Background
The role of children and adolescents in the transmission of severe acute respiratory syndrome coronavirus type 2 (SARS-CoV-2) is still not fully understood [
1,
2]. Children mostly have mild or no symptoms of coronavirus disease 2019 (COVID-19) if infected by SARS-CoV-2 [
3] and long COVID does occur but is much less frequent than in adults [
4‐
6]. While children are rarely seriously affected themselves, the extent to which they can spread the virus and put more vulnerable groups at risk remains debatable.
Household transmission accounts for a substantial number of SARS-CoV-2 infections [
7‐
9]. However, robust data on transmission patterns in household is scarce [
10] and evidence on the association between living in a household with children and SARS-CoV-2 seropositivity is inconsistent. Large population-based studies in the UK, Denmark, and France suggest a higher prevalence of SARS-CoV-2 infections for people living in a household with children [
11‐
14]. In contrast, studies among healthcare workers showed an opposite effect [
15,
16]. This indicates that there may also be protective effects related to living in a household with children. Whether protective- or risk-increasing effects prevail might depend on the presence of other factors (e.g., time spent at home due to employment status/role as caregiver, behavioral factors such as non-household contacts, adherence to social distancing and hygiene measures). These factors are known to affect both intra- and extra household SARS-CoV-2 transmission risk, but their role is still unknown and thus needs further exploration.
Most studies that have assessed the association of living with children and SARS-CoV-2 infections so far stand out for their large sample sizes. However, many of them are retrospective and based on data from reverse transcriptase polymerase chain reaction (RT-PCR) confirmed infections. As a result, they could overrepresent symptomatic infections, while asymptomatic infections are misclassified, as individuals with subclinical infections are less likely to undergo PCR testing. Furthermore, testing policies have varied throughout the pandemic phases due to reagents availability and testing capacities. Serological studies have been pointed out as reliable tool to understand the full spectrum of symptoms of COVID-19, and the spread of SARS-CoV-2 infections [
17,
18]. In combination with a random selection of participants, and pre-defined questionnaires, serological studies allow a more targeted assessment of variables and are less prone to recall- and other information biases.
This study used SARS-CoV-2 serology results and questionnaire data from working age participants (18–64 years) of the
Corona Immunitas research program in Switzerland [
19]. The main objective of this study was to investigate whether living in a household with children is associated with SARS-CoV-2 seropositivity in adults. More specifically we investigated first if SARS-CoV-2 seropositivity in adults is associated with the number of children in the household, and with specific age groups of children and secondly, if this association varies by frequency of contacts outside the household, employment status, and gender.
Discussion
In this cross-sectional study, we analyzed the association between SARS-CoV-2 seropositivity and living in a household with children, based on questionnaire and serology data from population-based SARS-CoV-2 seroprevalence studies in 11 cantons in Switzerland. Our results suggest that for the general population overall odds of SARS-CoV-2 seropositivity may be higher when a household is shared with at least one child of any age. However, despite this association in the general population, risk of SARS-CoV-2 infection related to living in a household with children might vary between subgroups. Indeed, interaction analyses indicated higher odds of SARS-CoV-2 seropositivity when living in a household with children for men than for women.
Although the precision (described by the confidence interval) of the overall analysis limits our ability to make definite general statements, the finding of a positive association of living with children and SARS-CoV-2 seropositivity in our sample of general adult population in Switzerland is consistent with other population-based SARS-CoV-2 studies [
11‐
14]. Results of the multivariable analysis in our study were very similar to what has been observed by Carrat et al. based on seroprevalence data in France, where a comparable population-based SARS-CoV-2 antibody testing approach has been followed [13]. A positive association of living with children and SARS-CoV-2 infection was also reported in population-based studies in Denmark [
12] and during the second wave in the UK [
11], which have calculated hazard ratios based on data from PCR-confirmed infections. While taken these results together, all studies agree that for the general population an association is very likely, our study in line with the other studies indicates that the strength of this association is only weak to moderate. Neither adjusted odd ratios in our study (OR 1.25 [0.99–1.59]) and reported by Carrat et al. from France (OR 1.3 [1.11–1.53]) nor hazard ratios reported by Husby et al. from Denmark (hazard ratio 1.05 [1.02–1.09] and by Forbes et al. from the second wave in the UK (hazard ratio 1.06 [1.05–1.08] and 1.22 [1.20–1.24] for living with children aged 0–11 and 12–17, respectively) indicates a strong increase in risk of SARS-CoV-2 when a household is shared with children [
11‐
13].
