Pilot of a Computerised Antithrombotic Risk Assessment Tool Version 2 (CARATV2.0) for stroke prevention in atrial fibrillation

Background: The decision-making process for stroke prevention in atrial fibrillation (AF) requires a comprehensive assessment of risk vs. benefit and an appropriate selection of antithrombotic agents (e. g., warfarin, non-vitamin K antagonist oral anticoagulants [NOACs]). The aim of this pilot-test was to examine the impact of a customised decision support tool -the Computerised Antithrombotic Risk Assessment Tool (CARATV2.0) using antithrombotic therapy on a cohort of patients with AF. Methods: In this prospective interventional study, 251 patients with AF aged = 65 years, admitted to a teaching hospital in Australia were recruited. CARATV2.0 generated treatment recommendations based on patient medical information. Recommendations were provided to prescribers for consideration. Results: At baseline (admission), 30.3% of patients were prescribed warfarin, 26.7% an antiplatelet, 8.4% apixaban, 8.0% rivaroxaban, 3.6% dabigatran. CARATV2.0 recommended a change of therapy for 153 (61.0%) patients. Through recommendations of CARATV2.0, at discharge, 40.2% of patients were prescribed warfarin, 17.7% antiplatelet, 14.3% apixaban, 10.4% rivaroxaban, 5.6% dabigatran. Overall, the proportion of patients receiving an antithrombotic on discharge increased significantly from baseline (admission) (baseline 77.2% vs. 89.2%; p < 0.001). Prescribers moderately agreed with CARATV2.0' s recommendations (kappa = 0.275, p < 0.001). Practical medication safety issues were cited as major reasons for not accepting a desire to continue therapy with CARATV2.0' s recommendations. Factors predicting the prescription of antiplatelets rather than anticoagulants included higher bleeding risk and high risk of falls. An inter-speciality difference in therapy selection was detected. Conclusions: This decision support tool can help optimise the use of antithrombotic therapy in patients with AF by considering risk versus benefit profiles and rationalising treatment selection.