Incidence of adverse events following COVID-19 vaccination in observational studies

Introduction This systematic review and meta-analysis aimed to estimate and synthesize real-world incidence measures of adverse events following immunization (AEFI) associated with COVID-19 vaccines and to examine how methodological and epidemiological factors influence reported estimates. Methods Following PRISMA 2020 guidelines and a prospectively registered protocol (PROSPERO), eight data sources including grey literature were searched for observational studies reporting AEFI incidence in the general population. A total of 160 studies met inclusion criteria. Incidence metrics were grouped by unit of analysis: events per notification, per exposed individual, per 100 doses, and cumulative incidence of individuals experiencing at least one AEFI. Subgroup analyses evaluated variation by age group, vaccine platform, number of doses, data source, study design, and denominator type. Risk of bias was assessed using the JBI tools and ROBINS-I. Results Pooled estimates differed according to metric definition. Event-to-exposure ratios yielded a combined estimate of 0.118 (95% CI: 0.066–0.211; 154 studies), whereas cumulative incidence proportions yielded 0.561 (95% CI: 0.336–0.763; 6 studies). Heterogeneity was extreme (I² ≥ 99%), partly reflecting denominator type: ratios ranged from 0.002 (notification-based) to 0.378 (dose-based) ( p < 0.001). Most studies had moderate-to-high risk of bias; excluding high-risk studies reduced the Set 1 estimate by approximately 41% to 0.069 (95% CI: 0.033–0.146). Conclusions Real-world AEFI incidence varies according to the epidemiological metric, denominator structure, and methodological quality of primary studies; event-to-exposure ratios and cumulative incidence proportions capture distinct dimensions of AEFI occurrence and should not be directly compared.