Identifying Signals Of Potential Drug-Drug Interactions Involving Oral Anticoagulants Using Real World Clinical Data

Drug-drug interactions (DDIs) with oral anticoagulants may lead to increased risk of serious bleeding and/or thromboembolism. Most oral anticoagulant users are older adults, who are likely to be taking multiple medications and therefore more susceptible to DDIs. Warfarin, the mainstay anticoagulant therapy for decades, is susceptible to numerous DDIs via various mechanisms. Direct-acting oral anticoagulants (DOACs) were introduced as alternatives and are increasingly adopted in clinical practice. Few studies have examined clinically significant DDIs involving DOACs. We aimed to systematically identify signals of potential oral anticoagulant DDIs to cause serious bleeding and/or thromboembolism, and to affect international normalized ratio (INR, a quantitative biomarker of warfarin-induced anticoagulation) using real world clinical data. We conducted three high-throughput pharmacoepidemiologic screening studies in parallel using a large healthcare claim database in the United States. In all three studies, we examined all oral medications frequently co-prescribed with oral anticoagulants as potential interacting precipitants. First, we conducted a series of self-controlled case series studies to identify which precipitants were associated with increased rate of serious bleeding. Second, we conducted a series of self-controlled case series studies to identify which precipitants were associated with increased rate of thromboembolism. In both studies, conditional Poisson regression was used to estimate rate ratios comparing precipitant exposed vs. unexposed time for each anticoagulant-precipitant drug pair. Pravastatin was examined as a quantitative negative control object drug. Third, we conducted a series of cohort studies to identify which precipitants were associated with an increased or decreased INR in warfarin users. Change in INR after vs. before precipitant initiation was estimated using conditional mixed effect model adjusted for warfarin dose change. We identified 71 and 81 potential DDI signals associated with increased rate of serious bleeding and thromboembolism, respectively. We found 63 and 51 precipitants associated with elevated and reduced INR following precipitant initiation, respectively. We reproduced previously documented DDIs, which supports the validity of our approach. The newly identified potential DDI signals need to be examined in future studies. The pharmacoepidemiologic screening approaches of biomarker and clinical endpoints could be useful for systematic identification of DDIs involving other drugs