Abstract P241: Why Do We Expect Clinicians to Perform the Work of Computers? Developing an Automated Stroke Registry

Abstract

Background: Stroke registries are established with the primary goal of improving health care quality and performance improvement through increased adherence to evidenced based guideline recommendations. As a Joint Commission certified organization, the Comprehensive Stroke Center uses the American Heart Association’s Get With The Guidelines® (GWTG) stroke patient management tool to collect performance improvement data on patients admitted with a principle diagnosis of stroke. Data abstraction is laborious and time consuming which requires manual entry by multiple clinicians. Therefore, we developed an automated internal stroke database to improve efficiency and augment accuracy. Methods: We collaborated with software engineers and EPIC report writers to create an innovative stroke database with the capability to extract data from an Electronic Health Record (EHR) for upload to GWTG. The clinicians identified data elements for automation in the EHR. Based on final ICD-10 coding the report writer retrieved end-user data, which were organized into tables and sent to the software engineer. These variables were processed to match GWTG specifications and then presented to clinicians in the user interface for validation. Next, a CSV file was generated and uploaded to GWTG, where the remaining data was entered. Results: The case completion time was reduced from 45 minutes to one hour to an average of 25 minutes (median 23 minutes), thus achieving our primary goal. Due to the variance in the number of required data elements based on diagnosis and patient specific variables, we aggregated the data elements, automated versus required, and calculated an automation rate of 54%. Approximately half of the 414 fields required in GWTG have the potential for automation, and we continue to strive toward this goal. Due to the complexity of GWTG questions and the way data is captured in the EHR, some data elements require manual entry. Conclusions: An automated stroke registry has the potential to enhance efficiency and limit the opportunity for human error. By minimizing the clinicians’ data entry tasks to those of higher-level data abstraction, there is more time for performance improvement.