Abstract 013: A Model for Stroke Prediction Using Claims Data in a Contemporary Cohort of Patients With Atrial Fibrillation Receiving Oral Anticoagulation

    loading  Checking for direct PDF access through Ovid

Abstract

Background: Oral anticoagulants (OACs) are recommended for AF patients for the prevention of thromboembolic events, including stroke. Stroke risk stratification scores (e.g., CHA2DS2-VASc) are used to tailor therapeutic recommendations for patients in different risk groups. However, these were derived before the advent of direct OACs. At present, there is no tool that estimates a patient’s stroke risk, given their individual characteristics, by type of OAC.

Methods: We used healthcare utilization data from two independent US databases (MarketScan and Optum) to construct and validate a predictive model of ischemic stroke in patients with AF initiating OACs. Patients with non-valvular AF initiating OACs were identified from the MarketScan data for years 2007-2015. Using bootstrapping methods and backward selection of 44 candidate variables, we developed a model which selected variables predicting stroke. The final model was validated in patients with non-valvular AF in the Optum Clinformatics database in the period 2009-2015.

Results: Among 135,523 patients with AF initiating OAC in the MarketScan dataset, 2,028 experienced an ischemic stroke after anticoagulant initiation. The stepwise model identified 15 variables (including type of OAC) associated with ischemic stroke (Table). The discrimination (c-statistic) of the model was adequate [0.68, 95% confidence interval (CI) 0.66-0.70], showing excellent calibration (χ 2= 7.7 p=0.57). The model was then applied to the 84,549 AF patients in the Optum data set (1408 stroke events). The model showed similar discrimination (c-statistic 0.67, 95%CI 0.65-0.70) and calibration (χ 2= 12.8 p=0.17). However, previously-developed predicted models had similar discriminative ability (CHA2DS2-VASc 0.67, 95%CI 0.65-0.68; ATRIA 0.67, 95%CI 0.65-0.68)

Conclusion: A novel model using extensive administrative healthcare data for the identification of patients at higher risk of ischemic stroke by type of anticoagulant did not perform better than established simple models.

Related Topics

    loading  Loading Related Articles