The systematic review and meta-analysis were performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. It was registered with the International Prospective Register of Systematic Reviews (PROSPERO) (Registration Number: CRD42022315126). PubMed, Embase and the Latin American databases LILACS, BRISA/RedTESA, IBECS, LIPECS, Sec. Est. Saúde SP, Scielo were searched from inception to March 6th, 2022 for published studies describing subjects ≤ 19 years old who developed pericarditis or myocarditis following COVID-19 vaccination. The search terms for PubMed were: ((pericarditi s [Title/Abstract]) OR (myocarditis [Title/Abstract]) OR (cardiac [Title/Abstract])) AND (covid19 [Title/Abstract]) AND (vaccin*[Title/Abstract]) AND ((child*[Title/Abstract]) OR (adolescent*[Title/Abstract]) OR (paediatric [Title/Abstract]) OR (young [Title/Abstract])), and those for Embase were: (pericarditis:ab,ti OR myocarditis:ab,ti OR cardiac:ab,ti) AND (covid-19’:ab,ti AND vaccin*:ab,ti) AND (children:ab,ti OR adolescent:ab,ti OR paediatric:ab,ti OR young:ab,ti). The Latin American databases LILACS, BRISA/RedTESA, IBECS, LIPECS, Sec. Est. Saúde SP, Scielo were also searched using the following search terms: 1. (miocarditis) AND (vacuna) AND (COVID) OR (coronavirus); 2. (pericarditis) AND (vacuna) AND (COVID) AND (coronavirus) (Supplementary Table 1).

Search results for PubMed, Embase and the Latin American databases were independently screened by team members. Each entry was screened by two members separately and assessed for compliance with the inclusion criteria. The exclusion criteria during data screening are included in Supplementary Table 2. Any disagreements would be brought to the attention of an independent reviewer not involved in the initial screening (GT). Data from the different studies were entered in Microsoft Excel. The initial plan was to search the largest passive vaccine surveillance systems, EudraVigilance, Vaccine Adverse Event Reporting System (VAERS) and VigiBase databases for adverse events. However, both VAERS and VigiBase were analysed by the included studies, and therefore these databases were not searched ultimately to avoid the duplication of data. In this meta-analysis, the extracted data elements consisted of: (1) last name of the first author and year of publication; (2) study type, (3) country of study; (4) number of events and number of at-risk individuals, (5) follow-up duration of the study, (6) sex and age, (7) vaccine type, (8) number of doses received, (9) delay between vaccination and onset of carditis, (10) definition of carditis, (11) treatment received, (12) number of cases that recovered with or without treatment, (13) length of hospital stay by the myocarditis/pericarditis cases. Subgroup analysis based on the type of vaccine was performed. Quality evaluation of case reports/case series was assessed using the case-report (CARE) 13-item guideline [24, 25]. Case–control and cohort studies were evaluated using the Newcastle–Ottawa Quality Assessment Scale [26].

Summary statistics were used to summarise the number of events, age, sex, vaccine types and outcomes for all cases described by all studies. From population-based studies, the number of events and where available, at-risk populations were extracted. Rate ratios were stratified according to dose and gender and expressed as a rate per million doses received. The confidence interval of the rates was calculated using the Poisson Exact Method. The heterogeneity across the studies was determined using Cochran’s Q value and the I² statistics. The fixed-effect model was used to conduct the meta-analysis. However, the random-effects model was used to calculate the pooled rate across the studies when the heterogeneity was significant (I² > 50%). The rate ratios were calculated to compare the rate of carditis after COVID-19 vaccination and infection. The analysis was performed using RStudio (version 1.4.1103).

Twelve population-based studies were included in the study (Table 1). In total, 12,602,625 doses, including 8,050,898 first doses and 4,551,727 s doses, were administered after excluding overlapping cohorts. Overlapping cohorts were identified from analysing the methodology of the studies and defined as patient cohorts in the same country and/or hospitals within a similar time period. All of the subjects (49.95% male) were vaccinated with BNT162b2. Five population-based studies involving individuals receiving COVID-19 vaccines from the United States, Canada, Israel and Hong Kong, China were selected after excluding the overlapping cohorts or cohorts without providing information on the doses administered.

