(P-001) Signal Detection by Time-to-onset Method in Conjunction with Disproportionality Analysis: A Study of Spontaneous Reports from Japan

Background: The disproportionality analysis (DPA) has been widely used for detecting safety signals from spontaneous reports on suspected adverse drug reactions (ADRs). Recently, an additional signal detection method based on time-to-onset (TTO; time from the start of the drug to the onset of the adverse event) has been developed and assessed for both vaccines and non-vaccines.

Objectives: To study whether a TTO-based method, when used in conjunction with DPA, is more efficient in detecting and/or evaluating signals from spontaneous reports on suspected ADRs.

Methods: We used Japanese Adverse Drug Event Report database (JADER) and identified 515,707 spontaneous reports including 133,142 drug-event pairs between April 2004 and December 2014. Out of the 133,142 pairs, we analyzed 7,966 pairs that [1] reported one of the top 50 events (in terms of the number of reports per event in the whole database) and [2] had 3 or more reports with information on TTO per drug-event pair. A DPA-based signal was detected when the lower limit of 95% confidence interval of reporting odds ratio (ROR) was > 1. A TTO-based signal was detected when the distribution of TTO of a drug-event pair was significantly different (p < 0.05) from that of all other pairs with the same event by Kolmogorov-Smirnov two-sample test. Using the information on ADRs in package insert as a gold standard, sensitivity and specificity of the DPA and/or TTO-based signals were calculated.

Results: Out of the 7,966 drug-event pairs analyzed, 6,968 (87.5%) were an ADR while 998 (12.5%) were not an ADR according to the gold standard (package insert). The sensitivity of the DPA- and TTO-based signals were 48.8% (3402/6968) and 33.5% (2336/6968), respectively, and 13.2% (920/6968) ADRs were detected as a signal only by the TTO-based method. The specificity of the DPA- and TTO-based signals were 67.5% (674/998) and 76.0% (758/998), respectively. When a signal was detected only when both methods were positive, the specificity was 91.1% (909/998) though the sensitivity was low (20.3%).

Conclusions: Complementary use of TTO-based method may detect signals for ADRs not detected by DPA. In addition, signals with a high specificity may be detected when both methods are positive.