Modern pharmacovigilance transforms how we detect emerging drug safety concerns

Pharmacovigilance has evolved considerably from a passive, retrospective activity into a dynamic, multi-source discipline central to the identification, characterization, and management of drug safety signals. While RCTs remain the gold standard for establishing efficacy and identifying common adverse events. their limited sample sizes, short durations, and restrictive inclusion criteria mean that rare, delayed-onset, or population-specific adverse drug reactions frequently escape detection prior to approval. Postmarketing surveillance—anchored by spontaneous reporting systems (SRSs) such as the FDA Adverse Event Reporting System (FAERS), the WHO’s VigiBase, and the EMA’s EudraVigilance—has therefore become an indispensable complement to premarket evaluation, enabling the discovery of safety signals that only become apparent once a drug is used in real-world populations

The scale of these databases has grown dramatically; VigiBase alone had accumulated approximately 40 million reports by early 2025, and FAERS approximately 30 million reports, corresponding to nearly 5 million reported medicinal product–event pairs. This exponential growth has been enabled by increased regulatory emphasis on reporting, greater public accessibility of these databases, which has in turn fueled a substantial expansion of published pharmacovigilance research.

The principal analytic engine of modern pharmacovigilance is disproportionality analysis, which quantifies whether one adverse event is reported more frequently for a given drug than would be expected based on its reporting frequency for other drugs. Common metrics include the reporting odds ratio, and Bayesian confidence propagation neural network/information component approaches. These frequentist and Bayesian techniques allow to flag statistical associations for further clinical scrutiny.