At the beginning of the COVID-19 epidemic, initial data from China indicated that children were significantly less affected by COVID-19 than adults and more frequently presented with mild and asymptomatic forms of the disease than adults (1). This has been later confirmed by many other publications and by seroprevalence studies from different parts of the world (2, 3). However, there has been a suggestion that children older than 10 years and adolescents exhibit features closer to that of young adults (2). In the USA, the weekly incidence of COVID-19 cases among adolescents aged 12–17 years was approximately twice compared with that of children aged 5–11 years between March and September 2020 (4). Evidence demonstrated that the SARS-CoV-2 viral-specific antibody response profiles vary with age and differ between younger children, adolescents, and young adults (5).

Most studies combine children in different age groups and lack sufficient numbers to explore them in detail age specificities. Here, we report data on a population-based sample of 2,555 children at the pivotal age of 9 years to gain more insight into the transition between younger and older child-type responses to SARS-CoV-2 infection.

The data originated from the child cohorts of the SAPRIS (“SAnté, Perception, pratiques, Relations et Inégalités Sociales en population générale pendant la crise COVID-19”) project, a consortium of prospective French cohort studies involving three general population-based adult cohorts, aside from the two child cohorts (6). The goal of the present analysis is to describe (1) the symptoms compatible with a SARS-CoV-2 infection in the two involved cohorts of children born in 2011 during or right after the first lockdown in France (from 17 March 2020 to 11 May 2020) and (2) the prevalence of SARS-CoV-2 ELISA Spike seropositivity in children and one of their parents.

In April 2020, the participants in two French nationwide cohorts of children, namely, Elfe and Epipage2, were invited to take part in an online survey about COVID-19-related symptoms and family life during the lockdown.

Both cohorts had enrolled children at birth in 2011. Children born at term, moderate preterm, or late preterm (after 33 weeks of gestation) were participants in the Elfe cohort and enrolled on 25 selected days over 2011, from a random sample of 320 maternity units in mainland France (7). Children born extremely preterm (below 35 weeks) were participants in the Epipage2 cohort. All maternity units in French mainland regions (but one) and in overseas regions were involved in the Epipage2 inclusion phase over periods that lasted from 5 weeks to 8 months according to three gestational age strata (8).

All families who had not withdrawn from the cohorts and had an email address known by the management teams were eligible for the COVID-19 survey.

Elfe children born before 35 weeks of gestation were excluded from the present analyses as these preterm children are best represented by the Epipage2 cohort.

Eligible participants for the COVID-19 survey were 12,814 and 2,922 in the Elfe and Epipage2 cohorts, respectively. Among them, 3,894 (30%) and 841 (29%), respectively, responded to the second survey questionnaire and were proposed to participate in the serological survey. The acceptance rate of children for the serological survey was approximately 60% (n = 2,358 and 498, respectively), which was similar for the two cohorts. A total of 2,828 DBS from children were received at the Biobank. Among them, 2,680 (95%) could be punched and were sent for analysis, and 2,642 serological results were returned. After the exclusion of the missing values of the sampling date and children that could not be included in the weighing procedure, data collected from 2,555 children were included in the statistical analysis. Among 2,400, one parent had also provided a DBS that could be analyzed.

The crude and weighted characteristics of the children are presented in Table 2. The crude prevalence of a positive ELISA Spike serology was not statistically different in the preterm children of the Epipage2 cohort (2.4%) and the term children of the Elfe cohort (1.7%, p = 0.40); therefore, results were given for the pooled sample.

The weighted prevalence of a positive ELISA Spike serology was 2.8% in 9-year-old children (95% CI: 1.7%–4.0%). It showed the expected gradient according to the regional viral contamination rates before the lockdown, ranging from 6.9% (2.4%–11.4%) in Grand Est to 0.1% (0.0%–0.3%) in Nouvelle Aquitaine (Table 3). In the whole sample, a positive serology was found in 8.6% (7.4%–9.7%) of parents who provided blood (the mother in 76% and the father in 24% of cases), which was more than three times higher than that in their children.

Table 4 shows a significant association (p < 0.001) between the serologies of the child and the parent from the same household with an odds ratio of 13.8 (7.9–24.2).

The clinical symptoms of the “possible COVID-19” definition that lasted for at least 1 day had been more frequently declared (13.4%) during the lockdown when ELISA Spike serology was positive than when it was negative or indeterminate (3.8%, p = 0.02).

Among the symptoms proposed in the questionnaire, those that lasted for at least 1 day and had been significantly more frequently declared in case of positive serology were anosmia/ageusia, fever, fatigue, dyspnea, and chest pain (Table 1).

Of the children with positive serology, 74% (43/58) (CI: 63%–86%) reported no symptoms in the past 14 days (i.e., during the lockdown), and of them, 71% (CI: 58%–83%) reported no fever nor cough earlier in the year.

