Sinds de uitroeping van de COVID-19-pandemie door de Wereldgezondheidsorganisatie (WHO) op 11 maart 2020 [1] zijn wereldwijd meer dan 13,5 miljard doses COVID-19-vaccins toegediend [2]. In november 2023 had ten minste 70,5% van de wereldbevolking ten minste één dosis van een COVID-19-vaccin gekregen [2]. Dit ongeëvenaarde scenario onderstreept de dringende behoefte aan uitgebreide monitoring van de veiligheid van vaccins, aangezien zeer zeldzame bijwerkingen die verband houden met COVID-19-vaccins pas aan het licht kunnen komen na toediening aan miljoenen individuen.

Vooruitlopend op deze ongekende wereldwijde uitrol van COVID-19-vaccins, heeft het Safety Platform for Emergency vACcines (SPEAC)-initiatief een lijst opgesteld van mogelijke COVID-19-vaccinbijwerkingen van speciaal belang (AESI) in 2020 [3]. AESI-selectie was gebaseerd op hun vooraf vastgestelde associaties met immunisatie, specifieke vaccinplatforms of adjuvantia, of virale replicatie tijdens wildtype ziekte; theoretische zorgen met betrekking tot immunopathogenese; of ondersteunend bewijs van diermodellen met behulp van kandidaat-vaccinplatforms [3].

Een flexibele benadering voor het beoordelen van AESI is de vergelijking van de waargenomen AESI-percentages na de introductie van een vaccinprogramma met de verwachte (of achtergrond)percentages op basis van historische perioden vóór de uitrol van het vaccin [4], [5]. Dergelijke vergelijkingen kunnen snel worden uitgevoerd en kunnen een sleutelrol spelen bij de vroege opsporing van mogelijke veiligheidssignalen van vaccins of wanneer regelgevende en volksgezondheidsinstanties een snelle beoordeling van een opkomend veiligheidssignaal nodig hebben [4], [6]. Waargenomen versus (vs.) verwachte (OE) analyse was een integraal onderdeel van het identificeren van trombose met trombocytopeniesyndroom (TTS) als veiligheidssignaal, wat leidde tot de opschorting van het gebruik van het ChAdOx1 (AstraZeneca COVID-19-vaccin) op 11 maart 2021 in Denemarken en Noorwegen [7], [8].

Deze evaluaties zijn niet alleen waardevol in een vroeg stadium van de grootschalige inzet van vaccins, maar ook naarmate het vaccinatieprogramma volwassener wordt, vooral als ze in een meerlandencontext kunnen worden uitgevoerd. We hebben een wereldwijde cohortstudie uitgevoerd naar aanleiding van het Observed vs. Expected Analyses of COVID-19 Adverse Events of Special Interest Study Protocol [9] met gegevens van 10 locaties in acht landen die deelnemen aan het unieke Global COVID Vaccine Safety (GCoVS) Project [10] van het Global Vaccine Data Network™ (GVDN®) [11]. Het GCoVS-project, gestart in 2021, is een door de Centers for Disease Control and Prevention (CDC) gefinancierde wereldwijde samenwerking van onderzoekers en gegevensbronnen uit meerdere landen met het oog op het toezicht op de veiligheid van COVID-19-vaccins.

1. Downloaden: Afbeelding met hoge resolutie downloaden (301KB)
2. Downloaden: Afbeelding op volledige grootte downloaden

Figuur 1. Aantal voorvallen en OE-ratio’s (met 95% betrouwbaarheidsinterval) voor homologe schema’s per dosis 1-4, neurologische aandoeningen. AESI: GBS = syndroom van Guillain-Barré, TRM = myelitis transversa, BP = faciale (Bell’s) parese, ADEM = acute gedissemineerde encefalomyelitis, FSZ = koortsstuipen, GSZ = gegeneraliseerde aanvallen. Vaccins: AZD = Oxford/Astra Zeneca/Serum Institute of India ChAdOx1, BNT = Pfizer/BioNTech (BNT162b2), MOD = Moderna (mRNA-1273).

