TYPE Original Research PUBLISHED 24 November 2022 DOI 10.3389/fphar.2022.1038043 Myocarditis and pericarditis OPEN ACCESS associated with EDITED BY Christopher Blanchette, University of North Carolina at SARS-CoV-2 vaccines: A Charlotte, United States REVIEWED BY population-based descriptive Antonina Argo, University of Palermo, Italy Patrick C. Hardigan, cohort and a nested Nova Southeastern University, United States self-controlled risk interval study *CORRESPONDENCE Miriam C. J. M. Sturkenboom, m.c.j.sturkenboom@umcutrecht.nl using electronic health care data † These authors have contributed equally to this work and share first authorship from four European countries SPECIALTY SECTION This article was submitted to Sophie H. Bots 1†, Judit Riera-Arnau 2,3†, Svetlana V. Belitser 1, Pharmacoepidemiology, a section of the journal Davide Messina 4, Maria Aragón 5, Ema Alsina 2,3, Ian J. Douglas 6, Frontiers in Pharmacology Carlos E. Durán 2,3, Patricia García-Poza 7, Rosa Gini 4, RECEIVED 06 September 2022 Ron M. C. Herings 8, Consuelo Huerta 7, ACCEPTED 31 October 2022 PUBLISHED 24 November 2022 Malede Mequanent Sisay 2,3, Mar Martín-Pérez 7, CITATION Ivonne Martin 2,3, Jetty A. Overbeek 8, Olga Paoletti 4, Bots SH, Riera-Arnau J, Belitser SV, Messina D, Aragón M, Alsina E, Meritxell Pallejà-Millán 9, Anna Schultze 6, Patrick Souverein 1, Douglas IJ, Durán CE, García-Poza P, Karin M. A. Swart 8, Felipe Villalobos 5, Olaf H. Klungel 1 and Gini R, Herings RMC, Huerta C, Sisay MM, Martín-Pérez M, Martin I, Miriam C. J. M. Sturkenboom 2,3* Overbeek JA, Paoletti O, 1 Pallejà-Millán M, Schultze A, Division of Pharmacoepidemiology and Clinical Pharmacology, Utrecht Institute for Pharmaceutical Souverein P, Swart KMA, Villalobos F, Sciences, Utrecht University, Utrecht, Netherlands, 2Department of Datascience and Biostatistics, Klungel OH and Sturkenboom MCJM Julius Center for Health Sciences and Primary Health, University Medical Center Utrecht, Utrecht, (2022), Myocarditis and pericarditis Netherlands, 3Clinical Pharmacology Service, Vall d’Hebron Hospital Universitari, Vall d’Hebron associated with SARS-CoV-2 vaccines: Barcelona Hospital Campus, Universitat Autònoma de Barcelona, Barcelona, Spain, 4Agenzia A population-based descriptive cohort Regionale di Sanitá, Florence, Toscana, Italy, 5Fundació Institut Universitari per a la recerca a l’Atenció and a nested self-controlled risk interval Primària de Salut Jordi Gol i Gurina (IDIAPJGol), Barcelona, Spain, 6Faculty of Epidemiology and study using electronic health care data Population Health, London School of Hygiene & Tropical Medicine, London, United Kingdom, 7Spanish from four European countries. Agency for Medicines and Medical Devices (AEMPS), Madrid, Spain, 8PHARMO Institute for Drug Front. Pharmacol. 13:1038043. Outcomes Research, Utrecht, Netherlands, 9Unitat de Suport a la Recerca Tarragona-Reus, Fundació doi: 10.3389/fphar.2022.1038043 Institut Universitari per a la recerca a l’Atenció Primària de Salut Jordi Gol i Gurina (IDIAPJGol), Barcelona, Spain COPYRIGHT © 2022 Bots, Riera-Arnau, Belitser, Messina, Aragón, Alsina, Douglas, Durán, García-Poza, Gini, Herings, Huerta, Background: Estimates of the association between COVID-19 vaccines and Sisay, Martín-Pérez, Martin, Overbeek, Paoletti, Pallejà-Millán, Schultze, myo-/pericarditis risk vary widely across studies due to scarcity of events, Souverein, Swart, Villalobos, Klungel and especially in age- and sex-stratified analyses. Sturkenboom. This is an open-access article distributed under the terms of the Methods: Population-based cohort study with nested self-controlled risk Creative Commons Attribution License (CC BY). The use, distribution or interval (SCRI) using healthcare data from five European databases. reproduction in other forums is Individuals were followed from 01/01/2020 until end of data availability (31/ permitted, provided the original 12/2021 latest). Outcome was first myo-/pericarditis diagnosis. Exposures were author(s) and the copyright owner(s) are credited and that the original first and second dose of Pfizer, AstraZeneca, Moderna, and Janssen COVID-19 publication in this journal is cited, in vaccines. Baseline incidence rates (IRs), and vaccine- and dose-specific IRs and Frontiers in Pharmacology 01 frontiersin.org Bots et al. 10.3389/fphar.2022.1038043 rate differences were calculated from the cohort The SCRI calculated calendar time-adjusted IR ratios (IRR), using a 60-day pre-vaccination control period and dose-specific 28-day risk windows. IRRs were pooled using random effects meta-analysis. Findings: Over 35 million individuals (49·2% women, median age 39–49 years) were included, of which 57·4% received at least one COVID-19 vaccine dose. Baseline incidence of myocarditis was low. Myocarditis IRRs were elevated after vaccination in those aged < 30 years, after both Pfizer vaccine doses (IRR = 3·3, 95%CI 1·2-9.4; 7·8, 95%CI 2·6-23·5, respectively) and Moderna vaccine dose 2 (IRR = 6·1, 95%CI 1·1-33·5). An effect of AstraZeneca vaccine dose 2 could not be excluded (IRR = 2·42, 95%CI 0·96-6·07). Pericarditis was not associated with vaccination. Interpretation: mRNA-based COVID-19 vaccines and potentially AstraZeneca are associated with increased myocarditis risk in younger individuals, although absolute incidence remains low. More data on children (≤ 11 years) are needed. KEYWORDS myocarditis, pericarditis, COVID-19 vaccine, adverse drug reaction, pharmacovigilance Introduction This study reports myocarditis and pericarditis incidence rates (IRs) before and after vaccination with four EMA-approved In July 2021, the European Medicines Agency’s (EMA) COVID-19 vaccines and assesses the association between each Pharmacovigilance Risk Assessment Committee (PRAC) vaccine dose and myo- and pericarditis risk using