Exploring the Use of ClinicalTrials.gov Trial Results Data for Pharmacovigilance

Pharmacovigilance aims to monitor drug safety using sources such as spontaneous reporting systems, biomedical literature or electronic health record data. Clinical trials represent a source of drug-event pairs data complementary to these sources for signal detection in pharmacovigilance platforms. The advantage of ClinicalTrials.gov (CTG) over other pharmacovigilance sources is the large number of negative drug-event pairs (explicit evidence that a given drug is not causing a particular adverse event; count of 0 is reported in the deposited results). With 208,959 trials registered and 20,025 trial results, CTG is the largest repository of trial summary results (with more than 3700 new trial results deposited per year). CTG trial registration data provide information about trial type, sponsor, arms and interventions. CTG results data further provide trial participant counts, baseline characteristics, outcome measures and, most importantly for our study, significant adverse events. Adverse events (AE) are recorded separately by trial arm (or trial group). Although some data submitted to CTG are structured, such as number of arms or intervention type (e.g., drug vs. procedure), many elements are collected as free text (e.g., the drug(s) used in the trial). This preliminary investigation explores the selection of clinical trials of interest for pharmacovigilance and the feasibility of extracting drug concepts from CTG trial registration data.