COVID-19 Infections Among Students and Staff in New York City Public Schools

Schools provide large health, social, and economic benefits to children, families, and societies. The American Academy of Pediatrics recommends that jurisdictions strive to ensure that “students [be] physically present in school.”¹  However, schools might also serve as a setting in which respiratory infections transmit readily, amplifying community incidence. The 2019 novel coronavirus disease (COVID-19) pandemic led many jurisdictions to close in-person school instruction for several months or the entire 2020 to 2021 academic year.²  Closures might reduce community transmission; in part because they are often instituted along with other restrictions on businesses and gatherings, any added value has been difficult to measure.^3,4  Evidence has recently emerged that in-person schooling can be conducted in a way that minimizes COVID-19 transmission among students and staff and that the harms of school closure might outweigh a potential benefit in reducing community transmission.^5,6 

In response to rapidly accelerating transmission of COVID-19, New York City (NYC) closed public schools on March 16, 2020, and transitioned all students to remote (online) learning. NYC’s first epidemic wave was among the most lethal in the world, resulting in >23 195 confirmed and probable COVID-19 deaths during February 29 to June 1, 2020.⁷  During subsequent months, NYC reduced, but did not eliminate, COVID-19 transmission and opened its public schools to in-person instruction on September 21, 2020, with the implementation of substantial preventive measures.

To continuously assess the safety of its public schools, the city instituted a program to monitor the prevalence of COVID-19 infection by testing a sample of asymptomatic students and staff physically present in school each day. We analyzed data from this monitoring program and from cases occurring among students and staff diagnosed with COVID-19 through community-based testing during October 9 to December 18, 2020.

The NYC Department of Education (DOE) directly oversees 1607 schools, located in 1400 buildings. These schools enroll ∼1.1 million students annually. For the 2020 to 2021 academic year, all families were given the option of either fully remote or “hybrid” learning, which was defined as 1 to 3 days of in-person school per week combined with remote learning on the other days. No students received full-time, in-person learning. During October 12 to November 20, 288 199 students attended hybrid learning, and 80 876 adults were employed and physically present in schools as teachers, staff, or administrators. Schools were closed from November 19 to December 6, and only elementary and special education schools were reopened in December. More details are available in the Supplemental Information. During December 7 to 18, these numbers were 164 673 students and 44 634 adults (Supplemental Table 6).

For in-person learning, DOE made comprehensive changes to school policy, practices, and facilities. Details are in the Supplemental Information. Multiple city agencies involved in school health established a joint coordination center, hereafter referred to as the Situation Room, to receive, investigate, and act on reports of COVID-19 infection in students, teachers, or staff associated with the public school system.

Reports of COVID-19 cases associated with schools were received in the Situation Room through 3 sources. First, as mandated by law, clinical laboratories are required to report to the NYC Department of Health and Mental Hygiene (DOHMH) information about any city resident who has a severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) viral diagnostic test performed. For new laboratory-confirmed cases, students (or their parent or guardian) are interviewed and asked about any association with a school, and school-associated cases are notified to the Situation Room. Second, all school-affiliated persons were required and actively encouraged to notify their school if they are diagnosed with COVID-19, and schools notified the Situation Room about these cases; such reports are necessary if the person is not a NYC resident. Third, polymerase chain reaction (PCR) testing was performed in each school at least monthly, with results reported directly from clinical laboratories to the Situation Room.

Each public school was assigned a random day when testing would be performed among asymptomatic persons in a school. Testing was performed at least once per month in each school; the frequency changed to once per week for some schools in October to November, then for all schools in December. Further information about the change in testing frequency, consent, and sampling proportions is in the Supplemental Information. In each school, a single swab of the right and left anterior nares was collected and underwent PCR by using standard, US Food and Drug Administration (FDA)–authorized methods at 1 of 2 commercial laboratories (BioReference; Fulgent/Color Genomics).

