Introduction
A substantial number of children and teens worldwide have been infected with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), which causes coronavirus disease 2019 (COVID-19) [
1,
2]. Children and teens are susceptible to SARS-CoV-2 infection but are frequently asymptomatic or paucisymptomatic [
3]. In a small percentage of cases, complications such as pediatric inflammatory multisystem syndrome have been reported [
4,
5]. Although there are detailed descriptions of the acute clinical course in children in the medical literature [
3‐
5], few evidence-based studies have been published about the possible long-term morbidity of COVID-19 in the pediatric population, that is symptoms experienced after the acute phase of COVID-19 [
6].
According to the World Health Organization [
6], a Long- (or Post-) COVID condition is said to occur in individuals with a history of probable or confirmed SARS-CoV-2 infection, for a typical duration of 3 months after the onset of COVID-19. Long-COVID is considered to last for at least 2 months and cannot be explained by an alternative diagnosis. However, there are deviations in the definition of the time frame of Long-COVID [
6]. Typical symptoms include fatigue, shortness of breath, and cognitive dysfunction that generally have an impact on everyday functioning [
6]. In the adult population, there is growing evidence of Long-COVID morbidity. Antonelli et al. [
7] found that 10·8% of all patients who tested positive for the Delta variant of SARS-CoV-2 virus experienced Long-COVID, and 4·5% of all patients infected with the Omicron variant experienced Long-COVID. Greenhalgh et al. [
8] reported similar proportions, with approximately 10% of all patients experiencing Long-COVID. A subset of this population had serious sequalae that required intensive care but most Long-COVID patients reported mild symptoms such as cough, low grade fever, fatigue, and shortness of breath. In a prospective cohort study of 277 adults who recovered from COVID-19, Moreno-Perez et al. [
9] detected symptoms of Long-COVID in approximately 50% of the cohort. The most commonly reported symptoms were fatigue, respiratory complaints, and neurological complaints. Carvalho-Schneider et al. [
10] found that about two-thirds of all adults with non-critical COVID-19 were still experiencing some Long-COVID symptoms 60 days after the onset of the disease. Accordingly, Mendelson et al. [
11] argued that health care systems should develop approaches to address the need for continued care for Long-COVID patients.
Evidence for Long-COVID in the pediatric population surfaced later than data for the adult population. The first type of documentation published were case reports [
12‐
14], followed by small studies based on self or parental reports. Buonsenso et al. [
15] examined a sample of 129 individuals aged 18 or younger, and reported that 20 of the 30 participants who were assessed 60–120 days after infection had persistent symptoms.
More recent studies of Long-COVID in children have used larger cohorts, but the findings are inconsistent. Molteni et al. [
16] collected voluntary parental reports using a mobile application on 1734 children who tested positive for SARS-CoV-2, and their matched controls. They found that 4·4% of the recovering children had symptoms that lasted more than four weeks, whereas less than 1% of the matched controls were symptomatic for more than 28 days. Borch et al. [
17] defined Long-COVID as symptoms lasting more than 4 weeks after diagnosis and found that 0·8% of the children in the sample self-reported Long-COVID symptoms on an electronic questionnaire. The most common symptoms reported in this study were fatigue, loss of smell, loss of taste, muscle weakness, dizziness, respiratory problems, and chest pain. Pinto Pereira et al. [
18] administered questionnaires to compare 11–17 years olds, 6, and 12 months after their infection date and reported Long-COVID symptoms, especially tiredness, shortness of breath, poor quality of life, poor well-being, and fatigue. A meta-analysis of Long-COVID studies in children was conducted by Lopez-Leon et al. [
19] The most prevalent clinical manifestations were mood changes (16·5%), fatigue (9·6%), and sleep disorders (8·4%). Racine et al. [
20] conducted a meta-analysis on symptoms of depression and anxiety during the first year of the COVID-19 pandemic. Stephenson et al. [
21] found that adolescents who were infected with COVID-19 were more likely to experience mental and physical symptoms 3 months after the infection date. In a national cross-sectional study, Kikenborg et al. [
22] found a tendency towards better quality-of-life in the case group than in the controls, although more lasting symptoms were reported in the former.
