BACKGROUND: Detection of undiagnosed atrial fibrillation (AF) after ischemic stroke through extended cardiac monitoring is important for preventing recurrent stroke. We evaluated whether a tool that displays clinically predicted AF risk to clinicians caring for stroke patients was associated with the use of extended cardiac monitoring. METHODS: We prospectively included hospitalized ischemic stroke patients without known AF in a pre-intervention (October 2018 - June 2019) and intervention period (March 11, 2021 - March 10, 2022). The intervention consisted of an electronic health record (EHR)-based best-practice advisory (BPA) alert which calculated and displayed 5-year risk of AF. We used a multivariable Fine and Gray model to test for an interaction between predicted AF risk and period (pre-intervention vs. intervention) with regards to incidence of extended cardiac monitoring. We compared the incidence of extended cardiac monitoring within 6-months of discharge between periods, stratified by BPA completion. RESULTS: We included 805 patients: 493 in the pre-intervention cohort and 312 in the intervention cohort. In the intervention cohort, the BPA was completed for 180 (58%) patients. The association between predicted clinical risk of AF and incidence of 6-month extended cardiac monitoring was not different by time period (interaction HR = 1.00 [95% Confidence Interval (CI) 0.98
1.02]). The intervention period was associated with an increased cumulative incidence of cardiac monitoring (adjusted HR = 1.32 [95% CI 1.03-1.69]). CONCLUSIONS: An embedded EHR tool displaying predicted AF risk in a post-stroke setting had limited clinician engagement and predicted risk was not associated with the use of extended cardiac monitoring. CLINICAL TRIAL REGISTRATION: NCT04637087.