From this sample and according to ELISA Spike serology, it is estimated that 2.8% (95% CI: 1.7–4.0) of 9-year-old children born in France in 2011 had produced antibodies against the SARS-CoV-2 spike protein before the summer of 2020 in France, which is three times less than that of the parents living in the same household. In the presence of one infected parent in the household, only 14% of children had a positive serology.

In France, schools closed from 16 March 2020 to June, except for children of first-line workers. Reopening took place on 22 June 2020 for the whole country but started on May 11 in districts with low SARS-CoV-2 circulation. After the first lockdown, the incidence rate of COVID-19 as assessed by positive PCR tests in the 7 previous days decreased to below 5/100,000 inhab (11), before rising again at the end of the summer. We can assume that in our study a majority of the children and their parents with positive SARS-CoV-2 serology had been contaminated before or during the first weeks of the lockdown as most DBS were performed before the end of July.

Serological data on children after the first COVID-19 wave are scarce in France, and none has a national scope. To the best of our knowledge, only two are population-based. The first study was performed in the city north of Paris where the first COVID-19 outbreak in schools was recognized in France at the beginning of the epidemic. A serological survey performed in schools in the Haut de France region in March–April 2020 before any protection measure was taken found that 35% of 664 children aged 12–17 years who participated in the study were seropositive in high school where the outbreak occurred and only 9% of 1,340 children in primary schools of the city (12).

The second study was undertaken in the city of Nancy in a random sample of households drawn from the electoral list. It yielded a SARS-CoV-2 seropositivity prevalence of 1% (95% CI: 0.1–3.5) in 5–19-year-old children (n 203) in June 2020 (13). Two other studies were performed in clinical settings. In a study based on the serum samples collected from children requiring hospital admission in North Eastern France for other causes than COVID-19 between February and August 2020, seropositivity was estimated at 4.0% in 1,400 children aged ≤19 years (14). In the Ile-de-France region, 27 ambulatory pediatric practices participated in a fingerstick serological survey in April–May 2020. Data were collected from 605 children at a mean age of 5 years, of whom 204 were asymptomatic, 118 had COVID-19-compatible symptoms between 7 days and 2 months before enrollment, and 283 were symptomatic. SARS-CoV-2 seropositivity was 6.4% (95% CI: 3.4–10.7), 23.7% (16.4–32.7), and 8.5% (5.5–12.5), respectively (15). Although all these studies used different methods for SARS-CoV-2 serology, they are coherent with our regional estimates in 9-year-old children that rank from 0.1% to 6.9% in the general population depending on the intensity of SARS-CoV-2 circulation.

A number of studies have found that the seroprevalence rate is lower in children and, for some studies, in adolescents than that in adults (2). We confirmed this result in 9-year-old children and adults sharing the same household. In a household transmission survey in Geneva (16), compared with 20–49-year olds, 5–9-year olds had less than half the odds ratio of seropositivity in the presence of a positive household member, which confirms their low susceptibility to SARS-CoV-2 infections. On the other hand, in our study, parental seropositivity was seven times higher (53%) when their child was seropositive compared with negative or indeterminate. A similar finding was found in the Haut de France survey for primary school pupils in contrast to high school pupils (12). Closer contacts between young children and their parents than between adolescents and their parents has been hypothesized to explain this finding, and, in that matter, 9-year-olds seem to behave as younger children. Although we cannot determine who was the index case in the household, it was more likely the parent or another adult member of the household, given the differences in the infection rates between children and adults. This suggests that close contacts with an infected adult explained a substantial proportion of infections with the initial SARS-CoV-2 variant in 9-year-old children and therefore the high frequency of infected adults in their surroundings.

However, children have a more robust innate and/or mucosal immune response to SARS-CoV-2 than adults, which translates into the absence of adaptive immune solicitation and antibody production. The difference in seroconversion between children and adults may also be due in part to a difference in cellular immune responses (T and B cells) (17). Therefore, the absence of ELISA Spike seropositivity may not always indicate an absence of infection. In addition, our estimates are based only on ELISA Spike seropositivity, which does not have 100% specificity and sensibility, which may induce some misclassification.

The clinical symptoms associated with seropositivity in our study are in accordance with those described in other studies in children (3, 12, 18, 19) except for gastro-intestinal symptoms that were not statistically different between seropositive children and the others, although in the expected direction. Most seropositive children reported no symptoms in the first 2 weeks of the lockdown or earlier in the year. In contrast to the SAPRIS study performed in adult cohorts, only 19% of seropositive adults reported no symptoms (20), pointing again to the difference in the immune response between 9-year-old children and adults.

To conclude, we have shown that 9-year-old children had a lower susceptibility to SARS-CoV-2 infection than adults with the initial Chinese strain, similar to younger children, and estimated that approximately 3% of them have developed antibodies against SARS-CoV-2 in France after the first wave of the COVID-19 epidemics.