Skip to main content

SARS-CoV-2 infection in primary schools in northern France: A retrospective cohort study in an area of high transmission

Our take —

This retrospective cohort study, available as a preprint and thus not yet peer reviewed, examined the infection attack rate from 6 primary schools in Crepy-en-Valois, France among 1340 total participants. They found an attack rate of 8.8% among students, 7.1% among teachers and 11.9% among all parents. This study may have had selection bias due to a low response rate, and recall bias for symptoms and symptom onset. Importantly, in contrast to data from high school students, there was no evidence of onward transmission among pupils to one another or teachers/staff, potentially due to the mild course of infections experienced. These data give hope that at least for primary schools, onward spread of infection by young children may be limited, however caution is warranted based on the limited time period for which potential infections may have been observed prior to school closures.

Study design

Retrospective Cohort

Study population and setting

The study objective was to understand the level of SARS-CoV-2 transmission in 6 primary schools which invited 1047 children and 51 teachers to participate, and had a final study sample of 510 children (48.7%), and 42 teachers (82.3%). A total of 651 parents, 119 relatives of students, and 28 administrative school staff also participated. From April 28 to 30, 2020, researchers constructed a retrospective cohort in Crepy-en-Valois, north of Paris, using a questionnaire for history of recent symptoms and history of COVID-19 diagnosis from January 13, 2020 until April 28 to 30, 2020, when interviews were conducted. Blood samples were taken from all participants and tested for SARS-CoV-2 antibodies; this antibody test was previously reported to have 99.4% sensitivity for SARS-CoV-2 antibodies and 100% specificity. Sociodemographic and underlying medical history were also collected.

Summary of Main Findings

Among 1340 total participants, the total attack rate was 10.4% (n=139), which did not differ statistically by age or participant type (e.g., student, parent, staff, etc.). For students, the attack rate was 8.8% (n=45), for teachers 7.1% (n=3), and for all parents, 11.9% (n=76). From the epidemic curve, there was no clear pattern and no clear impact from school holiday closures on February 14. Parents of infected children had a higher attack rate (61.0%) compared to parents of non-infected children (6.9%), and relatives of infected children had an attack rate of 44.4% compared to relatives of non-infected children, with an attack rate of 9.1%. Loss of taste and smell were reported among 48% of adults, however in children, only fatigue and diarrhea were associated with testing positive for SARS-CoV-2 antibodies, and a high proportion of infected children were asymptomatic (41.4%). Only two people were hospitalized in the study, both parents, and there were no deaths reported. There was no evidence of transmission from primary school children to one another or to teachers/staff.

Study Strengths

This study tested for SARS-CoV-2 antibodies, which adds validity to their retrospective case definition. It also collected data from numerous participants from the schools, including children, their families, and teachers/staff, which helps to understand the broader effects of transmission in school besides transmission only among children. They also built an epidemic curve to understand the timing of infection through the school, which is especially important given there was little report of transmission in this community during January or February 2020.


The study may have recall bias for symptoms experienced and symptom onset used to construct the epidemic curve. Using antibody tests are important to identify past infection, however it is not clear when this infection occurred from the antibodies alone, therefore the researchers had to rely on self-report. There may also be selection bias, given the low rate of response among children (<50%), which may result in children/parents who are more concerned or who are more likely to have been sick to volunteer to participate. Additionally, school closure was somewhat soon (two weeks) after initial cases developed, suggesting with additional time further dynamics may have been observed.

Value added

This study is one of the first to use antibody tests to study infections in schools, including before and after holiday closures, and before major social distancing efforts were enforced.

This review was posted on: 15 July 2020