Each case prompted an investigation to determine if there were additional cases, and contact tracing was initiated by using established protocols in collaboration with school administrators. For public schools, a close contact was defined as any person physically present in a classroom for any duration with the case or any person that spent at least 10 minutes cumulatively within 6 feet during the infectious period (from 2 days before through 10 days after symptom onset date or, if asymptomatic, specimen collection date).⁸  Classrooms and buildings were closed and cleaned following DOE protocols.

A COVID-19 case was defined as a person who tested positive for SARS-CoV-2 on an FDA-authorized PCR or antigen test who had not previously tested positive within the past 90 days. We excluded persons who were not positive on either a PCR or antigen test, including probable deaths and symptomatic persons with known COVID-19 exposure. Testing performed in schools only involved PCR. Testing in the community included PCR and antigen testing.

A COVID-19 event was defined as ≥2 cases occurring within the same school during a 7-day period regardless of whether the cases were known or presumed to be epidemiologically linked. We inferred index cases and possible direction of transmission using dates of diagnosis, symptom onset, interaction, and the absence of other explanations for infection, such as household contact.

The type of school was classified by using terminology of the NYC DOE to describe the range of grades included in a school, because schools can have a wide range of different grade levels: early childhood (preschool to grade 2); elementary (preschool to grade 5), junior high, intermediate, middle (grades 6–8); high school (grades 9–12), kindergarten through grade 8 (K–8); and kindergarten through grade 12 (K–12).

School period prevalence was calculated by dividing the number of persons who tested positive for SARS-CoV-2 by the number of persons tested for a given school week (Monday–Friday) when testing was performed. For comparison, we calculated the period prevalence for the community by using estimates of the number of persons with COVID-19 infection in NYC, then dividing by the city’s population to obtain an estimate of period prevalence.^9,10  The model by the Shaman group permitted analysis by age group for a given week.

We defined incidence as all cases reported during the evaluation period divided by the corresponding population. School incidence was calculated by using the results of any testing done on a student or staff person associated with in-person learning (ie, in-school testing, community testing, and verified self-report), then dividing all cases associated with the schools by the number of persons estimated by DOE to have been physically present in school during the evaluation period. We compared school incidence per week to community incidence using data for COVID-19 cases (PCR positive or antigen positive) from DOHMH. Incidence was stratified by age group and borough and calculated by school type. Because of the change in policy regarding schools and testing detailed in the Supplemental Information, we divided the incidence analysis into 2 periods (October 12–November 19 and December 7–18).

All statistical analysis was conducted in R (version 4.0.3).¹¹ 

Of 234 132 asymptomatic persons tested for SARS-CoV-2 infection in 1594 NYC public schools during October 9 to December 18, 2020, 986 (0.4%) tested positive. Except for the first and last weeks of testing, test positivity increased each week, from 0.19% between October 12 and 16 to 0.67% between December 14 and 18, consistent with rising test positivity citywide during the same period (Table 1). The median age of all case patients was 31 years, with 25% to 75% interquartile range (IQR) of 10 to 47 years; for staff, it was 43 years (IQR: 34–53), and for students it was 9 years (IQR: 7–11).

Among staff who tested positive for COVID-19, test positivity was highest among K–8 schools, followed by elementary and early childhood schools (Table 2). Among students who tested positive, test positivity was highest among students at elementary schools, followed by early childhood schools and K–8 schools.

Period prevalence in the schools never exceeded both model estimates for community prevalence; school prevalence was lower than community prevalence in 4 of the 8 weeks for both models and between the 2 model estimates for the other 4 weeks (Table 3). Stratified by age, the school period prevalence was similar or lower for all groups compared with estimated community period prevalence except for persons aged 65 to 74 years old (Supplemental Table 7).