electronic announced that mRNA-based COVID-19 vaccines may health record data from four European countries. elevate the risk of myo- and pericarditis (EMA press office, 2021) and this was added to the vaccine’s Summary of Product Characteristics (European Medicines Agency, 2021a; Methods European Medicines Agency, 2021b). By August 2022, approximately 1,2 billion COVID-19 vaccine doses had been Databases distributed within the European population, of which 76% were mRNA-platform based (European Centre for Disease Prevention We used data from five European data sources: the Dutch and Control, 2022). Current evidence of the association between PHARMO Database Network (NL-PHARMO) (Kuiper et al., COVID-19 vaccines and myocarditis comprised many case 2020), the Spanish Base de Datos para la Investigación reports and case series (Lee et al., 2022) and six observational Farmacoepidemiológica en Atención Primaria (ES-BIFAP) studies (Barda et al., 2021; Husby et al., 2021; Le Vu et al., 2021; database (Maciá-Martínez et al., 2020), the Spanish Sistema Patone et al., 2021; Witberg et al., 2021; Karlstad et al., 2022). d’Informació per el Desenvolupament de la Investigació en These suggest an elevated risk of myocarditis and potentially Atenció Primària (ES-SIDIAP), the Italian Agenzia Regionale pericarditis in the first 7–14 days after vaccination, especially in Di Sanità della Toscana (IT-ARS) database, and the British younger men after the second dose. Most of the cases resolved Clinical Practice Research Datalink (UK-CPRD) Aurum without problems (EMA press office, 2021; Klamer et al., 2022). (Clinical Practice Research Datalink, 2022). For more details Uncertainty remains around dose, age, sex, and platform see Table 1. effects as two observational studies do not provide data by vaccine dose (Barda et al., 2021; Husby et al., 2021) and three focussed on mRNA-based COVID-19 vaccines only (Barda et al., Study populations and design 2021; Husby et al., 2021; Le Vu et al., 2021). The largest study up to date was limited in their ability to assess effects of the This study combines a retrospective cohort and a nested self- adenovirus-based COVID-19 vaccines because they used data controlled risk interval (SCRI) study. For the cohort, all from Nordic countries where vaccine strategies relied almost registered individuals who had at least 365 days of data exclusively on mRNA-based vaccines (Karlstad et al., 2022). availability prior to 1 January 2020 were eligible for inclusion. Moreover, none of the studies were able to evaluate the effect Follow-up ended at the first myo-/pericarditis diagnosis, death, of the Janssen vaccine. or end of data availability, whichever came first. Person-time was Frontiers in Pharmacology 02 frontiersin.org Bots et al. 10.3389/fphar.2022.1038043 TABLE 1 Characteristics of the participating data sources. IT-ARS ES-BIFAP ES-BIFAP- ES-SIDIAP NL-PHARMO UK-CPRD HOSP Geographical Italy, Tuscany region Spain, multiple Spain, multiple Spain, Catalonia Netherlands United Kingdom location regions regions region National 100% 9% 9% 80% 45% 20% Population coverage End of data August 2021 October 2021 October 2021 June 2021 December 2021 December 2021 availability Myo- and Discharge diagnoses & emergency General practice & General practice General practice & General practice & General practice & pericarditis visits specialist & discharge discharge specialist specialist information diagnoses information diagnoses diagnoses information reported reported back reported back back Diagnosis coding ICD9CM SNOMED, ICD10CM IC10CM ICPC* SNOMED, RCCD2 system ICD9CM COVID-19 Immunisation register COVID-19 COVID-19 Clinical records GP records GP records vaccination vaccination register vaccination register COVID-19 COVID-19 PCR-RT register PCR-RT tests PCR-RT tests PCR-RT tests Diagnoses & PCR- GP records diagnoses RT tests Relevant linked In- and outpatient pharmacy Pharmacy Pharmacy Community information dispensing, diagnostic tests and dispensing dispensing pharmacy invoices, sources procedures, mortality register, specialist referrals mental health service register *This coding system does not differentiate between myocarditis and pericarditis, GP: general practitioner. categorised into non-vaccinated and vaccinated post dose 1 or Outcomes dose 2. Dose-specific follow-up was 28 days but was censored if administration of the next vaccine dose happened within this The main outcomes were a recorded diagnosis of myocarditis interval. or pericarditis, defined by using data source-specific code lists The SCRI design is a modification of a self-controlled case (Supplementary Table S2). Individuals experiencing both events series (SCCS), which uses a fixed control window set before the first contributed to both the myocarditis and pericarditis analysis. vaccine dose (Baker et al., 2015). For the SCRI, all individuals enrolled in the study cohort with at least one COVID-19 vaccine dose and a myo-/ pericarditis diagnosis during follow-up were eligible for inclusion. Covariates Patients with a heterologous second dose (0.81%) or unknown brand were excluded since the group was too small to look at We extracted information on age, sex, and COVID-19 separately and to avoid misclassification of brands. Data from ES- disease (diagnosis or PCR-RT test). We also extracted BIFAP-HOSP were excluded to avoid duplicating cases from ES- information on diagnoses of cardiovascular disease, cancer, BIFAP. Per design, only cases occurring between the start of the chronic lung disease, HIV, chronic kidney disease, diabetes control window and the end of the study period were included. The (type 1 and 2), severe obesity, sickle cell disease, and the use 60-day control window started 90 days before the first vaccine dose, and of immunosuppressants prior to cohort entry. Each