1. Downloaden: Afbeelding met hoge resolutie downloaden (280KB)
2. Downloaden: Afbeelding op volledige grootte downloaden

Figuur 2. Aantal voorvallen en OE-ratio’s (met 95% betrouwbaarheidsinterval) voor homologe schema’s per dosis 1-4, hematologische aandoeningen. AESI: THR = trombocytopenie, ITP = idiopathische trombocytopenie, PEM = longembolie, CVST = cerebrale veneuze sinustrombose, SVT = splanchnische veneuze trombose. Vaccins: AZD = Oxford/Astra Zeneca/Serum Institute of India (ChAdOx1), BNT = Pfizer/BioNTech (BNT162b2), MOD = Moderna (mRNA-1273).

1. Downloaden: Afbeelding met hoge resolutie downloaden (199KB)
2. Downloaden: Afbeelding op volledige grootte downloaden

Figuur 3. Aantal voorvallen en OE-ratio’s (met 95% betrouwbaarheidsinterval) voor homologe schema’s per dosis 1-4, cardiovasculaire aandoeningen. AESI: MYO = Myocarditis, PER = Pericarditis. Vaccins: Vaccins: AZD = Oxford/Astra Zeneca/Serum Institute of India (ChAdOx1), BNT = Pfizer/BioNTech (BNT162b2), MOD = Moderna (mRNA-1273).

Deze multi-country cohortstudie werd uitgevoerd in de unieke setting van het GVDN. Tot op heden is het aantal van dergelijke grote systematisch gecoördineerde studies op verschillende geografische locaties en populaties beperkt. Verschillende onderzoeken hebben echter eerder de risico’s van de geïdentificeerde veiligheidssignalen na COVID-19-vaccinatie beoordeeld, voornamelijk op één locatie. We onderzochten het verband tussen COVID-19-vaccinatie en 13 AESI’s, bestaande uit neurologische, hematologische en cardiovasculaire aandoeningen, op 10 locaties in acht landen, waaronder Europa, Noord-Amerika, Zuid-Amerika en Oceanië. In deze studie, waarbij meer dan 99 miljoen mensen waren gevaccineerd tegen SARS-CoV-2, was het risico tot 42 dagen na vaccinatie over het algemeen vergelijkbaar met het achtergrondrisico voor de meeste uitkomsten; Er werden echter enkele potentiële veiligheidssignalen geïdentificeerd. We hebben potentiële veiligheidssignalen voor GBS en CVST waargenomen na de eerste dosis ChAdOx1 op basis van meer dan 12 miljoen toegediende doses.

Over het algemeen hebben studies van de op vectoren gebaseerde vaccins, zoals de ChAdOx1, een hogere incidentie van GBS na vaccinatie waargenomen in vergelijking met de achtergrondincidentie; overwegende dat de meeste studies van de mRNA-vaccins, zoals BNT162b2 en mRNA-1273, geen toename van GBS hebben waargenomen [15], [24], [14], [25], [26], [27]. Atzenhoffer et al. [24] rapporteerden een verhoogde OE-ratio > 2,0 voor adenovirus-gevectoriseerde COVID-19-vaccins, in alle landen die bijdragen aan VigiBase, een internationale database van bijwerkingen en Patone et al. [27] rapporteerde 38 extra gevallen van GBS per 10 miljoen blootgestelde personen in de risicoperiode van 1-28 dagen na vaccinatie met ChAdOx1 in Engeland. De auteurs zagen geen verhoogd risico bij degenen die BNT162b2 kregen. Daarentegen toonde een studie van Li et al. [28] geen verhoogd risico op GBS voor ChAdOx1, terwijl alleen SARS-CoV-2-infectie geassocieerd was met een hoger risico. De discrepantie, vergeleken met de resultaten van Patone et al. [27], kon echter worden verklaard door een kleinere steekproefomvang en verschillende uitkomstmaten. Over het algemeen ondersteunt dit bewijs onze bevindingen van een GBS-veiligheidssignaal na ChAdOx1-vaccinatie. Hoewel zeldzaam, werd deze associatie erkend door de WHO, het Europees Geneesmiddelenbureau (EMA) en de Therapeutic Goods Administration (TGA) van Australië, waardoor GBS werd vermeld als een zeldzame bijwerking na blootstelling aan ChAdOx1 [15], [29], [30].