Next to our observations on the association of living with children and SARS-CoV-2 infection in general, we observed that the odds of SARS-CoV-2 seropositivity increases with every additional child in the household. A similar finding has been made by Husby et al. based on PCR-data in Denmark, which has shown a positive trend of higher hazard ratios for SARS-CoV-2 infection when living in a household with one, two, and three or more children, respectively [
12]. Odds ratios were similar for all children’s age groups. Results were comparable to the observations made in the overall analysis, although confidence intervals were wider (likely due to the lower number of cases in this subanalysis). Therefore, our results do not indicate that living with children of a particular age group has a stronger association with seropositivity than living with children of other age groups. Differences in number of non-household contacts have been proposed as a potential behavioral reason for an altered SARS-CoV-2 infection risk, when living with children. For example, Husby et al. suggested that living with children might lead to more non-household contacts as caregivers accompany their children on playdates [
12] while another study suggests people living with children have less non-household contacts, because they are more likely to stay at home and spend time as a family [
15]. In our study, the odds of SARS-CoV-2 seropositivity were not higher for people with at least one non-household contact reported within the previous seven days. Thus, differences in contact patterns likely do not explain an altered SARS-CoV-2 infection risk related to living in a household with children. In line with other studies [
14,
31], being unemployed or working from home was inversely associated with seropositivity, which is likely explained by the substantially lower work and commute-related risk of SARS-CoV-2 transmission. Interaction analyses for both non-household contacts and employment status showed no clear picture of an effect modification. Consistent with findings from other studies, SARS-CoV-2 seropositivity related to living with children was dependent on gender [
11,
12]. While in our analysis SARS-CoV-2 seropositivity was not associated with gender itself, male participants had substantially higher rates of SARS-CoV-2 seropositivity if they lived in a household with children.
As antibody tests cannot identify the exact timepoint of infection and participants’ household members were not tested, we were unable to define transmission routes or determine the extent to which living with children and SARS-CoV-2 infections in adults are causally related. Children being able to get infected and transmit the virus to others could impose a direct risk to adults living in the same household. At the same time, there are other individual characteristics and behavioral factors that are associated with sharing a household with children, which could be of importance regarding SARS-CoV-2 infection risk.
There is some evidence indicating that children could increase the risk of a SARS-CoV-2 infection in household members by being a direct source of infection. PCR-based studies have shown a higher secondary attack rate of paediatric compared to adult index cases in households [
37,
38]. With milder clinical manifestations [
39‐
41], SARS-CoV-2 infections in children tend to remain undiagnosed [
42] and unnoticed SARS-CoV-2 infections in children could keep households from implementing necessary isolation measures to interrupt transmission chains. Further, young children cannot be isolated from their caregivers when being sick [
43], and social distancing and hygiene measures are more difficult to implement. Also, restrictions in extra-curricular activities in early 2021 in Switzerland were limited only for children older than 12 years [
2] and apart from a relatively short period during the first wave in March–April 2020, schools in Switzerland have remained open. School-aged children and adolescents have shown to be more mobile and tend to have more close contacts with individuals outside the household [
44]. As such, the relative risk of children to acquire a SARS-CoV-2 infection outside the household might have been substantially higher in comparison to adults.
On the contrary, several studies have shown that children only account for a very small proportion of index cases within household clusters [
45‐
47] and infectivity is lower when index cases are asymptomatic [
9,
37,
48], as it is frequently the case in children. Our results provide no evidence that the odds of SARS-CoV-2 seropositivity are substantially increased in individuals spending more time at home with contact to children (e.g., people being unemployed or performing work from home, people having no non-household contacts). Furthermore, our results and other studies [
15,
16] show that although there is an association in the general population not everybody living with children is at increased risk of SARS-CoV-2 infection and for certain groups even the opposite is the case (i.e., odds of SARS-CoV-2 infections are lower when living with children). In our study, SARS-CoV-2 seropositivity for men living with children was substantially increased. A negative association between living with children and SARS-CoV-2 seropositivity was also observed among healthcare workers in Scotland [
15] and Switzerland [
16].