During October 9 to November 19, 2020, 44 091 persons in the city were diagnosed with COVID-19; for the same period, there were 1259 COVID-19 cases in persons associated with the schools (including 458 diagnosed from in-school testing) (Table 4). During this period, 3% to 6% of persons aged ≥18 years had a PCR performed each week across the community¹² ; in contrast, 11% to 27% of staff in schools had a PCR performed in school each week. Incidence among the school population was 341.1 cases per 100 000 population compared with incidence among the general community of 528.9 cases per 100 000 population. For persons 5 to 17 years old, incidence was 168.6 per 100 000 for the school population versus 383.7 per 100 000 for the community; for persons ≥18 years, it was 955.8 for the school population and 581.5 for the community. Stratified by age groups, incidence was lower in the school community for all persons except in those aged 65 to 74 years. Of all school cases identified during this period, 37.3% were missing age data.

During December 7 to 18, 2020, 42 485 persons in the city were diagnosed with COVID-19; for the same period, there were 972 COVID-19 cases in persons associated with the schools (including 529 diagnosed from in-school testing). During this period in December, 4% to 6% of persons aged ≥18 years had a PCR performed each week across the community¹² ; in contrast, 41% to 51% of staff in schools had a PCR performed in school each week. Incidence among the school population was 464.4 cases per 100 000 population compared with incidence among the general community of 509.6 cases per 100 000 population. For persons 5 to 17 years old, incidence was 244.7 per 100 000 for the school population versus 367.3 for the community; for persons ≥18, incidence was 1274.8 for the school population and 560.1 for the community. Stratified by age groups, incidence was lower in the school community for persons <18 years old and higher for persons aged 18 to 44, 45 to 64, and 65 to 74 years. Of all school cases identified during this period, 38.4% were missing age data.

A total of 36 423 persons were classified as school-based close contacts of a case with an exposure date during October 9 to December 18, 2020. Of those 36 423 close contacts, 191 (0.5%) tested positive for COVID-19 during the 14 days of quarantine (Table 5). For these 191 case patients, 132 (69%) had sufficient information about illness onset date, exposure period, and exposure locations for both cases and contacts to infer the likely direction of transmission. For these 132 case patients, 67 (51%) likely involved transmission from staff to staff, 36 (27%) from staff to student, 18 (14%) student to staff, and 11 (8%) from student to student.

During 8 weeks of in-person school, we found that persons associated with public schools had an overall burden of COVID-19 infections that was no higher than the burden in the general community and that transmission within schools was not common. Although our observation period was short, it involved a large, demographically diverse population that underwent extensive testing and case investigations.

We assessed burden by analyzing both prevalence and incidence of COVID-19. In both analyses, using statistical models or official case notifications for comparison, we found that the overall population of persons participating in in-person learning at public schools had lower or similar measures of COVID-19 infection than the NYC community. We found that staff may have an elevated risk of COVID-19 infection relative to the community but that this risk is not clearly attributable to transmission in schools. Overall incidence was higher for staff than the general community and markedly higher for persons aged 45 to 64 years and 65 to 74 years during December. The fact that prevalence remained similar to or less than model estimates for persons 18 to 64 years old, however, suggests that increased incidence was attributable to increased ascertainment of infection (staff were tested at far higher rates, including 10 times the rate in December, than adults in the community) or acquisition of infection outside of school settings because of rising community incidence. Both prevalence and incidence were higher among persons aged 65 to 74 years, although the small number of cases makes it difficult to draw definitive conclusions.

Our evaluation cannot explain why we observed that in-person school was not associated with an overall increased COVID-19 burden, particularly among students. It is possible that the population that chose in-person learning adhered more rigorously to individual protective measures (such as mask wearing, physical distancing, and hand hygiene) or had other unmeasured characteristics, such as previous infection, that could explain lower levels of COVID-19 infection. It is also possible that the hours spent during school, however limited, help reduce the overall risk of infection by reducing the opportunity for high-risk activities in the community. Longer-term analysis of school COVID-19 monitoring data and further evaluations, such as testing a representative population of students attending remote learning, could help elucidate factors that explain these findings.

