Background
Information on when in time an adverse drug reaction (ADR) is most likely to occur in relation to when a drug therapy is initiated (denoted time-to-onset in this study) is of clinical importance. It helps patients and health care professionals to be aware of when they should be particularly vigilant in following up on new prescriptions. Within pharmacovigilance, time-to-onset is one of the most fundamental criteria when assessing the likelihood of a causal relationship between a suspected ADR and a drug. Time-to-onset in case series analysis is crucial in order to determine whether a typical pattern exists, which can provide clues to the pharmacological mechanisms behind the ADR.
There have been suggestions on using time-to-onset data from large collections of individual case reports to develop methods for detecting safety signals based on statistical analysis [1,2]. Maignen et al. used methods from survival analysis to fit mathematical models to the distribution of reported time-to-onset for different drug-ADR pairs [1]. Van Holle et al. instead identified drug-ADR pairs with empirical distributions of reported time-to-onset that deviate from other ADRs with the drug of interest and of the ADR of interest with other drugs [2]. However, time-to-onset information from reports is known to be heterogeneously collected [3]. Additionally, such large collections include a multitude of drugs with various durations of use, which might affect any comparison of aggregated reported time-to-onset, if not explicitly adjusted for. Some adverse reactions only occur during treatment. In general, adverse events are less likely to be suspected and reported after treatment ends. Thus, the observed time-to-onset may depend on both the latency of the event and the expected duration of treatment. The aim of this study was to investigate the impact of differences in duration of treatment on the reported time-to-onset of angioedema and hepatitis in a spontaneous reporting system, to inform future signal detection and analysis based on aggregated reported time-to-onset.