Abstract 55: Automated, Real-Time, Electronic Health-Record Initiated Alerts Expedite Study Team Notification of Potentially Eligible Subjects for an Acute Stroke Trial

Introduction: Patients must be quickly identified to be eligible for enrollment in acute stroke studies. Clinical providers are often too busy with patient care to be responsible for identification of study patients. Electronic health records quickly incorporate a wealth of data, including laboratory results, into patient level records and can be used to potentially identify subjects in real-time. Our objective was to describe the implementation and performance of an automated screening tool for an acute stroke trial. Methods: The Stroke Hyperglycemia Insulin Network Effort (SHINE) trial (http://www.nett.umich.edu/nett/shine) is an ongoing NIH funded, multicenter, acute stroke trial. Two health systems that use EPIC electronic health record (EHR) developed automated screening (AS) alerts that notify the study team when potentially eligible patients meet the basic criteria for the study in real time. The procedures used are summarized in the figure. Our primary outcome measure was the number of alerts per month at each site. Results: From January 2014 through July 2014, a total of 331 patients (UM:259, UVA:72) were identified using AS algorithms; UM: 47.7, UVA: 10.3 per month. A total of 10 patients were eligible (UM:2, UVA:8) during this time period. Both UM patients were identified by the AS alert, along with four at UVA (transfers from other hospitals not captured).. Conclusions: While automated screening for acute stroke trials using the electronic health record is feasible, we found substantial implementation issues at two sites. At one site, less than 1% of alerts were for eligible patients and at the other site, half of actual eligible patients were missed. Despite challenges, the systems were useful to the research teams and helped study coordinators taking call from home identify cases which would be the highest yield to travel to the hospital to screen. EHR based acute stroke screening tools for research have both promise and a need for further refinement.