condition 28-day risk windows were defined after each vaccine dose. If the second was defined based on a combination of diagnosis codes and proxy dose was administered within the first dose risk window, the second medications (Willame et al., 2021). The cohort used age groups 5- dose took precedence. To account for potential event-dependency of the 11, 12-17, 18-29, and ≥ 30 years. The SCRI used < exposure, a pre-exposure buffer window starting 30 days prior to the 30 and ≥30 years due to the scarcity of events in younger age first vaccine dose was included (Supplementary Figure S1). groups. Exposure Data management and quality control The exposures of interest were the first and second dose the All five data sources converted their local data to the Pfizer/BioNtech (Comirnaty), Moderna (Spikevax), AstraZeneca Conception common data model (Thurin et al., 2022) (Vaxzevria), and Janssen (COVID-19 Vaccine Janssen) vaccines. resulting in structurally harmonised local datasets that were Frontiers in Pharmacology 03 frontiersin.org Bots et al. 10.3389/fphar.2022.1038043 quality checked (Hoxhaj, 2022a; van den Bor, 2022; Hoxhaj, Governance and transparency 2022b). Analyses were performed locally, and the aggregated output was uploaded to a Digital Research Environment for The protocol (Sturkenboom, 2021), R scripts for data quality standardisation and meta-analysis. checks (Hoxhaj, 2022a; van den Bor, 2022; Hoxhaj, 2022b) and data transformation are publicly available (Gini, 2022). This study was exempt from ethics review due to its observational Statistical analyses nature and anonymized data use. This study was conducted rapidly on request of the European Medicines Agency. There was Descriptive characteristics are given as mean (standard no direct involvement of patients or members of the public in the deviation), median [interquartile range], or number design, analysis, or reporting of this study. Main study results (percentage). Incidence rates (IR) were calculated with the were shared and discussed with members of the PRAC and the number of cases as numerator and person-years as Medicines and Healthcare Products Regulatory Agency from the denominator. Confidence intervals (CI) were calculated using United Kingdom. exact methods (Ulm, 1990). For persons aged ≥ 30 years, age- standardised rate differences were calculated by subtracting the age-standardised background myo-pericarditis incidence from Results the age-standardised incidence after each vaccination instance using the dsr package with the Eurostat population as reference. Cohort–baseline incidence and Due to heterogeneity in vaccine programmes targeted at children, vaccination uptake incidence rate differences were calculated using the fmsb package (Nakazawa, 2022) and included only persons without COVID-19 The study cohort comprised 35,369,669 persons (49·2% diagnosis to avoid confounding. women) with a median age of 39–49 years across the four The SCRI analysis was adjusted for calendar time in 30-day countries. Approximately 25%–35% of the study population time periods. Analyses are presented for the whole had at least one risk factor for severe COVID-19 disease at population by outcome type and vaccine brand and dose, the start of the study (Table 2). as well as stratified by sex and age group. Conditional Across data sources, the 2020 background myocarditis Poisson regression was implemented using R code based incidence in persons without COVID-19 disease ranged from on the SCCS package (Dif, 2022) adapted to include multiple 0·5–2·9/100,000 person-years (PY) in children aged 5–11 years, doses and vaccine brands. Aggregated data were pooled via 1·2-9·9 in persons aged 12–17 years, 2·8–6·4 for persons aged random effects meta-analysis using the meta package 18–29 years, and 2·7–4·5 for individuals ≥ 30 years (Figure 1A). (Balduzzi et al., 2019). The assumptions underlying the For pericarditis, these values were 0·6–5·2, 2·8–12·6, 10·0–21·9, SCRI design were tested. We performed two sensitivity and 11·6–29·7, respectively (Figure 1B). analyses: 1) excluding those with COVID-19 disease In total, 20,484,323 individuals (57.9%) received at least one before the control window to check the effect of COVID- COVID-19 vaccine dose, of whom 77·6% received at least two 19 disease; 2) using weekly risk windows (1–7 days, doses (0·81% heterologous) (Table 3). The distribution of vaccine 8–14 days, 15–21 days, 22–28 days) to check for a brands varied across countries, with the Pfizer vaccine potential time-response effect. The 7-day window was accounting for two-thirds of first doses in Italy, Spain, and the chosen based on literature suggesting the majority of Netherlands, whereas the AstraZeneca vaccine comprised 50% of myocarditis events occur within a week after vaccination. first doses in the United Kingdom. Only 7·3% and 2·7% of Both sensitivity analyses were performed in the whole individuals received the Moderna and Janssen vaccines, population only. Supplementary material S1 provides the respectively. Distance between vaccine doses differed between calculation for the number of excess cases. countries, with median distances being longest in the United Kingdom and shortest in Spain (Table 3). Vaccines brands were channelled to specific population groups based Missing data on country, age, and comorbidities (Supplementary Table S3). Individuals with missing data on sex, date of birth, or date of database entry or exit (< 1%) were excluded (Supplementary Cohort–incidence rates after COVID-19 Table S1). Due to the clinical nature of the data, missing data on vaccination exposure or outcome was interpretated as informative missingness and thus coded as absence of exposure or In children 5–11 years we did not observe any myo- or outcome, respectively. Missing data on covariates were