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. (Cardiol J 2017; 24, 2: 176-187).