Safety outcomes following COVID-19 vaccination and infection in 5.1 million children in England

The UK began offering COVID-19 vaccines to children aged 12 years and older in September 2021, extending to 5–11-year-olds in April 2022 primarily as a one-off primary course for non–high-risk children. Uptake in children lagged behind adults. Most vaccinated children received BNT162b2, with ChAdOx1 generally not recommended below age 40 and mRNA-1273 not recommended below age 16 during most of the rollout. Doses were 30 µg for those aged ≥12 and 10 µg for 5–11-year-olds. Given ongoing parental hesitancy and evolving JCVI guidance limiting routine boosters in lower-risk younger populations, clarifying the comparative safety of vaccination versus infection in children is important. This study aimed to quantify the risks of pre-specified serious outcomes leading to hospitalisation following COVID-19 vaccination and SARS-CoV-2 infection in 5–17-year-olds, overall and by age group and sex.

Prior work has identified increased risks of myocarditis, especially in young males, after mRNA vaccination and elevated risks of several complications following SARS-CoV-2 infection. Regulatory surveillance and observational studies have evaluated outcomes such as myocarditis, pericarditis, MIS-C, thrombocytopenia, neuroimmunological events, and anaphylaxis. Clinical trials in adolescents demonstrated acceptable safety for BNT162b2 and mRNA-1273. However, evidence specific to younger children and direct comparisons of risks from vaccination versus infection in large population-based cohorts of children and adolescents remain limited. This study builds on SCCS analyses in adults and mixed-age cohorts to provide paediatric-specific estimates using England-wide linked data.

Design: Primary analyses used a self-controlled case series (SCCS) design to assess associations between COVID-19 vaccines (BNT162b2, mRNA-1273, ChAdOx1) and pre-specified outcomes resulting in hospitalisation or death, and between SARS-CoV-2 infection and these outcomes. Each outcome was analysed separately. The SCCS inherently controls for fixed individual characteristics (e.g., sex, ethnicity). Age was treated as fixed given the short study period. A secondary matched cohort analysis compared vaccinated children to age- and sex-matched unvaccinated controls. Data sources: Nationwide linked datasets included the NHS COVID-19 vaccination records, Hospital Episode Statistics for admissions, Office for National Statistics mortality data, and RT-PCR testing records (SGSS). Additional robustness checks used the QResearch primary care database. Population and period: All children aged 5–17 years in England who received at least one relevant vaccine dose or had a positive SARS-CoV-2 test between 8 December 2020 and 10 August 2022 were eligible. A comparative analysis was conducted in young adults aged 18–24 years. For each SCCS outcome analysis, included individuals had a first hospital admission or death for that outcome during the study period; those with a hospitalisation for the same outcome in the two years before 8 December 2020 were excluded. Exposures and risk windows: Exposures were first, second, and third doses of BNT162b2, mRNA-1273, or ChAdOx1, and SARS-CoV-2 infection defined by a positive RT-PCR test. The primary risk window was 1–42 days post-exposure. Day 0 (test date) was treated as its own period for infection. To mitigate bias from delayed vaccination after illness, a pre-risk period of 1–28 days before each exposure was excluded from baseline. If two vaccine doses were administered within two days, the prior dose’s risk period was truncated. Outcomes: Pre-specified outcomes were selected based on regulatory signals and literature: myocarditis, MIS-C, myositis, epilepsy, acute pancreatitis, acute disseminated encephalomyelitis (ADEM), angioedema, demyelinating disease, appendicitis, immune/idiopathic thrombocytopenia (ITP), Guillain–Barré syndrome (GBS), and anaphylaxis (positive control). Outcomes were identified from ICD-10 codes; only the first event per person in the study period was analysed. Statistical analysis: Conditional Poisson regression with an offset for exposure risk period length estimated incidence rate ratios (IRRs) comparing the 1–42 day risk window to baseline (all other observation time, excluding pre-risk). The observation period was split into two-week intervals to adjust for seasonality. Follow-up was censored at death, fourth dose, or study end. Excess events per million exposures were estimated from IRRs where at least five events occurred in exposure windows. Primary analyses were stratified by age group (5–11, 12–17) and sex; 18–24-year-olds provided comparison estimates. Robustness checks assessed pre-vaccination periods and the anaphylaxis positive control. A matched cohort analysis estimated incidence rate ratios post-vaccination versus unvaccinated comparators with adjustment for ethnicity, deprivation quintile, and comorbidity.

