By Michael H. Crawford, MD, Editor

SOURCE: Chang SH, Chou IJ, Yeh YH, et al. Association between use of non-vitamin K oral anticoagulants with and without concurrent medications and risk of major bleeding in nonvalvular atrial fibrillation. JAMA 2017;318:1250-1259.

Little is known about the risk of bleeding with the new oral anticoagulants (NOACs) in atrial fibrillation (AF) patients on multiple other drugs for various comorbidities. Investigators evaluated the bleeding risk in AF patients on NOACs associated with the concurrent use of 12 commonly prescribed medications that share the metabolism pathways of the NOACs, such as CYP3A4 inhibitors and P-glycoprotein competitors. This was a retrospective analysis of the Taiwan National Health Insurance Administration database, which includes robust clinical information. More than 91,000 patients were identified who demonstrated nonvalvular AF and had received at least one NOAC prescription between 2012 and 2016. The primary outcome was major bleeding, excluding trauma bleeding. The analysis model used a propensity score to adjust for covariates. The mean age of the patients was 75 years. Fifty-six percent were men. Their mean baseline CHA2DS2-VASc score was 4, and their HAS-BLED mean score was 3.3. Comorbidities such as heart failure, diabetes, and cerebrovascular disease were common. The NOACs used during the four years were rivaroxaban in 59%, dabigatran in 50%, and apixaban in 14%. There were 4,770 major bleeds per 447,037 person-quarter of years of follow-up. Significant increases in adjusted bleeding rate differences compared to NOAC use alone were seen with amiodarone (14 per 1,000 person-years), fluconazole (138), rifampin (37), and phenytoin (52). Atorvastatin, digoxin, erythromycin and clarithromycin were associated with reduced incidence rates. Other combinations, such as with diltiazem, verapamil, cyclosporine, and other azoles, neither increased nor decreased bleeding rates. There were no differences in bleeding rates on the various combinations between the three NOACs. The authors concluded that in patients taking NOACs, concomitant use of amiodarone, fluconazole, rifampin, and phenytoin is associated with a significant risk of major bleeding compared to NOAC use alone.


Many hoped that the NOACs would free clinicians from the concern about drug interactions with warfarin. The authors of this study noted that drug interactions are reduced, but not eliminated, by NOACs. Clearly, there are drugs clinicians should avoid prescribing for patients on NOACs because of a marked increase in the risk of bleeding. These are in the order of risk: fluconazole, phenytoin, rifampin, and amiodarone. The data on amiodarone are different from data presented in the ARISTOTLE study (apixaban vs. warfarin) in which no increase in bleeding risk was found. However, the authors of ARISTOTLE studied far fewer patients and this was the lowest increased risk found in this study. On the other hand, several drugs that are known to increase plasma levels of NOACs in pharmacodynamic studies and were predicted to increase bleeding risk did not in this analysis: diltiazem, verapamil, cyclosporine, and other azoles. Paradoxically, some drugs actually appeared to lower the bleeding risk: atorvastatin, digoxin, and mycin antibiotics. This may be because other positive effects of these drugs lowered bleeding risk or chance, but their use does not seem to be a concern. Of the four drugs that accounted for 20% of the concomitant drug use (digoxin, diltiazem, amiodarone, and atorvastatin), only amiodarone was of concern. The strength of this study is that it was based on a very large and comprehensive nationwide database in a country with a one-payer system. Potential limitations are that this country (Taiwan) represents an Asian population and that the data may not reflect other ethnicities. Also, drug dosages were not considered in the analysis because of the overwhelming complexity this would pose. Additionally, liver and renal function data were not analyzed for the same reason. Finally, there are no data on edoxaban, which was not approved for use in Taiwan during the study period. However, there was no difference between the other three NOACs in the observed interaction patterns, so there is no reason to suspect edoxaban would act differently.

The use of amiodarone, fluconazole, rifampin, and phenytoin with NOACs increased the risk of major bleeding, whereas other commonly used drugs in AF patients did not.