Comparing Existing Algorithms for Retrieving Pregnancy-related Adverse Event Reports

Abstract Background Post-marketing surveillance is essential for complementing the safety profiles of medicinal products, especially for populations generally excluded from clinical trials such as pregnant individuals. However, the absence of a standardised pregnancy indicator in the electronic transmissions of adverse event reports hampers their correct identification in pharmacovigilance databases and complicates the study of safety concerns related to pregnancy exposures. Three recently developed rule-based algorithms with the common aim to systematically retrieve pregnancy-related reports differ in scope and are tailored to different databases (A. FAERS, B. EudraVigilance, C. VigiBase). Aim To compare the design and outputs of the three pregnancy algorithms. Methods This study was a collaboration among the authors of the three pregnancy algorithms. We harmonised their rules, implemented them in an R package to enable execution in both VigiBase and FAERS, and analysed key characteristics of reports flagged by each algorithm. Results The pregnancy algorithms A, B, and C flagged 235653, 279515, and 446957 reports respectively in VigiBase, and 265015, 260734, 350479 in FAERS. Reports exclusively retrieved by each algorithm (994, 3248, and 142324 in VigiBase, and 1528, 1100, and 59643 in FAERS) were mostly explained by Algorithm A having no age restriction, Algorithm B excluding normal pregnancy and ineffective contraception, and Algorithm C excluding paternal exposure. Conclusions Differences in flagging were largely related to varying scopes. Understanding commonalities and differences is crucial for empowering professionals working with pregnancy-related pharmacovigilance to select and use the most appropriate algorithm for their specific needs. Key points Three independently developed algorithms were designed to retrieve pregnancy-related adverse event reports and support research into pregnancy-specific safety concerns. By applying these algorithms to VigiBase and FAERS, we highlighted overlaps and differences in the reports they flag, reflecting heterogeneous scope and implementation. Awareness of these distinctions is essential to select and apply the most suitable algorithm for their specific needs.