Factors associated with hospital and intensive care admission in paediatric SARS-CoV-2 infection: a prospective nationwide observational cohort study

Coronavirus disease 2019 (COVID-19) is usually less severe in children compared to adults. This study describes detailed clinical characteristics, treatment and outcomes of children with COVID-19 in a non-hospitalised and hospitalised setting and quantifies factors associated with admission to hospital and intensive care unit in children with SARS-CoV-2 infection on a nationwide level. Data were collected through the Swiss Paediatric Surveillance Unit from children < 18 years with confirmed SARS-CoV-2 infection. All 33 paediatric hospitals in Switzerland reported non-hospitalised and hospitalised cases from March 1 to October 31, 2020 during both pandemic peaks. In total, 678 children were included. The median age was 12.2 years (IQR 5.0–14.6), 316 (46.6%) were female and 106 (15.6%) had comorbidities. Overall, 126 (18.6%) children were hospitalised of whom 16 (12.7%) required ICU admission. Comorbidities were the only factor associated with hospital admission in a multivariable regression analysis (odds ratio 3.23, 95%CI 1.89 to 5.50; p-value < 0.01). Children with preexisting comorbidities did not require ICU admission more often. Hospitalised children more often presented with fever (96 [76.2%] vs 209 [38.1%], p-value < 0.01) and rash (16 [12.8%] vs 6 [1.1%], p-value < 0.01). Anosmia/dysgeusia was more prevalent in non-hospitalised children (73 [13.3%] vs 3 [2.4%], p-value < 0.01). In hospitalised children, oxygen treatment was required in 34 (27.0%), inotropes in nine (7.3%) and mechanical ventilation in eight (6.3%) cases. Complications were reported in 28 (4.1%) children with cardiovascular complications being most frequent (12 [1.8%]). Three deaths were recorded. Conclusion: This study confirms that COVID-19 is mostly a mild disease in children. Fever, rash and comorbidities are associated with higher admission rates. Continuous observation is necessary to further understand paediatric COVID-19, guide therapy and evaluate the necessity for vaccination in children. What is Known: • Clinical manifestations of SARS-CoV-2 infection in children vary from asymptomatic to critical disease requiring intensive care unit admission. • Most studies are based on hospitalised children only; currently, there is limited data on non-hospitalised children. What is New: • The clinical spectrum and severity of COVID-19 is influenced by age: in children less than 2 years, fever, cough and rhinorrhoea are the most common symptoms and in adolescents, fever, cough and headache are more common. • Hospitalised children more often presented with fever and rash, while anosmia/dysgeusia is more prevalent in non-hospitalised children. • Children with pre-existing comorbidities are more frequently hospitalised but do not require ICU admission more often. Supplementary Information The online version contains supplementary material available at 10.1007/s00431-021-04276-9.

To date, the data on SARS-CoV-2 infection in children and adolescents either come from non-hospitalised children with limited clinical information or from hospitalised children only. This limits the knowledge in paediatric COVID-19 on factors associated with admission including clinical presentation and risk factors such as age, sex or comorbidities. This study presents epidemiological data from active surveillance of SARS-CoV-2 infections in non-hospitalised and hospitalised children in Switzerland and provides insights regarding risk factors for admission.

Study design and population
Our study is a prospective nationwide observational cohort study that describes detailed clinical characteristics and outcomes of children with laboratory-confirmed COVID-19 in a non-hospitalised and hospitalised setting. Paediatric SARS-CoV-2 infections are actively monitored in an observational study by the Swiss Paediatric Surveillance Unit (SPSU, http:// www. spsu. ch) since March 2020. The current analysis includes data from March 1 to October 31, 2020. All 33 paediatric and neonatological hospitals in Switzerland participated and notify cases monthly. Upon notification, the investigators sent the reporting centres an electronic clinical report form through RedCap or in paper form (see supplementary data Questionnaire) [16]. All data were recorded anonymously and reviewed by the investigators and further clarified with the reporting physician when needed. The study has received ethical approval by the Ethikkommission Nordwest-und Zentralschweiz (EKNZ 2020-01,130).