Studies have revealed that the strongest predictor of COVID-19 outbreaks in schools is the incidence of COVID-19 infection in the surrounding community.¹³  In our evaluation, we observed that the prevalence of infection in the school community rose with increases in community incidence. Our policies to prevent COVID-19 transmission required us to close a classroom for 14 days in response to an individual case and, in situations with >1 case within a 7-day period, to close the entire building for 14 days if we could not link transmission to an exposure outside the school. Given that COVID-19 has a maximum 14-day incubation period and many persons could not recall a specific exposure that led to infection, the number of building closures grew as community transmission increased. However, the overall proportion of buildings closed for 14 days remained a small proportion of all buildings (9% for October–November; 13% for December).

We estimated the secondary attack rate associated with school exposures was 0.5% and, notably, that a staff person was the likely index case for 78% of these secondary cases. Our findings align with experiences from other jurisdictions that adults are more likely to transmit infection in school settings than children, even in situations, such as in NYC, in which the prevalence of undiagnosed infection was highest in younger children.¹⁴  Schools should strengthen prevention and diagnosis of COVID-19 among staff, including strict adherence to masks and physical distancing in school and out of school and promotion of and priority access for routine periodic testing. Our estimate of the secondary attack rate is prone to error. It could be an overestimate because we do not have definitive proof that all cases arising during quarantine were acquired from the school-based index case that triggered quarantine, rather than from other sources of exposure either inside or outside the school. The secondary attack rate could also be an underestimate because close contacts do not universally undergo testing, although the city’s contact management program includes daily monitoring calls during quarantine that encourage contacts to get tested.

Our evaluation was subject to important limitations. First, investigations depend on interviews with patients (or their guardians) and their contacts, and such interviews rarely provide definitive information about the duration and source of infection. Second, in situations with >1 case within a school, we can only infer and cannot verify whether infections were linked and, if so, the direction of transmission. Third, during the October to November period, only 41% of parents provided consent to have their children tested. Our results could underestimate prevalence (but, importantly, not incidence for that period) if those students were systematically more likely to have undiagnosed COVID-19 infection. Fourth, more than one-third of school-associated cases had missing data about age, making it necessary to interpret analysis of school incidence by age strata with caution. Fifth, our analysis did not include seroprevalence data, which could provide additional information to assess prevalence and incidence in the school population. Sixth, although the 2 models of community prevalence provide plausible lower and upper-bound estimates to compare with school testing data, the models were not developed to estimate the proportion of persons who might test positive by PCR in a predominantly asymptomatic population. Finally, our analysis only covered a brief period of the school year and was not designed to assess the contribution of schools to community transmission. Multiple events that could increase community COVID-19 transmission occurred in NYC contemporaneous with the opening of public schools, including the opening of in-person learning at nonpublic schools and universities; the opening of higher-risk indoor activities, such as dining and fitness gyms (albeit with restricted capacity); the onset of cooler, lower humidity weather; and large increases in incidence in neighboring jurisdictions.

We found that in-person learning in NYC public schools was not associated with increased prevalence and incidence overall compared with the general community. Strict protocols for preventing, diagnosing, and managing school-associated cases might have contributed, but further studies are needed to understand which measures are most important to reducing transmission among students and staff. Longer follow-up and evaluation are also needed to understand how much increased incidence among staff is attributable to increased ascertainment through testing or increased exposures either in school or outside of school. Other jurisdictions seeking to open schools might wish to consider similar policies and practices for reducing transmission, periodic testing to monitor the effectiveness of COVID-19 safety measures, and use of multiagency operations centers, such as ours, to manage the complex process of receiving, investigating, and acting on COVID-19 cases continuously.

Staff in the NYC DOHMH and NYC Situation Room who conducted case and cluster investigations at all public schools.

COVID-19

2019 novel coronavirus disease

DOE

Department of Education

DOHMH

Department of Health and Mental Hygiene

FDA

US Food and Drug Administration

IQR

interquartile range

K–12

kindergarten through grade 12

K–8

kindergarten through grade 8

NYC

New York City

PCR

polymerase chain reaction

SARS-CoV-2

severe acute respiratory syndrome coronavirus 2