Associating Clinical Decision Support Use with Clinical Outcomes in a Visual Analytics Display

Project Summary Abstract Clinical decision supportCDSis a process for enhancing health related decisions and actions with pertinentorganized clinical knowledge and patient information to improve healthcare deliveryFor exampledoctors may have trouble remembering to order all guideline recommended care for sepsisCDS delivered via an order set in the electronic health recordEHRcan simplify this process and reduce mortality by bundling the recommended diagnostic and therapeutic orders togetherMore generallyCDS has the potential to improve patient outcomes by facilitating adherence to evidence based practicesEBPsHoweverit may fail to reach this potential becausethe CDS tool is underutilizedthe user may not follow the recommended action from the CDSthe recommended action may not lead to the EBPand orthe EBP may not translate to the expected outcome in a novel populationHealthcare organizations need an efficientrigorousand scalable process evaluation method to diagnose when and why CDS is not leading to the intended improvementsOur long term objective is to empower organizations to efficiently incorporate scientific knowledge into high value clinical care through incrementaldata driven improvements of CDS informed by understanding of the relationships between CDS useprocess measuresand patient outcomesIn Aimof this proposalwe will establish the technical feasibility of associating CDS use patterns with process and outcome metrics using EHR log data through a proof of concept demonstration focused on inpatient treatment of pediatric migraineBoth Children s Healthcare of AtlantaCHOAand Children s Hospital of PhiladelphiaCHOPhave well defined patient cohortslocal clinical guidelinesorder setsand outcome metrics for the care of inpatient pediatric migraine patientsPhrase Healthhas already developed a commercial CDS analytics product used byinstitutions that organizes EHR log data into an intuitive display that provides insight into how all an organization s alerts and order sets are usedWe will leverage these strengths to create a standard database schema for patient cohorts that links to Phrase Healths existing data model for order set use patternsWe will then verify the accuracy of this association through manual chart reviewWhile we use inpatient migraine as a modelthe architecture will generalize across clinical use casesIn Aimwe will use user centered design best practices to develop a visual analytics dashboard for rapid identification of CDS improvement targetsWe will then evaluate the effectiveness of the dashboard through scenario based summative testingin which we measure how accurately users identify CDS improvement targets with a functional dashboard based on real CDS data extracted in AimAt the end of this projectwe will have created and validated a novelscalable CDS process evaluation tool with demonstrated technical feasibility at two institutionsThis will advance healthcare organizationsability to improve patient outcomes through CDS and prepare us for a Phaseapplication focused on implementation Project Narrative Clinical decision supportCDShas the potential to improve patient outcomes by facilitating adherence to evidence based practicesTo achieve this goalhealthcare organizations need an efficient and rigorous process evaluation method to diagnose when and why CDS is not leading to the intended improvementsWe are creating and validating a novelscalable CDS process evaluation tool that will advance healthcare organizationsability to convert knowledge to high quality care through CDS