Case definition
Children and adolescents < 18 years of age who presented to a Swiss paediatric hospital and received ambulatory or hospitalised care were included if diagnosed with COVID-19 by detection of SARS-CoV-2 from a clinical specimen using a validated polymerase chain reaction (PCR) test or serology. Retrospective screening of patients with the following criteria was used for identification of potential PIMS-TS cases: PIMS-TS reported by the clinician, SARS-CoV-2 serology performed, ICU admission or cardiac changes.

Statistical analysis
Continuous data were summarised using median and interquartile ranges. Categorical data were presented as percentage. Categorical data were compared using the χ 2 test, with p-values < 0.05 considered as significant. Cooccurrence symptoms were clustered using the K-means clustering solution and visualised with a heat map. A multivariable logistic regression model of risk of admission was fitted by including all variables used for univariable analysis origin. For the assessment of age, the following groups were made 0 to < 2, 2 to < 5, 5 to < 10 and 10 to < 18 years of age. R (Version 1.2.5019) was used for statistical analyses.

Study population
A detailed dataset was returned for 682 cases, of which 678 were included in the final analysis. Reasons for exclusion were duplication in reporting (n = 3) and age ≥ 18 years (n = 1). The age of the children ranged from 7 days to 17.9 years with a median of 12.2 years (interquartile range (IQR) 5.0-14.6) (
Overall, most children (646 [95.3%]) did not receive medication. Specific treatment was given to 10 (1.8%) of non-hospitalised, 15 (13.6%) of hospitalised and 12 (75.0%) of ICU-admitted children. Among hospitalised children (non-ICU admitted), six (5.5%) received corticosteroids, two (1.8%) each hydroxychloroquine and intravenous immunoglobulins (IVIG) and one (0.9%) tocilizumab. Among ICU-admitted children, nine (56.3%) received biologicals (anakinra (n = 7), tocilizumab (n = 2)), seven (43.8%) each corticosteroids, IVIG and two (12.5%) hydroxychloroquine. A combination of IVIG and corticosteroids was given to five children with suspected PIMS-TS in the ICU group. No further treatment including remdesivir was given (supplementary data Questionnaire). The median duration of hospitalisation for non-ICU admit-   Table 2). Hospitalised children had significantly more comorbidities than non-hospitalised children (p-value < 0.01). Five (31.3%) children admitted to ICU had pre-existing comorbidities: three children had asthma/ bronchitis, a neonate had apnoea of prematurity (born at 29 weeks gestational age) and a 2-month-old infant had isolated microcephaly with a normal cerebral ultrasound. Children requiring admission to ICU did not have pre-existing medical conditions more frequently compared to non-ICUhospitalised children. A univariable regression analysis showed that children ≥ 2 years of age were less likely to be hospitalised, and comorbidities increased the risk of hospitalisation in both uni-and multivariable regression analysis (Table 3).

Symptoms
Overall, fever was the most frequent symptom observed among children with COVID-19 (305 [45.3%]) ( Table 1). In children aged less than 2 years of age, fever, cough and rhinorrhoea were the most common symptoms, and in adolescents between 10 and 18 years of age, fever, cough and headache were more commonly reported (a detailed distribution of symptoms according to age group is presented in Fig. 2). Fever and rash were more common in hospitalised compared with non-hospitalised children ( (Fig. 3). The first cluster represents an upper respiratory tract illness with fever, cough, rhinorrhoea and pharyngitis; the second a gastrointestinal illness with abdominal pain, diarrhoea and vomiting, and the third cluster corresponds to more constitutional symptoms with headache, myalgia and asthenia.

Complications
A total of 28 (4.1%) children with SARS-CoV-2 infection developed complications; this was more frequent in hospitalised   In hospitalised children, three (2.4%) deaths were reported during the study period. A 10-month-old infant with severe brain oedema, signs of hypoxic-ischemic encephalopathy and cerebral haemorrhages presenting with vomiting without fever, respiratory distress (with evidence of respiratory illness on chest radiography) and status epilepticus. A 2-month-old infant with cardiorespiratory arrest secondary to cerebral haemorrhage presenting with acholic stools, haematochezia, haematemesis and hepatosplenomegaly with abnormal liver function. A 6-year-old with an Epstein-Barr Virus (EBV)-associated hemophagocytic lymphohistiocytosis developed cerebral haemorrhage due to coagulation deficiency caused by liver failure. He presented with abdominal pain, fever and abnormal liver function 2 weeks after a SARS-CoV-2 infection and was initially suspected to have PIMS-TS.