not pericarditis events after COVID-19 vaccination during follow- imputed as they were only used for descriptive tables. up (Figure 1; Supplementary Table S4A). Frontiers in Pharmacology 04 frontiersin.org Bots et al. 10.3389/fphar.2022.1038043 TABLE 2 Baseline characteristics of the study population for the cohort study at 1 January 2020. IT-ARS ES-BIFAP ES-BIFAP-HOSP ES-SIDIAP NL-PHARMO UK-CPRD Study population (n, %) 3,490,375 (9.9) 11,996,689 (33.9) 9,211,907* 5,283,944 (14.9) 2,302,018 (6.5) 12,296,643 (34.8) Time of follow-up (person- 5,670,102 19,040,824 9,660,986 7,770,414 3,621,169 21,686,454 years) Demographic characteristics Female person-years (%) 47.9 48.2 47.7 49.3 49.0 50.5 Age in years (median, IQR) 49 [29-66] 44 [26-60] 45 [27-61] 43 [25-59] 44 [23-61] 39 [21-57] Age categories (%) 0–4 3.3 4.0 3.9 4.2 4.4 5.5 5–11 6.1 7.1 7.0 7.2 7.2 8.7 12–17 5.3 6.1 6.0 6.3 6.8 6.9 18–24 6.1 6.4 6.3 6.8 8.2 8.3 25–29 4.5 5.0 5.0 5.4 6.0 6.8 30–39 10.3 13.2 13.2 13.4 12.0 14.5 40–49 14.9 17.1 17.2 17.4 13.0 13.4 50–59 16.1 14.8 14.9 14.2 15.2 13.6 60–69 12.9 11.1 11.3 10.8 13.2 10.0 70–79 11.5 8.4 8.6 8.3 9.7 7.7 80+ 9.1 6.8 6.7 5.9 4.3 4.6 Risk factors for severe COVID-19 at 1 January 2020 (%) Cardiovascular disease 27.8 18.5 18.3 19.1 19.4 17.5 Cancer 2.4 1.6 1.9 2.1 2.2 1.3 Chronic lung disease 5.6 4.9 5.0 6.3 6.2 7.3 HIV 0.3 0.0 0.0 0.0 0.1 0.0 Chronic kidney disease 0.5 0.5 0.5 0.9 0.6 0.2 Diabetes 5.6 5.8 5.6 6.4 4.8 5.0 Severe obesity 0.2 0.9 1.0 2.5 0.2 0.7 Sickle cell disease 0.1 0.1 0.1 0.1 0.0 0.0 Use of immunosuppressants 6.0 1.4 1.4 1.8 2.8 0.4 At least one risk factor 34.4 23.9 23.9 26.2 26.2 24.1 *ES-BIFAP-HOSP, is subpopulation part of ES-BIFAP, therefore the study population percentage has not been calculated. CI: confidence interval, IRR: incidence rate ratio. IRRs, are adjusted for calendar time using 30-day periods, IQR: interquartile range; HIV: human immunodeficiency virus. In adolescents aged 12–17 years, the rate difference between 539·9). Myocarditis IR was also increased after both AstraZeneca post-vaccination and background rates of myocarditis and vaccine doses in UK-CPRD, but not statistically significant. No pericarditis rates was elevated in all data sources after both events were identified after the Janssen vaccine (763 PY follow- Pfizer vaccine doses. Non-significant increases in myocarditis up) (Figure 1; Supplementary Table S4A). rates were seen after Moderna vaccine dose 1 in ES-BIFAP and In individuals aged ≥ 30 years, age-standardized pericarditis after dose 2 in UK-CPRD. Follow-up after the AstraZeneca IRs were significantly decreased after AstraZeneca vaccine dose 1 in (1,262 PY) and Janssen (62·5 PY) vaccines was limited in this IT-ARS (RD = −21·7, −28·3; −15·0), ES-BIFAP-HOSP (RD = −16·7, age group and no events were observed after vaccination 95%CI −25·1; −8·3) and ES-SIDIAP (RD = −2·8, 95%CI −4·7; −0·9). (Figure 1; Supplementary Table S4A). We also observed lower pericarditis rates after Pfizer vaccine dose 2 in In individuals aged 18–29 years, the myocarditis and ES-BIFAP (RD = −5·9, 95%CI −11·0; −0·9) and Moderna vaccine pericarditis rate differences were elevated after both Pfizer dose 1 in IT-ARS (RD = −22·2, 95%CI −31·7; −12·7). Myocarditis vaccine doses, but not significantly in single data sources. A rates were only slightly elevated, and with heterogeneity across data significant increase in the rate of myocarditis was seen after sources (Figure 1; Supplementary Table S4B). In PHARMO Moderna vaccine dose 2 in IT-ARS (RD = 284·7, 95%CI 29·6- standardised myopericarditis IRs were significantly decreased after Frontiers in Pharmacology 05 frontiersin.org Bots et al. 10.3389/fphar.2022.1038043 FIGURE 1 The figure shows crude incidences for the age categories below 30 years and age-standardised rates for the > = 30 years category. Frontiers in Pharmacology 06 frontiersin.org Bots et al. 10.3389/fphar.2022.1038043 TABLE 3 Vaccinated persons and distances in days between dose one and for all COVID-19 vaccines. Total* N = IT-ARS N = ES-BIFAP ES-BIFAP- ES-SIDIAP UK-CPRD NL-PHARMO 35,369,669 3,490,375 N= HOSP N = N= N= N = 2,302,018 11,996,689 9,211,907 5,283,944 12,296,643 Total population 20,484,323 (57.9) 2,451,105 (70.2) 6,301,092 (52.5) 574,782 (6.2)& 2,793,996 (52.9) 7,466,594 (60.7) 1,471,536 (63.9) vaccinated (n, %) AstraZeneca (Vaxzevria) Dose 1 (n, %) 5,849,250 (28.6) 333,457 (13.6) 1,025,725 (16.3) 76,841 (13.4) 555,077 (19.9) 3,785,156 (50.7) 149,835 (10.2) Homologous Dose 82.5 91.7 63.8 96.9 52.4 92.5 48.4 2 (%) Heterologous Dose 2.5 4.6 1.4 1.1 2.1 2.7 <0.1 2 (%) Not received Dose 15.0 3.7 34.8 2.0 45.5 4.8 51.2 2 (%) Time between dose – 84 [84-84] 79 [74-86] 82 [72-84] 84 [79-92] 77 [69-79] 77 [76-84] 1 and dose 2 (median, IQR in days) Moderna (Spikevax) Dose 1 (n, %) 1,488,801 (7.3) 344,899 (14.1) 600,100 (9.5) 76,254 (13.3) 234,851 (8.4) 200,407 (2.7) 108,544 (7.4) Homologous Dose 77.6 66.1 86.9 91.0 74.0 84.9 57.4 2 (%) Heterologous Dose 0.2 0.1 0.1 0.1 < 0.1 0.6 <0.1 2 (%) Not received Dose 22.2 33.8 13.0 9.9 26.0 14.5 42.6 2 (%) Time between dose – 42 [28-42] 28 [28-28] 28 [28-28] 28 [28-29] 63 [57-77] 35 [35-35] 1 and dose 2 (median, IQR in days) Pfizer (Comirnaty) Dose 1 (n, %) 12,388,919 (60.5) 16,99,855 (69.4) 4,352,441 (69.1) 391,433 (68.1) 1,875,374 (67.1) 34,80,150 (46.6) 981,099 (66.7) Homologous Dose 78.8 81.3 81.8 93.7 67.4 81.8 71.8 2 (%) Heterologous Dose 0.1 < 0.1 < 0.1 < 0.1 < 0.1 0.4 < 0.1 2 (%) Not received Dose 21.2 18.7 18.2 6.3 32.6 17.8 28.2 2 (%) Time between dose – 42 [21-42] 21 [21-22] 21 [21-21] 21 [21-22] 74 [59-78] 35 [35-36] 1 and dose 2 (median, IQR in days) Janssen (Janssen COVID-19 vaccine) Dose 1 (n, %) 563,984 (2.7) 72,894 (3.0) 322,588 (5.1) 30,254 (5.3) 128,694 (4.6) 626 (0.0) 39,182 (2.7) Heterologous Dose 0.4 < 0.1 0.3 0.8 < 0.1 75.2 1.6 