Theoretical Framework and Key Considerations for Time-to-Onset Analysis in Spontaneous Reporting Systems

Spontaneous reporting databases play a central role in pharmacovigilance for monitoring the safety of drugs and vaccines. Conventional statistical signal detection has relied primarily on disproportionality analyses based on reporting frequencies, whereas information on the timing of adverse event onset has not been fully exploited. Time to onset (TTO), defined as the interval between the initiation of drug administration and the occurrence of an adverse event, provides complementary information that captures temporal patterns of event manifestation beyond simple occurrence counts. This review summarizes the definition, calculation, characteristics, and limitations of TTO analyses in spontaneous reporting databases and provides an overview of statistical signal detection methods incorporating TTO information. In particular, nonparametric distribution-comparison approaches, such as the Kolmogorov-Smirnov and Anderson-Darling tests, are well suited to spontaneous reporting data, in which the underlying population and exposure size are unknown. These methods enable the detection of abnormalities in the temporal structure of adverse event onset that may not be identifiable through frequency-based analyses alone. Furthermore, disproportionality analysis and TTO-based approaches are not competing methods but complementary strategies that capture different dimensions of safety signals-reporting frequency and temporal patterns-and their combined use may improve both sensitivity and interpretability of signal detection. The review also discusses survival analysis-based methods and Weibull modeling for TTO data, outlining their theoretical background and applications while emphasizing their inherent limitations when applied to spontaneous reporting systems. Because of reporting bias, incomplete time information, and the absence of non-event cases, such methods should not be used to estimate population-level risks or to infer causality. In conclusion, TTO analyses using spontaneous reporting databases should be positioned as exploratory tools for characterizing onset patterns, generating hypotheses, and informing the design of subsequent epidemiological and safety studies, rather than as a direct basis for clinical or regulatory decision making.