff infections, including both school-acquired and community-acquired infections.

Statistical Analysis

We conducted descriptive statistical analysis on reporting district demographics for each study period. For each study period, we calculated weekly district-reported infection rates for all included districts for every week that the district reported. We then extracted the community weekly infection rate for the corresponding counties of those districts included above for each week that the district reported. Next, we calculated by district the reporting ratio for each week. We then calculated an average reporting ratio across reporting districts for each week of the study period. Finally, we calculated a summary reporting ratio by averaging the weekly reporting ratios for each of the 3 study periods. We compared the 3 summary reporting ratios by computing differences and 95% confidence intervals (CIs) for the differences. Analyses were conducted using Stata statistical software (Release 16.1; StataCorp, LLC, College Station, TX). Data collection and analysis were performed under the ABC Science Collaborative of North Carolina Plan A protocol (Pro00108073) and ABC Science Collaborative Program protocol (Pro00108129), both deemed exempt by the Duke University Institutional Review Board.

RESULTS

Across the 3 study periods and 53 weeks of reporting, 141 districts from 114 unique counties and 7 states met inclusion criteria with a total of 1 686 991 students enrolled and staff employed. Districts varied in population size with a range of 242 to 164 382 students and staff. Most districts reported for over half of a study period for which they were included, with a median of 10 weeks reported per district (Table 1).

Table 1.

Characteristics of Reporting School Districts’ COVID-19 Data Across 3 Study Periods (March 2021–June 2022)

	Spring 2021a	Fall 2021b	Spring 2022c	All Study Periods
Total districts includedd	98	66	48	141
Total population	1 432 949	899 436	701 287	1 686 991
Length of study period, weeks	11	19	23	53
Median weeks reported per district (Q1, Q3)	8 (7, 10)	17 (13, 19)	14.5 (10, 21)	10 (7, 27)
Median district population size (Q1, Q3)	6568 (3150, 14 537)	4139 (1401, 8636)	4878 (1730, 12 600)	5512 (2232, 12 364)
Unique counties represented	69	51	44	114
Unique states represented	1	7	7	7

^aMarch 15, 2021 to May 28, 2021.

^bAugust 16, 2021 to December 17, 2021.

^cJanuary 10, 2022 to June 3, 2022.

^dFor a district’s data to be included for a study period, they had to have reported data for at least 4 weeks (being excluded from 1 study period did not exclude a district from the others).

The weekly average district-reported and community rates across each study period are graphically displayed in Figure 1. In spring 2021 (Figure 1A), the average weekly community rate was overall greater than the average weekly district-reported rate, with an average reporting ratio of 1.40 (standard deviation 0.43). During the fall 2021 and spring 2022 periods, however, the average weekly district-reported rate was greater than the average weekly community rate (Figures 1A and 1B). The average weekly reporting ratios (Figure 2) decreased significantly between spring 2021 and fall 2021 (−0.58, 95% CI −0.84 to −0.32) and between spring 2021 and spring 2022 (−0.73, 95% CI −0.96 to −0.48). There was no significant change in the average reporting ratio between fall 2021 and spring 2022 (−0.15, 95% CI −0.36 to 0.06).

Figure 1.

Weekly average school district reported and corresponding community COVID-19 infection rates. Blue dots represent weekly average school district-reported rates per 100 000 district population (students and staff) across reporting districts for that week. Red dots show the average community rates per 100 000 population for counties that reporting districts belonged to, obtained from Centers for Disease Control and Prevention (CDC) Community Levels by County dashboard. COVID-19, coronavirus disease 2019.

Figure 2.

Change in average COVID-19 reporting ratios over time (March 15, 2021 to June 3, 2022). Dots represents the average reporting ratio for each week of the study period. Triangles represents the average reporting ratio across all weeks. The reporting ratio is equal to the community-reported rate of COVID-19 infections in the counties that participating districts belonged to (per 100 000 county population) divided by the school district-reported rate of COVID-19 infections (per 100 000 student and staff population). COVID-19, coronavirus disease 2019.

DISCUSSION

During public health emergencies and pandemics, ensuring accurate case counts is crucial for assessing resource allocations and predicting trends that could place strains on limited health care resources. This study compared rates of COVID-19 infections reported by more than 1.5 million students and staff across 141 K-12 school districts with the corresponding COVID-19 infection rates from their surrounding communities; we found that district-reported rates increased relative to community rates reported by public health agencies over the study period. In spring 2021, the reported community infection rate was higher than the school infection rate. During this time, the majority of testing was conducted via interaction with a health care facility. However, during fall 2021 and spring 2022, the infection rate captured by school reporting was overall higher than the surrounding community rate. This finding may be in part explained by the increased use of home testing and the resultant gap in the community data captured by public health agencies. In future pandemics, K-12 school districts may be able to be leveraged to provide more comprehensive reporting of data and to fill known data gaps.