2 (%) Not received Dose 99.6 100 99.7 99.2 100 24.8 98.4 2 (%) Unknown brand 193,369 (0.9) – 238 (< 0.1) – – 255 (< 0.1) 192,876 (13.1) *excluding ES-BIFAP-HOSP. &low percentage as most of the regions with hospital data had only data until end of 2020, prior to vaccination. IQR: interquartile range. Frontiers in Pharmacology 07 frontiersin.org Bots et al. 10.3389/fphar.2022.1038043 TABLE 4 Pooled calendar-time adjusted incidence rate ratios for myocarditis and pericarditis from the self-controlled risk interval analysis, stratified by vaccine brand, dose, age, and sex. Myocarditis Pericarditis COVID-19 Unexposed/ First dose Unexposed/ Second Unexposed/ First dose Unexposed/ Second Vaccinations exposed cases IRR exposed cases dose IRR exposed cases IRR exposed cases dose IRR (95% CI) (95% CI) (95% CI) (95% CI) Whole population Control reference reference reference reference Pfizer 51/28 1.72 (0.96–3.09) 46/44 3.18 (1.65–6.12) 391/126 0.95 (0.74–1.22) 323/118 0.90 (0.67–1.21) AstraZeneca 37/11 0.96 (0.42–2.17) 29/17 2.42 (0.96–6.07) 143/58 0.93 (0.50–1.73) 119/55 1.22 (0.57–2.61) Moderna 14/7 4.00 (0.54–29.5) 11/13 5.28 (1.68–16.6) 68/25 0.99 (0.48–2.04) 57/21 1.11 (0.32–3.83) Janssen < 5/< 5 1.59 (0.12–21.6) 20/< 5 0.74 (0.16–3.42) 12–29 years Control reference reference reference reference Pfizer 9/10 3.30 (1.16–9.37) 5/14 7.78 (2.58–23.5) 41/26 1.73 (0.98–3.05) 30/22 1.74 (0.98–3.08) AstraZeneca 7/<5 0.32 (0.04–2.63) 6/<5 0.72 (0.14–3.69) 9/<5 1.14 (0.27–4.81) 8/<5 1.55 (0.41–5.87) Moderna <5/<5 3.49 (0.59–20.5) <5/10 6.05 (1.09–33.5) 11/9 1.84 (0.74–4.58) 8/5 0.48 (0.05–4.18) Janssen <5/<5 ≥ 30 years Control reference reference reference reference Pfizer 42/18 0.90 (0.43–1.89) 41/30 1.34 (0.70–2.59) 350/100 0.75 (0.57–0.98) 293/96 0.67 (0.49–0.91) AstraZeneca 30/10 0.82 (0.35–1.91) 23/15 1.92 (0.79–4.68) 134/55 0.83 (0.46–1.50) 111/51 0.88 (0.51–1.52) Moderna 11/<5 0.68 (0.05–9.01) 9/< 5 0.26 (0.04–1.76) 57/16 0.72 (0.22–2.38) 49/26 0.81 (0.42–1.54) Janssen < 5/< 5 0.88 (0.07–11.7) 20/<5 0.63 (0.16–2.53) Women Control reference reference reference reference Pfizer 19/10 1.03 (0.43–2.48) 19/13 1.54 (0.67–3.58) 162/49 082 (0.59–1.15) 139/43 0.69 (0.47–1.00) AstraZeneca 15/<5 0.17 (0.02–1.34) 12/6 1.29 (0.45–3.68) 60/25 0.98 (0.44–2.19) 45/21 1.19 (0.51–2.77) Moderna <5/<5 <5/<5 30/6 0.33 (0.11–0.96) 28/8 0.83 (0.36–1.94) Janssen <5/<5 11/<5 Men Control reference reference reference reference Pfizer 32/18 1.90 (0.98–3.67) 27/31 2.78 (1.53–5.07) 229/77 1.02 (0.70–1.50) 184/75 1.06 (0.79–1.43) AstraZeneca 22/10 1.19 (0.52–2.74) 17/11 1.91 (0.81–4.51) 83/33 0.88 (0.58–1.35) 74/34 1.08 (0.69–1.71) Moderna 10/5 2.86 (0.63–13.1) 7/12 5.80 (1.62–20.7) 38/19 1.75 (0.96–3.19) 29/13 1.51 (0.48–4.75) Janssen <5/<5 1.24 (0.10–15.7) 9/<5 1.70 (0.33–8.63) CI: confidence interval, IRR: incidence rate ratio. IRRs, are adjusted for calendar time using 30-day periods. Moderna vaccine dose 1 (RD = −21·2, 95%CI −23·5; −18·9) and vaccine (57%). Twenty-five cases from NL-PHARMO were AstraZeneca vaccine dose 2 (RD = −21·2, 95%CI −23·5; −18·9) excluded because vaccine brand was unknown and (Supplementary Table S4C). three cases were censored between the end of the pre- exposure window and the start of the first risk window, resulting in a total of 2,675 myo- or pericarditis cases Self-controlled risk interval analyses included in the analyses. As the records from NL-PHARMO (n = 115) could not be separated into In total, 2,703 myo-/pericarditis cases were eligible for myo- and pericarditis, the outcome-specific analyses inclusion in the SCRI. UK-CPRD contributed most included 2,560 cases (492 myocarditis, 2,124 pericarditis, cases (47%), and the majority of cases received the Pfizer 56 both). Frontiers in Pharmacology 08 frontiersin.org Bots et al. 10.3389/fphar.2022.1038043 FIGURE 2 Meta-analysed incidence rate ratios of myo- and pericarditis following COVID-19 vaccination. Pooled myocarditis incidence rate ratios (IRRs) were increased for both doses of the Pfizer vaccine (IRR = 3·30, significantly increased after the 2 nd dose of the Pfizer 95%CI 1·16-9·37; IRR = 7·78, 2·58-23·5, respectively) and (IRR = 3·18, 95%CI 1·65-6·12) and Moderna (IRR = 5·28, Moderna vaccine dose 2 (IRR = 6·05, 95%CI 1·09-33·5). The 95%CI 1·68-16·6) vaccines in the whole population (Table 4). effect of the Pfizer vaccine was seen in both sexes although The same trend was observed for AstraZeneca vaccine dose the number of cases in women was very low. For the 2, albeit not statistically significant (IRR = 2·42, 95%CI 0·96- Moderna vaccine, the effect could not be estimated 6·07). Stratification by age showed that the Pfizer and separately in women due to lack of events (Supplementary Moderna vaccine effects were driven by individuals aged < Table S6). The excess number of cases per 1 million 30 years (Table 4). In this age group, myocarditis risk was vaccinated individuals was 32·3 (95%CI 10·6-87·0) for the Frontiers in Pharmacology 09 frontiersin.org Bots et al. 10.3389/fphar.2022.1038043 Pfizer vaccine and 33·7 (95%CI 3·8-153·0) for the Moderna 6·88, 95%CI 5·28–8·96) Pfizer vaccine dose (Le Vu et al., 2021). A vaccine. self-controlled case series using the English national There was no significant increase in myocarditis risk in those immunisation database also showed elevated myocarditis risk aged ≥ 30 years (Table 4). Stratification by both age and sex after both Pfizer vaccine doses after 1–7 and 1–28-day risk showed that the effect of the AstraZeneca vaccine on myocarditis windows, although the effect sizes were closer to the Danish was strongest in men aged ≥ 30 years (IRR = 8·69, 95%CI 2·54- than the French study (Patone et al., 2021). All five studies 29·7) (Supplementary Table S6). We did not find evidence for an situated in Europe found increased myocarditis risk after the