Diagnosis
Diagnosis of SARS-CoV-2 infection was mostly confirmed by PCR (663 [98.1%]). Of the 40 children in * Numbers of hospitalised children include those admitted to intensive care unit (ICU)   Table S1).

Transmission
In total, 309 children (45.6%) had a family member with a confirmed or suspected SARS-CoV-2 infection (supplementary data Fig. S4). Community-acquired infection (including school and day-care) was confirmed or suspected in 86 (12.7%) children. In one-third of children, the primary case was unknown 284 (41.9%).

Discussion
This nationwide study presents the first epidemiological data from active surveillance of SARS-CoV-2 infection in non-hospitalized and hospitalized children in Switzerland. Data have been collected during both pandemic waves: the first wave was marked by general lockdown with school closures. The second wave was characeristed by an important rise of cases in October, which was similar to the evolution of the pandemic in the adult population and was followed by a second lockdown in November [17]. Our findings suggest that comorbidities are an important factor associated with two times higher hospitalisation rates in children and adolescents. Data from the Centres of Disease Control surveillance of in the USA suggest that overall hospitalisation rates are six times higher among children with a reported underlying condition [18]. In children ≤ 9 years of age, the rate of underlying conditions was reported to be 4.1% overall and 22.3% in those hospitalised. We have found similar rates of comorbidities in hospitalised but higher rates in non-hospitalised children. The latter may be explained by the fact that our data are based on a hospital-centred surveillance, which Fig. 2 Symptoms distribution in non-hospitalised and hospitalised children with SARS-CoV-2 infections according to age group. Other symptoms included conjunctivitis, otalgia, cheilitis, hand oedema, thoracic pain, arthralgia, acrocyanosis, fainting, seizure, alguria, orchitis and macro-haematuria likely underestimates the burden of disease in non-hospitalised children. In addition, children with comorbidities were more likely tested and referred to hospitals, particularly in the early phase of the pandemic. This is also reflected by the fact that we observed a more severe spectrum of paediatric COVID-19 with a lower rate of asymptomatic children compared to other studies [19,20]. Another paediatric study from the UK reported much higher rates of comorbidities of 42% in hospitalised children, but this study did not include data on non-hospitalised cases [1]. The most frequent comorbidities reported in the UK study were neurological, asthma and haemato-oncological or immunological which compares to the spectrum reported in our study. Whether underlying conditions are associated with increased severity of disease or a lower threshold for admission remains unclear based on our data.
In our setting, 13% of hospitalised children required admission to ICU, which is in line with other European studies [1,19,21]. Similarly, to other studies, children with serious medical conditions (immunodeficiency, haematooncological, cardiac or metabolic disease) in our cohort did not develop severe COVID-19 requiring ICU admission more often than previously healthy children [22]. A recent systematic review suggests that children with underlying cardiac disease might be at increased risk of severe COVID-19 [23]. However, we did not find this in our study population, where children with underlying cardiac disease (including one patient after heart transplantation) did not experience a severe disease course. Three children with asthma/bronchitis required ICU admission due to respiratory failure. Evidence are sparse on whether asthma is a risk factor for severe COVID-19 in children; a systematic review found only two reports considering asthma as a risk factor for SARS-CoV-2 infection [24]. However, a large US retrospective paediatric study found that the risk of testing positive for SARS-CoV-2 was reduced in children with a respiratory disorder, including asthma [25]. Moreover, children with asthma are underrepresented during this pandemic, and studies in adults have not identified SARS-CoV-2 as a trigger for asthma exacerbations [26]. Several hypotheses have been proposed to explain this unexpected finding, including an impaired immune response to SARS-CoV-2. Children with asthma have lower levels of interferon-gamma and consequently reduced angiotensin-converting enzyme 2 (ACE-2) gene expression in airway epithelium, which act as the entry point of SARS-CoV-2 into the human body [26]. Other studies report reduced ACE-2 expression in patients with allergic asthma or after inhaled corticosteroids [27,28].
