Epidemiological investigation
Using a retrospective observational study, we investigated three primary school outbreaks (school 1, 2 and 3) in March–April 2021, following the reopening of primary schools on February 9. The total numbers of children ranged between 265–529 distributed over 9–19 classes, with 19–34 children per class. The schools housed children from grade 1–8 with ages ranging between 4 and 12 years old. Infection prevention at the time of the outbreaks included the use of medical face masks by teachers outside the classroom, availability of disinfectant hand gel in every classroom used prior to entering the classroom, and supervised cohorting of each individual class during lunch breaks with children (school 1 and 2 only). Inside the classroom no masks were used by staff members. Students did not wear masks inside or outside the classroom. No physical distance was maintained between students of the same class. No online schooling was given during the study period. All staff and/or parent meetings were held online. All outbreaks described were investigated by the South Limburg Public Health Service, the Netherlands. Additional information on infection prevention measures at the schools was obtained through visits at affected schools and interviews with concerned managerial staff.
Case definition of cases at schools
A case was defined as an individual with a real-time polymerase chain reaction (RT-PCR) test or antigen test positive for SARS-CoV-2. Cases were defined as symptomatic if they reported COVID-19 related symptoms, including common cold symptoms (nasal cold, runny nose, sneezing or sore throat), cough, elevated temperature or fever (temperature > 38 °C), loss of taste or smell, diarrhea, nausea, fatigue and headache [5]. Cases were defined as asymptomatic if they reported none of these symptoms at the time of their positive test and developed no symptoms in the seven days that followed.
The contagious period in symptomatic cases was defined as the two days preceding symptom development. In asymptomatic and pre-symptomatic cases, the date of the positive test was used as a proxy for disease onset. Cases were further classified into two different categories, i.e., children and staff members. Staff members were defined as teachers and managerial staff of affected schools. An outbreak was considered finished when no new cases were reported over a period of 28 days (two times maximum incubation period of SARS-Cov-2). Attack rate (AR) was calculated as the number children cases divided by the total number of children each class. In each class we assessed if a teacher had taught a class with positively tested children.
Case definition of secondary cases at households
Household members of cases were asked to quarantine at home and were advised to test when developing symptoms and on day 5 following last contact with the case. All household contacts testing positive within 14 days after diagnosis of the primary case counted as secondary cases. Exclusion criteria were single person households, households in which a household contact developed symptoms prior to the primary case and cases for which no household information was known. Sibling pair families were combined. Secondary attack rate (SAR) was calculated as the number of cases divided by the persons at risk.
Testing strategy
According to national outbreak management protocols for primary schools, classes are put into quarantine for ten days following a single case visiting the classroom from two days before symptom onset [6]. All children from affected classes including teachers are recommended to be tested twice, i.e., once as soon as possible after exposure, and once five days after exposure. If several classes are affected, closure of the entire school may be considered. Whole genome sequencing was performed on random samples from at least one child and one teacher of each affected class. In classes with more than 10 cases we tried to sample additional samples. Antigen tests and samples with a cycle threshold value > 32 were excluded as whole genome sequencing (WGS) was not possible. Samples not investigated at the Maastricht University Medical Centre (MUMC +) were requested at the relevant laboratories.
Laboratory testing
Laboratory confirmation of SARS-CoV-2 was performed via an RT-PCR assay or antigen test on a nasopharyngeal sample. Samples investigated in the MUMC + were determined using the following method. First, for RNA extraction, 900 µl of clinical sample were mixed with 900 µl of Chemagic Viral Lysis Buffer (Perkin-Elmer) and RNA was extracted from samples using the Chemagic Viral DNA/RNA 300 Kit H96 (Perkin-Elmer) on the Chemagic 360 system (Perkin-Elmer). A multiplex RT-PCR was performed using the N1-gene and E-gene as targets, including the immediate early gene of mouse cytomegalovirus as an internal control. cDNA synthesis and PCR amplification were combined using the TaqPath™ 1-Step RT-qPCR Master Mix, CG (Applied Biosystems, US). Thermal cycling was performed using the Quantstudio 5 Real-Time PCR System (Applied Biosystems, US). Oligonucleotides were synthesised and provided by Biolegio (Netherlands).
Whole genome sequencing
Samples that were tested positive for SARS-CoV-2 were stored at -80 degrees Celsius until RNA was isolated for sequencing. For RNA extraction, 90 µl of sample were mixed with 90 µl of Chemagic Viral Lysis Buffer (Perkin-Elmer), followed by extraction using the MagNA Pure 96 DNA and Viral NA Small Volume Kit 96 (Roche, Germany) on the MagNA Pure 96 system (Roche, Germany), without the addition of an internal extraction control.
Sequencing was performed using the PCR tiling of SARS-CoV-2 virus with Native Barcoding Expansion 96 (EXP-NBD196) protocol (Version: PTCN_9103_v109_revH_13Jul2020) of Oxford Nanopore technologies, with minor modifications and using the primers previously published by Oude Munnink et al. [7]. Briefly, the only modifications were extending the barcode and adaptor ligation steps up to 60 min and loading 48 samples per flow cell.
Bioinformatic analysis was performed using an in-house developed pipeline MACOVID v2.0 that is based on Artic v1.1.3. In brief, short and obvious chimeric reads are filtered with Cutadapt v2.5. The filtered reads were mapped to the reference genome MN908947.3 with Minimap2 v2.17 and quality checked with “align_trim” function of Artic v1.1.3. Mapped reads were split per primer pool using Samtools v1.9 and a consensus was created per primer pool with Medaka v1.0.3. Variants were called using Medaka v1.0.3 and Longshot v0.4.1. Low coverage regions (< 30x) were masked with “artic_make_depth_mask” function of Artic v1.1.3. A preconsensus was made with “artic_mask” and the final consensus sequence was made with bcftools v1.10.2. Documentation and source code are available from https://github.com/MUMC-MEDMIC/MACOVID under MIT license. The consensus sequences were used to construct a phylogenetic tree with ncov pipeline (https://github.com/nextstrain/ncov) v3 of nextstrain with all B.1.1.7 and Dutch genomes in the Global Initiative on Sharing All Influenza Data (15-apr-2021) as a reference. The WGS used in this paper was considered to be highly accurate as a 100% score was obtained in identifying single nucleotide polymorphisms and indels, Pango lineages and clusters between samples in an external quality assessment of SARS-CoV-2 WGS across 15 European Laboratories[8].
All unsuccessful sequenced samples were analyzed a second time to increase the success rate of the sequencing procedure. For cluster identification, a cut-off value of ≤ 2 single site mutations difference between genomes was applied. Isolates with genomic mutations within this cut-off were deemed be part of the same genotypic cluster, hereafter referred to as the same genotype. All FASTA files had been uploaded to GISAID (see Supplementary file 1 for accession numbers).