increased pericarditis risk after vaccination in the whole Moderna vaccine with estimates ranging from approximately 4- population (Table 4). Stratification by age and sex showed an fold in the Danish study to almost 26-fold in the French (Barda isolated finding in men aged < 30 years after Pfizer vaccine dose 2 et al., 2021; Husby et al., 2021; Le Vu et al., 2021; Patone et al., (Supplementary Table S6). There was no significant association 2021; Karlstad et al., 2022). The literature also supports our between vaccination and myopericarditis in NL-PHARMO observation that myocarditis risk was most strongly increased in (Supplementary Table S5). those aged 12–29 years, although estimates varied strongly here Using weekly risk windows suggested myocarditis risk was as well. The length of the risk window may play a role, as our highest 1–7 days after vaccination with the Pfizer and Moderna estimates for the Moderna vaccine were much higher when we vaccines and 8–14 days after vaccination with the AstraZeneca looked only at the first 7 days after vaccination compared to the vaccine (Supplementary Table S7). Excluding individuals with 28-day risk interval. For the age-specific analyses, different prior COVID-19 did not markedly change our conclusions definitions of the age groups may also explain some of the (Supplementary Figure S2). There was some heterogeneity variation in the estimates. across data sources (Figure 2). We could not exclude an increase in myocarditis IR after AstraZeneca vaccine dose 2, although it is difficult to tease out age- and sex-specific effects because this vaccine was Discussion predominantly administered to older individuals except in the United Kingdom. The association between the AstraZeneca This study, including over 35 million individuals from four vaccine and myocarditis risk has only been observed in the European countries, shows that myocarditis rates are elevated in English study (Patone et al., 2021), which reported an increase individuals aged < 30 years after both doses of the Pfizer vaccine in risk only after the first dose for both a 7-day (IRR = 1·76, 95% and the second dose of the Moderna vaccine. We could not CI 1·29–2·42) and a 28-day (IRR = 1·29, 95%CI 1·05–1·58) risk exclude an effect of AstraZeneca vaccine dose 2 on myocarditis, window. We are the first to study the association between the but this was mostly based on United Kingdom data where Janssen vaccine and myocarditis, but we had limited power to AstraZeneca vaccine was used in younger persons. The detect an effect due to low uptake of this vaccine. Janssen vaccine was not associated with myocarditis incidence, COVID-19 vaccines are highly effective in preventing severe but we had insufficient power due to limited distribution of this COVID-19 disease and are therefore a pivotal means for vaccine. Importantly, the absolute incidence of myocarditis was preventing mortality and morbidity in this pandemic. low resulting in a low number of excess myocarditis cases per However, accurate quantification of the risk of rare side million vaccinated. Pericarditis risk was not significantly elevated effects is crucial to be able to accurately establish benefit-risk after any of the COVID-19 vaccines. profiles of each of the COVID-19 vaccines in specific age groups. Our results are in line with previous work, although estimates Younger people are generally at a lower risk of serious vary strongly between studies. An Israeli target trial using data consequences of COVID-19, and the benefit risk balance from a payer-provider healthcare organisation found an increase needs to always be weighed carefully (Bilotta et al., 2022). The myocarditis risk within 42 days after the Pfizer vaccine (RR 3·24, EMA is responsible for monitoring the benefit risk balance and in 95%CI 1·55–12·44) and no effect on pericarditis (RR 1·27, 95% their assessment of the mRNA vaccines has stated that the 0·68–2·31) (Barda et al., 2021). A Danish nationwide cohort balance remains positive (European Medicines Agency, 2021b). study found no statistically significant elevation of myocarditis The main strengths of this study are its size and geographical risk within 27 days after the Pfizer vaccine (HR 1·34, 95%CI coverage, including previously unpublished data from four 0·90–2·00) (Husby et al., 2021), but did not distinguish by dose. A European countries (Italy, Spain, Netherlands, and Denmark). cohort from four Nordic countries (including Denmark) did find It combines a cohort design to calculate absolute IRs and RDs an elevated (but smaller) risk within 28 days after the first (IRR = with a nested SCRI to adjust for unmeasured confounding. The 1·38, 95%CI 1·12–169) and second (IRR = 1·75, 95%CI use of the Conception CDM and pipeline enables streamlined 1·43–2·14) dose of the Pfizer vaccine (Karlstad et al., 2022). data quality checks and local individual-level analysis. Our results Similarly, a French nested case-control study using nationwide allow for direct comparison between brand within one country reimbursement data found elevated myocarditis risk in the 7 days and across countries within one brand, showing consistent after both the first (OR 1·66, 95%CI 1·16–2·37) and second (OR findings across data sources. Lastly, the validity of the SCRI Frontiers in Pharmacology 10 frontiersin.org Bots et al. 10.3389/fphar.2022.1038043 relies on three important assumptions: 1) event-independent COVID-19 vaccination outweighs the risk of myocarditis in exposure; 2) no event-dependent censoring of observation younger persons (EMA press office, 2021). time; and 3) events are recurrent and independent. We showed that the risk of myocarditis was similar in the control window and the 30-day pre-exposure window, suggesting Data availability statement assumption (1) holds true. The other two