Statistical signal detection methods for post licensure vaccine safety surveillance: A review

Post licensure safety surveillance is crucial for vaccines, given their administration to healthy populations and the necessity of detecting rare adverse events following immunization. This review outlines statistical signal detection methods utilized in passive and active surveillance systems, referring to current frameworks in the United States and Japan. Regarding passive surveillance, including spontaneous reporting, we discuss frequentist and Bayesian disproportionality analysis. As for specific methods frequently utilized in vaccine signal detection, we detail (conditional) maximized sequential probability ratio test (MaxSPRT), self-controlled case series, case-centered analysis, and time-to-onset analysis. We illustrate these methods through practical examples, such as seasonal influenza vaccine safety monitoring and COVID-19 vaccine assessments. Furthermore, we implement a simulation study to evaluate the performance of MaxSPRT, demonstrating its utility for rapid signal detection. Finally, we briefly address methodological challenges and the potential of causal inference frameworks for future surveillance.