COVID-19 trends and severity among symptomatic children aged 0–17 years in 10 European Union countries, 3 August 2020 to 3 October 2021

We estimated risks of severe outcomes in 820,404 symptomatic paediatric COVID-19 cases reported by 10 European Union countries between August 2020 and October 2021. Case and hospitalisation rates rose as transmission increased but severe outcomes were rare: 9,611 (1.2%) were hospitalised, 640 (0.08%) required intensive care and 84 (0.01%) died. Despite increased individual risk (adjusted odds ratio hospitalisation: 7.3; 95% confidence interval: 3.3–16.2; intensive care: 8.7; 6.2–12.3) in cases with comorbidities, most (83.7%) hospitalised children had no comorbidity.

Understanding the burden of coronavirus disease (COVID-19) among children is essential for evidencebased decision-making regarding the vaccination of children and for assessing the importance of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) mitigation measures in specific settings, such as schools [1]. Here, we report on the burden and severity of symptomatic notified COVID-19 cases among children in the European Union (EU).

Understanding the burden of COVID-19 in children in Europe
We analysed, using R v4.1.1 (R Core Team, Vienna, Austria [2]), pooled case-based surveillance data reported to The European Surveillance System (TESSy) by 10 EU countries (Austria, Cyprus, Finland, Germany, Ireland, Italy, Luxembourg, Malta, Slovakia and Sweden) for symptomatic COVID-19 cases (reported as symptomatic or with a date of onset), between weeks 32/2020 and 43/2021. We restricted the analysis to weeks 32/2020 to 39/2021 to account for delayed reporting of the outcomes: hospitalisation, intensive care unit (ICU) (admission to ICU and/or requiring ventilation or extracorporeal membrane oxygenation (ECMO)) or death. We compared the cumulative number of reported deaths by each country to official data from public sources [3], with a minimum completeness threshold of 90% for inclusion in this study. Time series of reported hospital and ICU admissions were compared with those for admission or occupancy obtained from public sources [4], excluding countries reporting incomplete time series or with peaks that occurred at different times. Outcomes reported as 'unknown' were recoded to 'no' for deaths in all countries, and for hospitalisation/ICU in Ireland (following national practice) and Sweden (all cases were coded as 'yes' or 'unknown'). Hospitalisation/ICU status was recoded to 'no' if date of hospitalisation preceded date of onset, to minimise inclusion of incidental hospital admissions in the crude risk numerators. Cases with unknown outcome or sex were excluded.
We described trends by age group in weekly rates and proportions of case notifications and hospital admissions. Age-specific cumulative case notification rates per 100,000 population and crude risk (percentage of cases in each age group reporting the outcome) for hospitalisation, ICU admission (among all and hospitalised cases) and death were estimated by age group in years and in months < 2 years using data from seven countries that reported age in months (Austria, Cyprus, Finland, Ireland, Luxembourg, Slovakia and Sweden). We considered the overall study period and the period when the SARS-CoV-2 Delta variant of concern (Phylogenetic Assignment of Named Global Outbreak (Pango) lineage designation B.1.617.2) accounted for more than 95% of reported weekly sequences (median: weeks 28/2021 to 39/2021). Age-specific distributions of the outcomes by period and sex were compared using chi-squared tests.
Cases with unknown comorbidity were excluded from a subset of data from seven countries (Cyprus, Finland, Italy, Luxembourg, Malta, Slovakia and Sweden) reporting data on comorbidities (coded as cancer, diabetes, cardiac disease, lung disease, neuromuscular disease, HIV infection, asthma, kidney disease, hypertension, pregnancy, liver disease, obesity, smoker/history of smoking). We estimated crude risks stratified by the presence and absence of a comorbidity and adjusted odds ratios (aOR; via logistic regression) to compare outcomes for cases in each age group (< 1, 1-4, 5-11, 12-17, 0-17 years) with and without any comorbidity.