This suggests that risk of SARS-CoV-2 infection related to living with children is not only determined by the child’s infectivity, but other characteristics and behavioural factors do play a role, of which some can also be protective. One of those protective effects might be related to seasonally spreading human endemic coronaviruses, due to which children might possess some cross-immunity to SARS-CoV-2 [
49,
50], and more frequent co-infections with other viruses could interfere with the replication of SARS-CoV-2 [
49]. Individuals spending a lot of time at home taking care of children could profit of such cross-immunity and other competing infections, due to increased exposure. This effect could be particularly prominent in women who, already before the pandemic, more often spent time at home taking care of children [
33]. Living with children further leads to more part-time work, as more time is allocated to childcare. People working in professions with high risk of SARS-CoV-2 transmission such as healthcare [
51,
52] could thereby profit from living with children, as they spend less time at work being exposed to SARS-CoV-2. At the same time, in healthcare workers, social distancing and hygiene measures might have been better implemented, that have shown to significantly reduce transmission risk from a child, if infected [
53]. Although we were unable to assess these factors related to working in healthcare jobs in our study, this could to some extent explain the discrepancies of our findings in comparison with studies conducted among healthcare workers [
15,
16].
With our study design using retrospectively reported symptoms, we were unable to define if severity of COVID-19 is different for people living with children versus people living without. Number of SARS-CoV-2-related hospital admissions was low in our population-based cohort (total number: 10, of which 3 living with children). Thus, in contrast to previous studies [
11,
12,
15], we were unable to use this variable as a proxy to assess severity of COVID-19. An additional analysis on self-reported symptoms in participants revealed that most symptoms occur with similar frequencies in seropositive individuals with children compared to seropositive individuals without children (see additional file
1: Table S1). For seronegative individuals, symptoms were generally more frequent in people living with children compared to people living without, indicating that people living in a household with children might get more often sick, but this is likely not related to SARS-CoV-2.
Despite its strength of being a large nation-wide population-based study using a highly sensitive and specific SARS-CoV-2 antibody test, this study has some limitations. Overall participation rate was moderate and non-random willingness to participate in the survey is possible (e.g., higher participation of more highly educated individuals, individuals who experienced symptoms suggestive of COVID-19, individuals with a confirmed SARS-CoV-2 infection, or individuals who were exposed or believed themselves to be exposed to SARS-CoV-2). However, we have no reason to believe that non-random willingness to participate in the study has been substantially different among people living with compared to people living without children. Although some selection bias cannot be ruled out, we expect the effect, if any, to be of very small influence on the overall results. Due to the limited number of participants, some associations could not be estimated precisely. Especially results of the analysis of subgroups as well as interaction analysis should thus be interpreted with caution. Participants’ characteristics were furthermore self-reported, and some variables could be only assessed as approximations (see number of non-household contacts). Accuracy on some variables might be limited and, in some cases, recall bias could have occurred. Despite the high sensitivity and specificity of the antibody test used, some misclassification is possible and effect sizes may be biased towards the null and thus some associations could have been missed. Especially in case of SARS-CoV-2 infections that date back longer, SARS-CoV-2-specific antibodies might have waned. However, as we expect this to have happened irrespective of living with children, overall results are likely not substantially affected. In general, with serology results, we were unable to determine the timepoint when infection occurred. Infection risk related to living with children could have been different with different transmission dynamics at different stages of the SARS-CoV-2 pandemic in Switzerland. Our study most accurately depicts the pandemic situation during a time of high community transmission during the second wave in Switzerland. Given that SARS-CoV-2 antibodies take around 6–15 days to become detectable [
22], very few samples taken as part of our study (those that were collected in October 2020, when community transmission was just starting to rise in Switzerland), would not represent second-wave transmission dynamics. Transmissions within households with the delta and omicron variant of SARS-CoV-2 appears to be further increased [
54,
55]. However, these strains did not become prevalent in Switzerland until May 2021 (delta variant) and November 2021 (omicron variant), respectively [
21], later than our data collection. Next to variation in containment measures, intra household transmission could have been lower at later stages of the pandemic, because with increasing number of immune household members (due to previous infection or vaccination) also non-immune household members are protected [
56].
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.