Our study clearly shows that the clinical spectrum of COVID-19 is different in hospitalised and non-hospitalised children with fever and rash being more common in those admitted, but cough, rhinorrhoea and pharyngitis being comparable. Although fever was the most frequent symptom, especially among young children, the incidence was lower than reported by other multicentric studies [1,19,21]. This lower prevalence might reflect the higher proportion of adolescents in our cohort. Age is a further important factor influencing the clinical spectrum of disease with fever and respiratory symptoms of lower severity being more frequent in children below 2 years of age. Older children more likely present with non-respiratory symptoms, and anosmia/ dysgeusia was only rarely recorded in children less than 10 years of age. A similar spectrum was described in the UK paediatric cohort; however, this only included hospitalised children [1].
Our findings confirm that paediatric COVID-19 is mostly a mild illness. Several hypothesis have been proposed to explain the milder disease seen in children, including an age-related difference in the immune response, with a stronger innate immune response in children and agerelated differences in expression of ACE-2 [5,29]. Our data also captures cases with the suspicion of PIMS-TS corresponding to the most severe spectrum of COVID-19. Of the three deaths reported, two were being temporally related to SARS-CoV-2 infection without another disease identified.
Evidence-based treatment options for paediatric COVID-19 are still lacking [30]. In our study, the majority of non-hospitalised but also hospitalised children did not receive specific treatment. However, children admitted to ICU with PIMS-TS required inotropic support and received systemic corticosteroid and IVIG in combination with immunomodulators (including anakinra (interleukin-1 inhibitor) and tocilizumab (interleukin-6 inhibitor)). Similarly, most case series report using IVIG and/ or corticosteroid as main therapeutic option for children with PIMS-TS features, given the similitude with Kawasaki disease [14,15,[31][32][33][34]. The rational for using anticytokine agents (anakinra or tocilizumab) is that these toxic shock-like clinical presentations are caused by exaggerated immune response and excessive cytokine release, called cytokine storm, rather than by viral replication itself [35]. However, there are insufficient data to recommend for or against the use of specific treatment for PIMS-TS, and management remains centred on effective supportive care [30].
The strengths of our study are the multicentre nationwide study design capturing the widespread epidemiology of SARS-CoV-2 infection in a European country, the inclusion of a large number of children including non-hospitalised children and the detailed demographic and clinical data. However, our study overestimates the admission rate as we did not include patients tested in private practices and as some children were hospitalised for reasons other than SARS-CoV-2 infection. Furthermore, the threshold for admission might have been different depending on local guidelines, pre-existing comorbidities, physicians' discretion and presentations outside office hours. Data from the Federal Office of Public Health show a 0.9% admission rate in children in the same time period (personal communication from Mirjam Mäusezahl, Federal Office of Public Health, 23 March 2021). In addition, in the early weeks of the COVID-19 pandemic, SARS-CoV-2 testing was not considered a priority in children, and only children with severe symptoms or persistent fever were tested. Testing strategies also differed regionally which may have affected some of the observed regional differences. Analysis of weight and obesity as a potential risk factor for hospitalisation could not be included as weight was not recorded in most non-hospitalised cases. We are also unable to analyse children classified as PIMS-TS in detail as specific data on PIMS-TS cases are only prospectively collected since November 2020. Due to the important rise of cases in October 2020, data collection of SPSU SARS-CoV-2 for non-hospitalised children was stopped, and therefore, we are unable to include nonhospitalised children in the data analysis during the further second wave.

Conclusion
This study confirms that COVID-19 is mostly a mild disease in children and adolescents with low mortality. Fever, rash and comorbidities are associated with higher admission rates. The clinical spectrum and severity are influenced by age in paediatric COVID-19. Continuous observation is necessary to further understand paediatric COVID-19, guide therapy and evaluate the necessity for vaccination in children.