Of the pooled studies reporting on carditis, 343 patients developed carditis, corresponding to a rate per million doses per 7 days of 37.76 (95% confidence interval [CI] 23.57, 59.19) (Fig. 2). Further subgroup analyses by sex showed that the rate of carditis was higher among males (rate per million: 60.31; 95% CI 40.15, 90.61) than females (rate per million: 12.02; 95% CI 4.37, 33.07) with a rate ratio of 5.04 (1.40, 18.19) (Table 2). The rate of carditis was significantly higher among the second dose (rate per million: 70.82; 95% CI 49.47, 101.18) than the first dose (rate per million: 12.64; 95% CI 4.81, 33.24) with a rate ratio of 5.60 (95% CI 1.97, 15.89). From the studies providing myocarditis-only data, the rate per million doses of myocarditis was 22.34 (95% CI 8.46, 58.99). The rate per million doses of pericarditis was 5.72 (95% CI 3.34, 9.87).

The rate of carditis per 7 days in the background and after COVID-19 infection were carefully selected by matching the location of the selected studies. Overall, among paediatric patients, the rates of carditis after the COVID-19 vaccination were not significantly different from the COVID-19 infection (Table 3). The rate of carditis after the first dose of vaccination was lower than the COVID-19 infection. Meanwhile, the rate of carditis after the second dose was higher than the background and not different from the COVID-19 infection. For male patients, the rate of carditis was significantly higher than the background, similar to that after COVID-19 infection. For females, the rate of carditis was much lower than that after COVID-19 infection.

A total of 301 patients who developed carditis were included (Table 4). 281 cases were male with a mean age of 15.90 (SD 1.52) (Table 5). Of all doses vaccinated, 79.39% were BNT162b2 (Pfizer) vaccine, 19.91% were Moderna (mRNA-1273) vaccine, and < 0.23% were Ad26.COV2.S (Johnson & Johnson’s Janssen) vaccine (Fig. 3A). Some 240 (79.73%) had myocarditis, 5 (1.66%) had pericarditis and 56 (18.60%) had unclassified carditis. A total of 38 (12.62%) events occurred after the first dose and 263 (87.38%) events occurred after the second dose (Fig. 3B). Only 3 patients reported a previous history of cardiac diseases.

The most common presenting symptoms were: chest pain (97.34%), fever (37.54%), dyspnoea (21.26%), myalgia (16.94%) and headache (15.61%). Laboratory findings included elevated troponin (median: 9.62; interquartile range [IQR] 4.40, 828.09) and C-reactive protein levels (median: 7.58; IQR 4.06, 24.70) (Fig. 3C). The most common findings in the electrocardiogram included ST elevation and T wave abnormalities. Other findings such as ST depression, non-specific ST changes, interventricular conduction, PR depression, atrial tachycardia and left axis deviation were also reported. Four patients reported to have sinus tachycardia. Common CMR findings included oedema and late gadolinium enhancement in the myocardium. The echocardiogram abnormalities included decreased left ventricular ejection fraction and pericardial effusion.

Overall, 261 (86.71%) patients were reported to have received treatment. 53 (20.31%) patients required intensive care unit admission and the mean length of stay in the hospital was 3.91 days (SD 1.75) (Table 3). Among those receiving treatment, 226 (86.59%) received nonsteroidal anti-inflammatory agents (NSAID) and 50 (19.16%) received steroid treatment. Intravenous immunoglobulins were prescribed in 57 (21.84%) patients. Angiotensin-converting enzyme (ACE) inhibitors and colchicine were also used in 2 (0.77%) and 24 (9.20%) patients, respectively. Out of the 299 cases with longitudinal description, 298 (99.67%) patients had resolution of the myocarditis and/or pericarditis with or without treatment and none of the patients died as a result of the myocarditis and/or pericarditis.

The landmark trials of COVID-19 vaccinations demonstrated favourable safety profiles with complications being rarely reported, with no cases of post-vaccine myocarditis or pericarditis being observed [27, 28]. This could have been due to the small number of patients enrolled in the clinical trials, as well as the apparent rarity of this complication [29]. However, some concerns were raised following the reports by the CDC and VAERS of rare cases of myocarditis and pericarditis being associated with COVID-19 mRNA vaccinations, specifically the Pfizer-BioNTech mRNA vaccine (BNT162b2) and the Moderna mRNA vaccine (mRNA-1273) [14, 30‐35]. Since the COVID-19 vaccine was only approved for adults at the beginning, the majority of the reported side effects were studied in adults, with little attention paid to children and adolescents. Nonetheless, it was reported that the risk of myocarditis and pericarditis following COVID-19 vaccination was the highest among males and young adults [36‐38]. Our review further extends these observations, by demonstrating an increased risk among those aged 12–17 years and of the male sex.

