Impact of CYP3A4/P-gp interacting medications on clinical outcomes in nonvalvular atrial fibrillation patients managed on rivaroxaban

Access full-text files




Goble, Joseph Adam

Journal Title

Journal ISSN

Volume Title



Objectives: To analyze concurrent drug use and its association with outcomes in nonvalvular atrial fibrillation patients receiving rivaroxaban. Methods: We included patients ≥ 18 years of age who had at least one prescription for rivaroxaban between January 1, 2012 and December 31, 2016 along with at least two diagnoses of atrial fibrillation during the pre-index period. Those on rivaroxaban with concurrent exposure to CYP3A4/P-gp interacting medications were placed in the concomitant interacting medication (CIM) user group, and those without exposure to CYP3A4/P-gp interacting medications were placed in the CIM non-user group. Patients were excluded if they did not have continuous enrollment in the 365 days before and after the first prescription of rivaroxaban during the study period (allowing for ≤ 90-day gaps in coverage). Baseline characteristics, comorbidities, medication use, and occurrence of adverse events between the two groups were compared using inferential statistics. Multivariate logistic regression models with modified Poisson distributions and Cox proportional hazard models were used to address the hypotheses, assessing the effect of CIM on bleed and thromboembolism, adjusted for covariates. Results: A total of 287 and 180 patients were included in the CIM users group and the CIM non-users group, respectively. At baseline, compared to CIM non-users, CIM users had a higher proportion of pre-index cardiovascular issues: myocardial infarction (12.2% versus 6.1%), heart failure (43.6% versus 29.4%), and coronary artery disease (42.9% versus 32.2%). The average pre-index CHADS2VASc (4.2) and Charlson Comorbidity Index scores (2.7) in both cohorts were similar. In unadjusted analyses, the CIM user cohort had similar rates of bleed (13.9% versus 16.7%) and lower rates of thromboembolism (13.2% versus 28.3%) compared to the CIM non-user group. When controlling for covariates, relative to CIM non-users, CIM users had a comparable risk of bleed (risk ratio [RR]: 0.73; p = 0.14) and a lower risk of thromboembolism (RR: 0.58; p < 0.01). When adjusted for covariates and assessing time to first event, the hazard of thromboembolism was about half for CIM users (hazard ratio [HR] = 0.58; 95% confidence interval [CI]: 0.39, 0.87; p < 0.01). Conclusion: This study suggests that co-administered CYP3A4 and P-gp inhibitors reduce the risk of thromboembolism in patients prescribed rivaroxaban. While no alteration of bleeding risk was observed, this may be limited by event rates, sample size, and inability to assess risk at the individual drug level.


LCSH Subject Headings