Methods:

An alert-based CDS program was developed that identified hospitalized patients with increased risk of stroke. A best practice advisory was placed in the electronic health record at Brigham and Women’s Hospital. Eligible patents with AF at increased risk of stroke but without a current anticoagulation order were identified (n = 458). They were randomly assigned to alert-based CDS (intervention) versus no notification (control). Patients were identified by the CDS program to have new, paroxysmal, persistent, or permanent non-valvular AF or atrial flutter by problem list, medical history, or admitting diagnosis. Their CHA2DS2-VASc score was calculated from data in the electronic health record. CDS defined anticoagulant therapy for stroke prevention as therapeutic doses of dabigatran, rivaroxaban, apixaban, edoxaban, intravenous unfractionated heparin, low-molecular weight heparins, fondaparinux, intravenous direct thrombin inhibitors, or warfarin with a target international normalized ratio of 2-3.

Providers in the alert group received a best practice advisory when completing admissions orders that stated patient’s indication for anticoagulation, increased risk of stroke, and lack of active anticoagulation order. Three options were given: 1) access to order Food and Drug Administration approved anticoagulation regimens for stroke prevention, 2) a link to information regarding evidence-based practice guidelines for stroke prevention in AF, or 3) proceeding without anticoagulation after providing rationale. Prescription of anticoagulation therapy was assessed during hospitalization, at discharge, and at 90 days.

The primary efficacy outcome was frequency of new anticoagulation prescription orders in alert versus control groups. The secondary efficacy outcome was occurrence of the composite of MACE (cerebrovascular accident, systemic embolism, myocardial infarction, and all-cause mortality) at 90 days. The primary safety outcome was major bleeding or clinically relevant non-major bleeding at 90 days.