• Cohort: 5.1 million children (1,842,159 aged 5–11; 3,355,766 aged 12–17). Vaccination uptake: 32% of 5–11-year-olds received ≥1 dose (mostly BNT162b2); among 12–17-year-olds, 86% received dose 1, 67% dose 2, and 14% dose 3.
• 5–11 years: No increased risk detected for any pre-specified outcome within 1–42 days after BNT162b2, mRNA-1273, or ChAdOx1.
• 12–17 years (BNT162b2): Increased myocarditis risk confined to 1–14 days post-dose. Estimated excess events per million: approximately 3 (95% CI 0–5) after the first dose and 5 (95% CI 3–6) after the second dose. No deaths from myocarditis were observed in under-18s.
• 12–17 years (BNT162b2): Modest increase in hospitalisation with epilepsy after the second dose; estimated excess ≈12 (95% CI 0–23) per million exposed. Sex-stratified analyses indicated increased myocarditis predominantly in males after the second dose and a small increase in demyelinating disease in females after the second dose (≈4 excess per million; 95% CI 0–6).
• ChAdOx1 in adolescents: Increased risk of hospitalisation with epilepsy after the first dose (IRR ~1.9; estimated excess ≈700–800 per million exposed, particularly in females). Evidence for appendicitis risk was inconsistent and not supported in matched cohort analyses; BNT162b2 second-dose appendicitis was not significantly increased versus baseline.
• Infection risks (unvaccinated): SARS-CoV-2 infection was associated with increased risks of MIS-C, myocarditis, acute pancreatitis, ADEM, and ITP, with higher MIS-C risk in males; elevated risks were most evident 22–44 days after infection in children.
• Protection by vaccination: These infection-associated risks were largely absent or reduced among children and adolescents who had received at least one vaccine dose prior to infection.
• Matched cohort: Findings were broadly consistent with SCCS. Incidence rates for outcomes were significantly higher after SARS-CoV-2 infection than after vaccination in both age groups.

This large nationwide analysis provides age- and sex-specific safety estimates for children and adolescents. Vaccination with BNT162b2 showed no safety signals in 5–11-year-olds and only small increases in myocarditis and a modest increase in epilepsy-related hospitalisations in 12–17-year-olds, with myocarditis risk concentrated in the first two weeks post-dose and no associated deaths in under-18s. A small signal for demyelinating disease was observed in females after a second BNT162b2 dose. By contrast, SARS-CoV-2 infection in unvaccinated children and adolescents was linked to substantially higher risks of serious outcomes including MIS-C and myocarditis. Importantly, these infection-related risks were markedly reduced in those vaccinated prior to infection, supporting a net safety benefit of vaccination. Signals around ChAdOx1 in adolescents (epilepsy, appendicitis) should be interpreted cautiously given small exposed numbers and lack of corroboration in matched analyses. Overall, the results inform policymakers that the safety profile of mRNA vaccination in 5–17-year-olds is favourable compared with the risks posed by infection.

Among 5–11-year-olds, no increased risks for 12 pre-specified serious outcomes were detected following COVID-19 vaccination. In adolescents, BNT162b2 was associated with small increases in myocarditis and a modest increase in epilepsy-related hospitalisations, while vaccination substantially mitigated the elevated risks of MIS-C, myocarditis, and other outcomes associated with SARS-CoV-2 infection. These findings support the favourable safety profile of mRNA COVID-19 vaccination in 5–17-year-olds. Future research should refine risk estimates in younger children, assess variant-specific effects, evaluate equity and deprivation gradients, and continue active surveillance as vaccination strategies evolve.

• Potential underpower in 5–11-year-olds due to lower vaccine uptake in this age group.
• Outcome identification relied on hospital admission and death coding; events managed solely in the community or recorded only in primary care may be missed (possible misclassification).
• Infection ascertainment depended on RT-PCR testing; incomplete community testing and unrecorded infections could bias associations and misattribute risks.
• Lack of data on SARS-CoV-2 variants precluded variant-specific risk assessment; wave effects were not explicitly modelled beyond seasonality adjustments.
• Residual confounding by time-varying factors is possible despite SCCS design and seasonality adjustment.
• Limited numbers for certain vaccines/doses in adolescents (e.g., mRNA-1273, ChAdOx1) constrained precision and interpretation, especially in sex-stratified analyses.