abstract
- INTRODUCTION: This study tests the hypothesis that nonadherence with INR monitoring is associated with an increased risk for warfarin-related bleeding and thrombosis and describes patient characteristics associated with INR monitoring nonadherence. MATERIALS AND METHODS: This was a retrospective, longitudinal, matched cohort study wherein patients were categorized into adherent and nonadherent cohorts; adherent patients were matched 2:1 to nonadherent patients. The primary study endpoint was the first occurrence of bleeding or thromboembolism. Multivariate logistic regression modeling identified patient characteristics associated with INR monitoring adherence or nonadherence. RESULTS: A total of 4995 and 2544 patients contributed 10729 and 5385 patient-years of warfarin therapy in the adherent and nonadherent groups, respectively. The rate of thromboembolic events during follow up was higher in the nonadherent group than in the adherent group (0.95% vs. 0.62% per patient-year, respectively; p=0.019) and nonadherence to INR monitoring was associated with a moderately higher risk of thromboembolism (adjusted Hazard Ratio=1.51; 95% confidence interval=1.04 - 2.20). The difference in bleeding between the two groups was not statistically significant. CONCLUSIONS: Repeatedly missing INR tests is an easily identified clinical parameter that is associated with moderately increased risk for thromboembolism in patients taking chronic warfarin therapy. Clinicians should carefully consider the underlying thromboembolic risk and extent of nonadherence when weighing the benefits of continued warfarin therapy for a given patient.