Trends in case notification and hospitalisation
Pooled weekly notification rates increased sharply in all age groups from July 2021 ( Figure 1A). We observed concomitant rises in hospitalisation rates in all age groups, but starting from, and reaching, much lower levels in children aged 1-17 years than in adults or children younger than 1 year ( Figure 1C). Since January 2021, children have represented an increasing proportion of notified cases and hospital admissions ( Figure  1B and D).
Cumulative case rates reported among a total of 820,404 symptomatic paediatric cases (12.4% of 6,604,483 cases of all ages) increased with every additional year of age from 2 to 17 years (Table 1). Hospitalisation was reported for 9,611 (1.2%) cases,

Figure 1
Weekly age distribution and rates of notified symptomatic COVID-19 cases and hospital admissions reported to TESSy, pooled from 10 EU countries, weeks 32/2020 to 39/2021 (n = 6,604,483)

Table 1b
Age-specific counts and crude risks of severe COVID-19 outcomes by age, pooled from 10 EU countries, weeks 32/2020 to 39/2021 (n = 6,604,483) ICU admission for 640 (0.08% of all cases, 6.7% of hospitalised cases) and death for 84 (0.01%). Limiting the analysis to children aged 2-17 years, the overall risks of these outcomes were 0.8%, 0.06%, 7.5% and 0.01%, respectively. The risk of hospitalisation was highest among the youngest (0-2 months), with point estimates that decreased with increasing age to 9 years and then increased with each year from 12 to 17 years. Male children aged 0-17 years were slightly more likely than female children to be admitted to hospital (1.2% vs 1.1%; p < 0.0001) or ICU (0.09% vs 0.07%; p < 0.05), but these differences did not exist in all paediatric age groups. Results for periods coinciding with dominance of the Delta variant were consistent with the full study period, although hospitalisation was more common among children younger than 1 year (14.7% Delta vs 13.1% full period; p < 0.01).  dominant)), the adjusted odds of hospitalisation, ICU admission and death were seven, nine and 27 times higher, respectively, among cases with at least one comorbidity compared with those with none ( Table 2).

Discussion
This study highlights that the risk of a severe COVID-19 outcome is substantially elevated for children with underlying risk factors compared with healthy children [8]. However, among paediatric COVID-19 cases with information on comorbidities, 83.7% had no reported comorbidity, demonstrating a potential populationlevel impact of high levels of community transmission leading to large numbers of hospital admissions among healthy children.
The elevated risk of hospitalisation observed for children younger than 2 years may reflect lower thresholds for admission of infants and neonates in particular [9], and further research is required to clarify the risk of severe disease in this age group. Whether the higher crude risk of hospitalisation among children under 1 year in the period of dominance of the SARS-CoV-2 There are important limitations to this study. The analysis is based on surveillance data reported to TESSy with at least 90% completeness of cases and deaths compared with official national totals. Data on vaccination status of the children in this study was not available, however, during the study period, vaccines were only approved in the EU for use in children 12-17 years-old. As children are less likely to be symptomatic for COVID-19 than adults [11,12], reporting of cases may be biased towards those with severe disease. Although we recoded 640 cases with hospitalisation date before date of onset as not hospitalised and the TESSy reporting protocol defined a hospitalised case as one with severe COVID-19 requiring admission to hospital or ICU [13], it is possible that some remaining reported paediatric COVID-19 hospitalisations may have been for other causes. While this could overestimate the crude risk of hospitalisation, we are aware that hospitalisations were under-reported to TESSy in three of the countries in our study (50-76% complete for Germany, Ireland and Sweden) for which comparison against publicly available admission data was possible, which would have the opposite effect. Overall, our results are comparable to those reported among symptomatic cases in a recent systematic review, including studies from Europe and the US [10]. Reporting of comorbidities was less likely among cases with severe outcomes, making our effect estimates conservative for cases with a comorbidity. Low numbers of children with severe outcomes diminish the analytical power of the study, particularly for less common outcomes. This, together with incomplete reporting of comorbidities, prevented a detailed risk factor analysis.

Conclusions
Paediatric hospital admissions for COVID-19 increased as overall transmission rates increased. The individual risks of a severe COVID-19 outcome were substantially elevated for those with a comorbidity compared with healthy children, but most children hospitalised in this study with data on comorbidities had no reported comorbidity. This demonstrates the additive impact of high levels of community transmission and can inform decision making around paediatric COVID-19 vaccinations. Preventive measures to reduce transmission and severe outcomes in children remain critical, as does the submission of timely, complete surveillance data to facilitate assessment of severity following the emergence of new variants.