Abstract 16467: Can a Model-Based Decision Support Tool Improve Anticoagulation Outcomes for Atrial Fibrillation? Results From a Large Database Cohort
Background We recently reported a novel decision model that integrates both stroke and bleeding risk scores to guide optimal use of anticoagulation (AC) for stroke prevention in atrial fibrillation (AF) (Casciano 2011). The current study assesses the validity of model recommendations by comparing stroke and hemorrhagic outcomes in patients with warfarin exposure concordant or discordant with recommendations.
Methods The outcomes of 61,769 patients with recent-onset AF from the MarketScan database were simulated via a Markov model based on entry demographics, stroke risk (CHADS2), bleeding risk (ATRIA), and warfarin efficacy from clinical trials. Warfarin was recommended if predicted quality-adjusted life-years were higher than on aspirin. Warfarin use was identified by a prescription or prothrombin time claim <90 days after index AF diagnosis. Ischemic stroke and major (ie, hospitalized) hemorrhagic events were identified by inpatient claims with primary ICD-9-CM diagnoses. Cox proportional hazards models controlling for potential confounders compared risk for ischemic stroke and major bleed between patients who received warfarin versus those who did not, stratified by warfarin use concordant or discordant with model recommendations.
Results Among patients who were recommended to receive warfarin, ischemic stroke risk was significantly decreased for those who received it versus those who did not (HR=0.72, 95% CI=0.63-0.82). Among patients who were not recommended warfarin, ischemic stroke risk was not significantly different for those who received it versus those who did not (HR=0.90, 95% CI=0.65-1.24). Warfarin use was associated with higher risks of major bleed in both patients recommended (HR=1.52, 95% CI=1.35-1.71) and not recommended to receive warfarin (HR=1.59, 95% CI=1.26-2.00). Warfarin use concordant with recommendations was associated with an increase of 0.41 major bleeds and a decrease of 0.60 ischemic strokes, while discordant use was associated with an increase of 0.31 major bleeds but a decrease of only 0.09 ischemic strokes per 100 person-years.
Conclusions Our findings provide empirical support for use of an evidence-based decision model to help identify AF patients for whom the benefits of warfarin outweigh its risk.
- © 2011 by American Heart Association, Inc.