Het geïdentificeerde verhoogde risico op CVST na ChAdOx1-vaccinatie in deze studie wordt bevestigd door meerdere studies. Een verhoogde OE-ratio werd waargenomen in een landelijke cohortstudie uit Denemarken en Noorwegen, met verhoogde percentages veneuze trombo-embolische voorvallen, waaronder CVST met een overschrijding van 2,5 voorvallen per 100.000 vaccinaties na ChAdOx1 [7]. Op basis van verschillende methodologieën hebben andere onderzoeken ook een verhoogde incidentie van CVST na vaccinatie gerapporteerd [31], [32]. Uiteindelijk leidde dit zeldzame maar zorgwekkende veiligheidssignaal tot de terugtrekking van het ChAdOx1-vaccin uit de COVID-19-vaccinprogramma’s of de implementatie van op leeftijd gebaseerde beperkingen in meerdere landen [8].

It is crucial to acknowledge the significance threshold of prioritised safety signals applied in this study (LBCI > 1.5). This threshold was selected based on expert opinion within the GVDN and at CDC, to focus on those outcomes most likely to be true signals. Some observed events, although not fulfilling this threshold, may still hold clinical importance and require further investigation. For instance, ITP with an OE ratio > 1.0 and LBCI of 1.2 following vaccination with ChAdOx1 aligns with findings reported in the literature as a potential signal. This concurrence is highlighted in a study conducted in Victoria, Australia, which observed a substantially higher than expected rate of ITP following ChAdOx1 vaccination [33].

Moreover, we observed significantly higher risks of myocarditis following the first, second and third doses of BNT162b2 and mRNA-1273 as well as pericarditis after the first and fourth dose of mRNA-1273, and third dose of ChAdOx1, in the 0–42 days risk period. The elevated rates of pericarditis following ChAdOx1 vaccination identified in this study rely on a limited number of observed counts in the meta-analysis. The wide confidence interval underscores the substantial uncertainty of characterizing pericarditis as a safety signal following ChAdOx1 vaccination. However, our study confirms findings of previously identified rare cases of myocarditis and pericarditis following first and second doses of mRNA vaccines [21], [22], [23], [34]. A large cohort study of 23.1 million residents across four Nordic countries revealed an increased risk of myocarditis among young males aged 16–24 years, based on 4–7 excess events in 28 days per 100,000 vaccinees after a second dose of BNT162b2, and between 9 and 28 per 100,000 vaccinees after a second dose of mRNA-1273 [22]. Similarly, studies from British Columbia, Canada reported cases of myocarditis to be higher among those receiving a second dose compared with a third dose, and for those who received a second dose of the mRNA-1273 vaccine compared with the BNT162b2 vaccine [35], [36]. Patone et al. [37] estimated extra myocarditis events to be between one and 10 per million persons in the month following vaccination, which was substantially lower than the 40 extra events per million persons observed following SARS-CoV-2 infection period. A systematic review by Alami et al. [38] concluded that mRNA vaccinated individuals were twice as likely to develop myocarditis/pericarditis compared with unvaccinated individuals, with a rate ratio of 2.05 (95 % CI 1.49–2.82). Given the evidence, WHO issued updated guidance regarding these safety signals and mRNA COVID-19 vaccination, and EMA provided updates to the Product Information for BNT162b2 and mRNA-1273 vaccines [21], [23]. TGA as well as the CDC continue to monitor and review data on myocarditis and pericarditis following COVID-19 vaccination [39], [40].