Current studies of Long-COVID in pediatric populations tend to be based on small cohorts [
12‐
16] or on self- or parental-reports [
15‐
22], sometimes with a low response rate [
23]. Not all studies include a control group [
18‐
20,
24]. This makes it difficult to draw conclusions as to the profile or the prevalence of pediatric Long-COVID [
24]. Since studies on the general population suggest that Long-COVID places a considerable burden on health services, that may require similar preparation in the eventuality of future pandemics [
25], this study was designed to examine Long-COVID morbidity in a large pediatric population, based on objective clinical data, as compared to a matched control group. The findings can thus provide insights into the burden on service providers of Long-COVID in the pediatric population.
Discussion
This study investigated the associations between SARS-CoV-2 pediatric infections and recourse to health services 3–6 months after the infection, while controlling for socio-economic status, existing medical conditions, existing health service consumption, age, and date of diagnosis. Of the seven health services examined, the only statistically significant differences were for referrals for mental health services (RR 1·51 95%CI 1·15 − 1·96), and prescriptions (RR 1·03 95%CI 1·01–1·06), thus confirming the association between COVID-19 and long-term morbidity. Although the relative risk was high for referrals for mental health services, the rates of referrals in this age group were low, as was the risk difference (adjusted RD 0·001 95%CI 0·0006 − 0·002). Similarly, the risk difference for prescriptions was also small (adjusted RD 0·01 95%CI 0·004 − 0·02). There were no other statistically significant associations between the infection and other health services. This includes primary physician visits (RR 1·00 95%CI 0·98 − 1·01), specialty physician visits (RR 1·03 95%CI 1·00–1·07), ER visits (RR 1·1 95%CI 1·00–1·20), hospital admissions (RR 1·02 95%CI 0·86 − 1·21), and new diagnoses (RR 0·89 95%CI 0·79 − 1·01). However, some of these services are rarely used by children so that no definitive conclusions can be drawn despite the relatively large cohort in this study (N = 65,548 for the COVID-19 diagnosed group and for the control group). Hence, the evidence suggests that the impact of pediatric long- COVID on the health care system for this cohort was small.
The mechanism behind the increase in referrals for mental health services is not clear and can be attributed to either physiological or psychological reasons, or both. Another possible explanation may be related to the quarantine period (14 days), which was mandatory in Israel at the time for individuals with COVID-19. However, the quarantine in itself cannot explain the difference in referrals for mental health services because at the time of the study, anyone exposed to COVID-19 was required to quarantine, except individuals who had already had COVID-19. Note that the pediatric population of in the control group underwent more lab tests than the COVID-19 diagnosed group, both during the 3–6 months, and 1–3 months after the index date (see Figure
S3). Hence, the control group appears to have had a higher frequency of PCR tests as a result of exposure to individuals with COVID-19, thus leading to potentially higher frequencies of quarantine periods compared to quarantines in the COVID-19 diagnosed group.
One of the major strengths of this study was the use of nationwide data and objective outcome measures, unlike most other studies that have either been small in size [
12‐
16] or relied on self or parental reports [
15‐
22]. The large, detailed dataset made it possible to match the COVID-19 diagnosed group to a control group and adjust for additional confounders.
However, this study is not without limitations. Referrals to medical services are indirect measures of morbidity. In addition, data regarding the duration of the symptoms and the time to resolution of the medical complaints were unavailable. Further, some symptoms, such as fatigue, might not manifest in changes in the use of medical services patterns. Another possible limitation is that the COVID-19 diagnosed group was composed of children and teens who tested positive and did not necessarily include all the infected children at the time of the study. Nonetheless, during the first 3 waves of COVID-19 in Israel, testing was rather stringent, and infections would have been identified well [
30]. Note that despite the adjustments for past use of medical services, sex assigned at birth, age, date, societal sector, socio-economic status, and district, significant differences still remained between the COVID-19 diagnosed group and the control group 3–6 months prior to the index date. This was accounted for statistically in the main outcomes of this study by adjusting for these differences; however, other unmeasured risk factors may have been present.
Our analysis included controls who did not test positive before the index date. However, some controls tested positive for COVID-19 during the 6 month of post index-date follow-up period. To test the influence of these cases, we repeated the analyses while excluding the 2,348 matched pairs (3·58%) where controls tested positive in this timeframe. The reproduced results in Table
S4 showed very high congruence with our main results, affirming that our findings are robust with respect to the design choice.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.