Home testing provided an opportunity to reduce strain on health care systems and potential to interrupt chains of transmission sooner than relying on accessing health care institutions for testing. However, the approval of numerous at-home tests without a standard infrastructure for reporting the results of these tests resulted in large data gaps in COVID-19 rate estimates [11]. The Institute for Health Metrics and Evaluation estimated that true infection rates were nearly 14.5 times higher than what was being reported by the community [12]. While reporting results of home testing was possible, it was often burdensome and self-selective of specific populations—such as those with higher educational status, higher household income, and those who were more likely to be vaccinated [4]. A study from Say Yes! COVID Test suggested that even when digital methods were made available for individuals to order and report their home tests, only 8% of the study population who ordered tests reported their results [13]. Of those who reported their results, 75% also stated that they reported to their local public health agency [13]. Studies have shown that motivation to report results of at-home tests vary by socioeconomic status, with increased reporting among individuals with higher socioeconomic status [2]. When incentives such as gift cards were implemented for reporting, rates of reporting were higher [13].

Our study suggests that as the pandemic progressed and at-home testing became routine, schools captured relatively higher COVID-19 rates than public health agencies. Our findings are consistent with at least 1 prior study comparing laboratory-reported COVID-19 for children and adolescents in New York to K-12 school-reported cases [14]. This study found that the ratio of school-to-laboratory-reported cases increased with the expansion of at-home testing [14]. Once at-home testing was widely accessible, the burden of reporting cases to public health agencies fell on the individual. In future pandemics, consideration of methods that will provide access to testing while simultaneously leveraging community resources to report cases can minimize data gaps. K-12 schools in the USA often represent unique partners in public health; parents and families are accustomed to reporting illnesses to schools and this relationship could be leveraged to collect important community data. However, schools are often strained for resources. Comprehensive public health efforts would recognize the important role that schools can have in reporting and would allocate appropriate resources to schools to implement rigorous reporting methods. Ensuring that districts have equitable ability to report infections is critical for the generalizability and utility of school infection reporting to the surrounding community.

Our study has limitations. The data used represent real-world observations rather than a controlled study environment. The implementation of at-home testing was not a study intervention; therefore, the changes in reporting ratio cannot be causally linked to the adoption of at-home testing. Second, we used community rates from the surrounding county without specifying age range (eg, community rates in those age <18 years) because district-reported data included both students and staff; however, K-12 populations are not expected to be directly generalizable to the demographics of the community at large. Another possible contributing factor to the changes in school infections relative to community infections could be a change in the proportion of school-aged children acquiring infections during the study period. Although home tests are readily available to the public, we did not compare school or community factors that changed throughout the study to alter the number of used home tests. In addition, this was a voluntary response study. There may have been convenience bias in the data as districts with populations from higher socioeconomic status may have had more resources to allow consistent reporting of data in comparison with districts with less well-resourced populations without access to testing or mitigation strategies. Finally, the study population was dynamic in reporting to the ABC Science Collaborative; however, despite a changing population, inclusion was limited to those districts that followed standard collection and reporting methods and reported at least 4 weeks in a study period to minimize variability in quality of reporting.

The advent of at-home COVID-19 testing resulted in reduced interface with public health agencies for diagnoses and resultant gaps in case reporting. During a public health crisis such as the COVID-19 pandemic, reporting of cases by K-12 school districts may represent an important source of data to improve understanding of true community infection rates; however, rigorous reporting of cases requires numerous resources that may not be available to all school districts and could result in strain on the schools. Such resources that would be needed for future pandemics include funding for personnel to collect data, standardized electronic data collection and reporting methods, access to tests, and staffing and support to conduct contact tracing. Additionally, partnership with families is needed to relay the importance of reporting on a consistent basis and to build a school culture of reporting. Additional work is needed to build sustainable, scalable methods to access and use the important public health data that can be gleaned from K-12 schools.

Notes

Financial support . This work was supported by the Biogen Foundation and Duke Clinical Research Institute’s R25 Summer Training in Academic Research (STAR) Program (grant #5R25HD076475-10). This work was also funded in part by (U24TR001608) of the National Center for Advancing Translational Sciences Trial Innovation Network. This work was also funded under the National Institute of Child Health and Human Development (NICHD) contract (HHSN275201000003I) for the Pediatric Trials Network (PI Danny Benjamin). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Supplement sponsorship . This article appears as part of the supplement “STAR Program,” funded by the Biogen Foundation and Duke Clinical Research Institute’s R25 Summer Training in Academic Research (STAR) Program (grant #5R25HD076475-10). This work was also funded in part by (U24TR001608) of the NCATS Trial Innovation Network. This work was also funded under the National Institute of Child Health and Human Development (NICHD) contract (HHSN275201000003I) for the Pediatric Trials Network (PI Danny Benjamin). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Potential conflicts of interest . Dr Brookhart serves on scientific advisory committees for Amgen, Astellas/Seagen, Atara Biotherapeutics, Brigham and Women’s Hospital, Gilead/Kite, Intercept, NIDDK, and Vertex; he owns equity in Accompany Health and Target RWE. Dr Benjamin, Jr reports consultancy for Allergan, Melinta Therapeutics, and Sun Pharma Advanced Research Co. Dr Zimmerman reports funding from the National Institutes of Health (NIH) and US Food and Drug Administration (FDA). Dr Boutzoukas received salary support through the NIH US government National Institute of Child Health and Human Development T32 training grant (1T32HD094671) during the period of study, and the Biogen Foundation. All other authors report no conflicts of interest.