Another potential safety signal was identified for ADEM after the first dose of mRNA-1273 vaccine, with five more observed than expected events based on 1,035,871 person-years and 10.5 million doses administered; however, the number of cases of this rare event were small and the confidence interval wide, so results should be interpreted with caution and confirmed in future studies. Although some case reports have suggested a possible association between COVID-19 vaccination and ADEM, there was no consistent pattern in terms of vaccine or timing following vaccination, and larger epidemiological studies have not confirmed any potential association [41], [42], [43], [44]. Moreover, case reports may report on coincidental events and do not establish association nor indicate causality, thus larger observational studies are warranted to further investigate our finding. To address this, a follow-up study is currently being undertaken within the GVDN, focusing on a demographic not included in our analysis. Based on reports of rare ADEM cases to the European Database of Suspected Adverse Drug Reaction, EMA assessed the potential association of ADEM following vaccination with ChAdOx1 [45]. Frontera et al. [46] concluded that chances of having a neurological event following acute SARS-CoV-2 infection were up to 617-fold higher than following COVID vaccination, suggesting that the benefits of vaccination substantially outweigh the risks. A safety signal for generalized seizures was identified following Gamaleya Research Institute/Sputnik vaccination, however the number of vaccinations was relatively low compared with other vaccines in this study. Further studies are warranted to explore this potential safety signal.

Conducting a cohort analysis in the unique multi-country context of the GVDN leverages a vast and diverse data pool. Aggregating data from multiple countries on more than 99 million vaccine recipients has significantly increased the sample size and the statistical power compared with many previous safety studies. This enhances the ability to detect safety signals, especially for extremely rare adverse events, as the larger sample size provides greater precision in estimating observed rates.

Results based on data across Europe, North and South America and Oceania offer stronger external validity, enabling findings to be more generalizable to a broader range of populations and healthcare settings participating in the global COVID-19 vaccination programme. Moreover, multi-country analyses facilitate comparisons between countries with varying vaccination strategies, population demographics, and healthcare systems, yielding insights into how these factors may influence vaccine safety profiles. Data used in our analysis were drawn from multiple databases, including healthcare databases, national immunization registries, and vaccination dashboards, allowing the identification of potential safety signals from various sources.

The results from our study should, however, be interpreted considering multiple limitations. Our analyses inherently involve heterogeneity in data collection, quality, and reporting standards across countries. These differences in healthcare infrastructure and surveillance systems can introduce bias and affect the comparability of results. The participating sites across the eight countries implemented varied vaccination strategies, including vaccine types, dosing schedules, and prioritization of vaccine recipients. Moreover, the multi-country analyses are susceptible to population confounding factors, such as differences in pre-existing health conditions, genetic factors, ethnic profiles, and behavioural patterns, which was not possible to adjust for in our analysis. We consider our approach suitable for application in large datasets representing average populations. However, age- and sex-specific historic background rates that are not adjusted for factors like prior disease may not provide a suitable comparison, for example, in the early stages of a vaccination campaigns where people with co-morbidities were vaccinated prior to other population groups.

Potential underreporting across countries may have led to an underestimation of the significance of potential safety signals. It is important to recognize the potential for false negatives, especially when detecting associations with lower confidence intervals below 1.5 that maintain statistical significance. The safety signals identified in this study should be evaluated in the context of their rarity, severity, and clinical relevance. Moreover, overall risk–benefit evaluations of vaccination should take the risk associated with infection into account, as multiple studies demonstrated higher risk of developing the events under study, such as GBS, myocarditis, or ADEM, following SARS-CoV-2 infection than vaccination. Finally, the use of ICD-10 codes is subject to considerations about specificity and sensitivity, and application may vary by country.

Observed vs. expected analyses in a multi-country context of the GVDN and the GCoVS Project offers a larger and more diverse dataset, enhanced generalizability, and improved statistical power over single site or regional studies. It also presents challenges related to data heterogeneity, population confounding factors, and variations in vaccination strategies and reporting systems. The involvement of researchers and data sources from diverse regions of the world promotes inclusivity, reduces potential biases, and fosters collaboration in the pursuit of a shared public health goal. While our study confirmed previously identified rare safety signals following COVID-19 vaccination and contributed evidence on several other important outcomes, further investigation is warranted to confirm associations and assess clinical significance. This could be addressed by conducting association studies specific to individual outcomes by applying methodologies such as the self-controlled case series (SCCS) to validate the associations [6].

All analyses, inferences drawn, opinions, conclusions, and statements are those of the authors and do not necessarily represent the official views of, nor an endorsement by, CDC/HHS, or the U.S. Government. For more information, please visit cdc.gov.