assumptions should also hold true because mortality following myo-/pericarditis is The data analyzed in this study is subject to the following low and both events are rare, evidenced by the low absolute licenses/restrictions: The local data used for this study are not incidence rates. There are some limitations to address as well. publicly available. The protocol and scripts for data quality First, there was a lag in data availability because data available for checks, data transformation, and analysis are publicly available researchers was updated only periodically. To ensure that all data on GitHub: https://github.com/VAC4EU/CVM/tree/Release_ sources were able to run required quality checks, we truncated myocarditis_manuscript. follow-up at December 2021. However, some data providers had limited resources to update their data instances. For example, the data instance for ES-SIDIAP was locked in June 2021. Whenever Author contributions new data become available, we may update our analyses and include additional data sources. Moreover, vaccine exposure was All authors critically reviewed the manuscript. SoB, JR-A, limited in children aged 5–11 years so we could not provide an CD, AS, and MS wrote the manuscript. SB, JR-A, AS, SvB, DM, estimate for this population with the current data instances. ID, RG, MMS, MP-M, OK, and MiS were involved in design, Future updates of data may give more insight. Second, we may execution, and interpretation of data analysis. DM, MA, PG-P, have underestimated the background rates of myo-/pericarditis RG, RH, CH, MM-P, JO OP, MP-M, PS, KS, and FV were due to reduced access to care during lockdown periods, although involved in data collection and curation, running of quality check previous work suggests this effect was minimal (Willame et al., and analysis scripts, and had full access to the raw data. OK and 2021). This does not impact the SCRI as much since the control Mi S conceived the study. window was close to the risk windows, minimising differences in baseline rate due to changes in national COVID-19 policy. Third, data sources relying on GP data will record information from Funding specialist and hospitals with some delay. We censored these data sources at December 2021 to minimise this, however we may still The project received support from the European Medicines have missed some events and the cohort study may therefore Agency (EMA/2018/23/PE). have underestimated incidence rates. Fourth, we had limited power to estimate the effects of the AstraZeneca vaccines in persons aged below 30 and the Janssen vaccine in general due to Acknowledgments very low use of those vaccines. Future data instances may solve the former, but the latter is difficult to remedy because vaccine The research leading to these results was conducted as part of programmes simply did not distribute high quantities of Janssen the activities of the EU PE&PV (Pharmacoepidemiology and in the countries included in our study. Pharmacovigilance) Research Network (led by Utrecht Myocarditis incidence is elevated in individuals aged < University) with collaboration from the Vaccine 30 years after both doses of the Pfizer vaccine and the second Monitoring Collaboration for Europe network (VAC4EU). dose of the Moderna vaccine. AstraZeneca vaccine dose 2 may Scientific work for this project was coordinated by the also have an effect. Additional data on young children University Medical Center Utrecht. The project has (5–11 years) and adenovirus platform vaccines in younger received support from the European Medicines Agency individuals (< 30 years) are needed. Comparison between under the Framework service contract nr EMA/2018/23/ published studies was complicated by large variation in the PE. This report expresses the opinion of the authors and definition of risk window length and age groups. Given that may not be understood or quoted as being made on behalf of COVID-19 disease also poses a risk of life-threatening or reflecting the position of the European Medicines Agency complications in children, it is paramount to endorse vaccine or one of its committees or working parties. benefit-risk assessment studies in this population. Future studies focusing on younger age groups should consider harmonising these definitions to facilitate meta-analysis efforts. The absolute Conflict of interest incidence of myocarditis remains low and the number of excess cases per million vaccinated are around 30–35 per million The authors declare that the research was conducted in vaccinated, supporting the EMA decision that the benefit of the absence of any commercial or financial Frontiers in Pharmacology 11 frontiersin.org Bots et al. 10.3389/fphar.2022.1038043 relationships that could be construed as a potential conflict evaluated in this article, or claim that may be made by its of interest. manufacturer, is not guaranteed or endorsed by the publisher. Publisher’s note Supplementary material All claims expressed in this article are solely those of the authors and do not necessarily represent those of their The Supplementary Material for this article can be found affiliated organizations, or those of the publisher, the online at: https://www.frontiersin.org/articles/10.3389/fphar. editors and the reviewers. Any product that may be 2022.1038043/full#supplementary-material References Baker, M. A., Lieu, T. A., Li, L., Hua, W., Qiang, Y., Kawai, A. T., et al. (2015). A Klamer, T. A., Linschoten, M., and Asselbergs, F. W. (2022). The benefit of vaccine study design selection Framework for the postlicensure rapid immunization vaccination against COVID-19 outweighs the potential risk of myocarditis and safety monitoring program. Am. J. Epidemiol. 