Our findings suggest that the risk of post-vaccine carditis is highest among males aged 12–17 years old [14]. This is likely related to the age and sex distribution of non-vaccine-associated myocarditis in the general population, demonstrating a bimodal distribution of incidence peaking at infancy and adolescence more common in males [39‐41]. We also noted that the risk of post-vaccine carditis is higher in Asian populations than in Western populations, with Chua et al. and Li et al. reporting notably higher rates of post-vaccine carditis than in other studies [42, 43]. This is possibly due to the increased monitoring and reporting of cases in such areas, leading to increased detection of mild cases of carditis [44]. However, this may also relate to the general epidemiology of myocarditis, with the Global Burden of Disease study reporting the highest incidence of myocarditis in 2017 in East Asia and South Asia [23].

Although vaccine-related carditis has attracted widespread attention in the media, it is a rare complication. Its severity is usually mild to moderate with an early clinical diagnosis following (CMR) [45]. It was previously proposed that post-vaccination carditis can be due to an immune-mediated adverse response induced by the vaccination [46]. Alternatively, it was also suggested that the vaccination may induce heart-reactive autoantibodies, thus resulting in carditis [47].

It is important to note that while vaccine-related myocarditis has been receiving increased attention, previous case reports have suggested similar cardiovascular complications following COVID-19 infections [48]. A study by Puntmann et al. explored CMR outcomes in patients after COVID-19 infections and found that after two months of SARS-CoV-2 positivity, 78% of survivors had persistent cardiac involvement, with 60% showing ongoing signs of myocarditis on CMR [49]. In our study, the mean length of hospital stay was 3.9 days and 99.67% of patients studied achieved resolution of the myocarditis and/or pericarditis. These findings suggest that in contrast to infection-related myocarditis, the majority of vaccine-related myocarditis cases showed good and rapid recovery, with normalisation of ECG findings, left ventricular ejection fracture and cardiac markers often within a week [50, 51]. However, follow-up CMR was rarely performed for patients post-recovery and the long-term cardiovascular consequences of vaccine-related myocarditis remain unknown.

Nonetheless, our findings should raise the awareness of clinicians regarding the risk of developing carditis, which should be considered in individuals who present with chest pain usually around a week of vaccination, particularly within the younger male demographics. This bears increasing clinical importance as an increasing number of countries are recommending COVID-19 vaccinations for children and adolescents. Since the majority of carditis cases presented after the second dose of the COVID-19 vaccine, future research is needed regarding the safety and efficacy of additional doses in different age groups. A recent study from Israel demonstrated a 90% reduction in COVID-19-associated mortality in those who received a booster third dose of BNT162b2 at least 5 months after a second dose [52]. However, the incidence of post-vaccine carditis after an additional booster mRNA vaccination dose remains unknown and will require further investigation.

Given the global prevalence of the COVID-19 infection, in addition to the potentially long-lasting and occasionally life-threatening complications aside from carditis, it is critical to emphasize the potential benefits of the COVID-19 vaccination. Based on current evidence from this study and other published data, it largely outweighs the small, potential risk of carditis that is usually subclinical.

Several limitations of this study should be noted. The definitions of carditis differed between the studies, which can result in a discrepancy in the standard used for case collection. However, this is an inevitable limitation due to the difference in study design and healthcare resources across different countries. The calculation of the rate ratios was different depending on the study due to different age group stratification. Moreover, causal relationships cannot be established given the inclusion of observational studies. Additionally, it should be noted that the minimum age of vaccination varies between vaccines and countries, which can affect the interpretation of event risk amongst the younger age group. Furthermore, while the rate calculated was similar to Oster et al., the data from Krug et al. was derived from VAERS, which is a passive monitoring system, and thus, could be subjected to potential reporting and recall bias [14, 53, 54]. Finally, the minimum age of the paediatric patients in our included studies was 12 years old. There is still insufficient data regarding the carditis risk among patients < 12 years old and further studies are needed.