Parts of this material are based on data and/or information compiled and provided by the Canadian Institute for Health Information and the Ontario Ministry of Health. The analyses, conclusions, opinions, and statements expressed herein are solely those of the authors and do not reflect those of the funding or data sources; no endorsement is intended or should be inferred. Parts of this material are based data and/or information provided by the British Columbia Ministry of Health. All inferences, opinions, and conclusions drawn in this manuscript are those of the authors, and do not reflect the opinions or policies of the Data Steward(s).

The GCoVS project is supported by the Centers for Disease Control and Prevention (CDC) of the U.S. Department of Health and Human Services (HHS) as part of a financial assistance award totalling US$10,108,491 with 100 % per cent funded by CDC/HHS.

The Ontario site contributing to this study was supported by Public Health Ontario and by the ICES, which is funded by an annual grant from the Ontario Ministry of Health. JCK is supported by a Clinician-Scientist Award from the University of Toronto Department of Family and Community Medicine.

K. Faksova: Visualization, Writing – original draft, Writing – review & editing. D. Walsh: Data curation, Formal analysis, Investigation, Methodology, Software, Validation, Conceptualization, Writing – review & editing, Visualization. Y. Jiang: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Supervision, Validation, Visualization, Writing – review & editing. J. Griffin: Conceptualization, Writing – review & editing, Methodology. A. Phillips: Conceptualization, Methodology, Writing – review & editing, Investigation, Validation. A. Gentile: Data curation, Investigation, Supervision, Validation. J.C. Kwong: Data curation, Supervision, Validation, Writing – review & editing, Investigation, Methodology. K. Macartney: Data curation, Supervision, Validation, Writing – review & editing, Investigation, Methodology. M. Naus: Data curation, Supervision, Validation, Investigation, Methodology. Z. Grange: Data curation, Supervision, Validation, Conceptualization, Investigation, Methodology. S. Escolano: Data curation, Supervision, Validation, Investigation, Methodology, Writing – review & editing. G. Sepulveda: Data curation, Formal analysis, Software, Validation. A. Shetty: Data curation, Validation, Investigation, Methodology. A. Pillsbury: Data curation, Validation, Investigation, Methodology, Writing – review & editing. C. Sullivan: Data curation, Validation, Investigation, Methodology, Writing – review & editing. Z. Naveed: Data curation, Validation, Investigation, Methodology, Writing – review & editing. N.Z. Janjua: Data curation, Writing – review & editing. N. Giglio: Data curation, Investigation, Methodology, Validation. J. Perälä: Data curation, Investigation, Methodology, Validation. S. Nasreen: Conceptualization, Data curation, Validation, Writing – review & editing. H. Gidding: Conceptualization, Validation, Writing – review & editing, Investigation, Methodology. P. Hovi: Conceptualization, Validation, Writing – review & editing, Investigation, Methodology. T. Vo: Conceptualization, Validation, Formal analysis, Investigation, Methodology, Writing – review & editing. F. Cui: Conceptualization, Investigation, Methodology, Validation. L. Deng: Conceptualization, Investigation, Methodology, Validation, Writing – review & editing. L. Cullen: Conceptualization, Investigation, Methodology, Validation, Writing – review & editing. M. Artama: Conceptualization, Investigation, Methodology, Validation, Writing – review & editing. H. Lu: Data curation, Formal analysis, Software, Validation. H.J. Clothier: Conceptualization, Methodology, Validation, Writing – review & editing, Data curation, Formal analysis, Project administration. K. Batty: Conceptualization, Methodology, Project administration, Validation, Writing – review & editing. J. Paynter: Conceptualization, Methodology, Supervision, Data curation, Formal analysis, Writing – review & editing. H. Petousis-Harris: Conceptualization, Funding acquisition, Investigation, Methodology, Resources, Supervision, Writing – review & editing, Project administration. J. Buttery: Conceptualization, Funding acquisition, Investigation, Methodology, Resources, Supervision, Validation, Writing – review & editing, Project administration. S. Black: Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Writing – review & editing. A. Hviid: Conceptualization, Investigation, Methodology, Supervision, Validation, Writing – review & editing.