181 (8), 608–618. doi:10.1093/aje/ pericarditis. Neth. Heart J. 30, 190–197. doi:10.1007/s12471-022-01677-9 kwu322 Kuiper, J. G., Bakker, M., Penning-Van Beest, F. J. A., and Herings, R. M. C. Balduzzi, S., Rücker, G., and Schwarzer, G. (2019). How to perform a meta- (2020). Existing data sources for clinical epidemiology: The PHARMO database analysis with R: A practical tutorial. Evid. Based. Ment. Health 22 (4), 153–160. network. Clin. Epidemiol. 12, 415–422. doi:10.2147/CLEP.S247575 doi:10.1136/ebmental-2019-300117 Le Vu, S., Bertrand, M., Jabagi, M-J., Botton, J., Drouin, J., Baricault, B., et al. Barda, N., Dagan, N., Ben-Shlomo, Y., Kepten, E., Waxman, J., Ohana, R., et al. (2021). Association entre les vaccins COVID-19 à ARN messager et la survenue de (2021). Safety of the BNT162b2 mRNA COVID-19 vaccine in a nationwide setting. myocardite et péricardite. Chez les Pers. 12 à 50 ans Fr. Etude à données Système N. Engl. J. Med. 385 (12), 1078–1090. doi:10.1056/NEJMoa2110475 Natl. Données Santé (SNDS) 18, 65. Bilotta, C., Perrone, G., Zerbo, S., and Argo, A. (2022). COVID-19 vaccination in Lee, A. S. Y., Balakrishnan, I. D. D., Khoo, C. Y., Ng, C. T., Loh, J. K. X., Chan, L. pediatric population: A necessity or obstruction to the protection of the right to L., et al. (2022). Myocarditis following COVID-19 vaccination: A Systematic review health? Biojuridical perspective. Front. Public Health 10, 874687. doi:10.3389/ (October 2020-October 2021). Heart Lung Circ. 31 (6), 757–765. doi:10.1016/j.hlc. fpubh.2022.874687 2022.02.002 Clinical Practice Research Datalink (2022). CPRD Aurum january 2022 (version Maciá-Martínez, M. A., Gil, M., Huerta, C., Martín-Merino, E., Álvarez, A., 2022.01.001) [data set]. Clinical Practice Research Datalink. Available at: https:// Bryant, V., et al. (2020). Base de Datos para la Investigación Farmacoepidemiológica cprd.com/cprd-aurum-source-file-january-2022. en Atención Primaria (BIFAP): A data resource for pharmacoepidemiology in Spain. Pharmacoepidemiol. Drug Saf. 29 (10), 1236–1245. doi:10.1002/pds.5006 Dif, D. Using R for self-controlled case series studies [Internet]. Self-controlled case series studies. [cited 2022 Jun 2]. Available from: https://sccs-studies.info/r. Nakazawa, M. fmsb: Functions for medical statistics book with some html. demographic data [Internet]. 2022 [cited 2022 Jun 2]. Available from: https:// cran.r-project.org/web/packages/fmsb/index.html. EMA press office (2021). Comirnaty and spikevax: Possible link to very rare cases of myocarditis and pericarditis. [cited 2021 Dec 15]. Available from: https://www. Patone, M., Mei, X. W., Handunnetthi, L., Dixon, S., Zaccardi, F., Shankar-Hari, ema.europa.eu/en/news/comirnaty-spikevax-possible-link-very-rare-cases- M., et al. (2021). Risks of myocarditis, pericarditis, and cardiac arrhythmias myocarditis-pericarditis. associated with COVID-19 vaccination or SARS-CoV-2 infection. Nat. Med. 28, 410–422. doi:10.1038/s41591-021-01630-0 European Centre for Disease Prevention and Control (2022). COVID-19 vaccine tracker. [cited 2022 Jun 2]. Available from: https://vaccinetracker.ecdc.europa.eu/ Sturkenboom, M. Rapid safety assessment of SARS-CoV-2. vaccines E. U. Memb. public/extensions/COVID-19/vaccine-tracker.html#uptake-tab. States using Electron. health care datasources Protoc. V2.2 [Internet]. Eur. Netw. Centres Pharmacoepidemiol. Pharmacovigil. 2021 [cited 2021 Dec 15 25, 199. European Medicines Agency. Comirnaty. 2021 [cited 2021 Dec 15]. Available Available from: https://www.encepp.eu/encepp/viewResource.htm?id=42637. from: https://www.ema.europa.eu/en/medicines/human/EPAR/ comirnaty#authorisation-details-section. Thurin, N. H., Pajouheshnia, R., Roberto, G., Dodd, C., Hyeraci, G., Bartolini, C., et al. (2022). From inception to conception: Genesis of a network to support better European Medicines Agency (2021). Spikevax (previously COVID-19 vaccine monitoring and communication of medication safety during pregnancy and Moderna). [cited 2021 Dec 15]. Available from: https://www.ema.europa.eu/en/ breastfeeding. Clin. Pharmacol. Ther. 111 (1), 321–331. https://ascpt. medicines/human/EPAR/spikevax. onlinelibrary.wiley.com/doi/abs/10.1002/cpt.2476. Gini, R. VAC4EU/CVM [Internet]. GitHub. [cited 2022 Jun 2]. Available from: Ulm, K. (1990). A simple method to calculate the confidence interval of a https://github.com/VAC4EU/CVM/tree/Release_myocarditis_manuscript. standardized mortality ratio (SMR). Am. J. Epidemiol. 131 (2), 373–375. doi:10. Hoxhaj, V. IMI-ConcePTION/level-1-checks [Internet]. GitHub. [cited 2022 Jun 1093/oxfordjournals.aje.a115507 2]. Available from: https://github.com/IMI-ConcePTION/Level-1-checks. van den Bor, R. IMI-ConcePTION/level-2-checks [Internet]. GitHub. [cited Hoxhaj, V. IMI-ConcePTION/level-3-checks [Internet]. GitHub. [cited 2022 Jun 2022 Jun 2]. Available from: https://github.com/IMI-ConcePTION/Level-2-checks. 2]. Available from: https://github.com/IMI-ConcePTION/Level-3-checks. Willame, C., Dodd, C., Gini, R., Durán, C. E., Thomsen, R. M., Wang, L., et al. Husby, A., Hansen, J. V., Fosbøl, E., Thiesson, E. M., Madsen, M., Thomsen, R. (2021). Background rates of Adverse events of special interest for monitoring COVID- W., et al. (2021). SARS-CoV-2 vaccination and myocarditis or myopericarditis: 19 vaccines (2.0). Zenodo. doi:10.5281/zenodo.5255870 Population based cohort study. BMJ 375, e068665. doi:10.1136/bmj-2021-068665 Witberg, G., Barda, N., Hoss, S., Richter, I., Wiessman, M., Aviv, Y., et al. Karlstad, Ø., Hovi, P., Husby, A., Harkanen, T., Selmer, R. M., Pihlstrom, N., et al. (2021). Myocarditis after COVID-19 vaccination in a large health care (2022). SARS-CoV-2 vaccination and myocarditis in a nordic cohort study of organization. N. Engl. J. Med. 385 (23), 2132–2139. doi:10.1056/ 23 million residents. JAMA Cardiol. 7, 600–612. doi:10.1001/jamacardio.2022.0583 NEJMoa2110737 Frontiers in Pharmacology 12 frontiersin.org