Development of Predictive Informatics Tool Using Electronic Health Records to Inform Personalized Evidence-Based Pressure Injury Management for Veterans with Spinal Cord Injury

Abstract

ABSTRACT Background Pressure injuries (PrI) are serious complications for many with spinal cord injury (SCI), significantly burdening health care systems, in particular the Veterans Health Administration. Clinical practice guidelines (CPG) provide recommendations. However, many risk factors span multiple domains. Effective prioritization of CPG recommendations has been identified as a need. Bioinformatics facilitates clinical decision support for complex challenges. The Veteran’s Administration Informatics and Computing Infrastructure provides access to electronic health record (EHR) data for all Veterans Health Administration health care encounters. The overall study objective was to expand our prototype structural model of environmental, social, and clinical factors and develop the foundation for resource which will provide weighted systemic insight into PrI risk in veterans with SCI. Methods The SCI PrI Resource (SCI-PIR) includes three integrated modules: (1) the SCIPUDSphere multidomain database of veterans’ EHR data extracted from October 2010 to September 2015 for ICD-9-CM coding consistency together with tissue health profiles, (2) the Spinal Cord Injury Pressure Ulcer and Deep Tissue Injury Ontology (SCIPUDO) developed from the cohort’s free text clinical note (Text Integration Utility) notes, and (3) the clinical user interface for direct SCI-PIR query. Results The SCI-PIR contains relevant EHR data for a study cohort of 36,626 veterans with SCI, representing 10% to 14% of the U.S. population with SCI. Extracted datasets include SCI diagnostics, demographics, comorbidities, rurality, medications, and laboratory tests. Many terminology variations for non-coded input data were found. SCIPUDO facilitates robust information extraction from over six million Text Integration Utility notes annually for the study cohort. Visual widgets in the clinical user interface can be directly populated with SCIPUDO terms, allowing patient-specific query construction. Conclusion The SCI-PIR contains valuable clinical data based on CPG-identified risk factors, providing a basis for personalized PrI risk management following SCI. Understanding the relative impact of risk factors supports PrI management for veterans with SCI. Personalized interactive programs can enhance best practices by decreasing both initial PrI formation and readmission rates due to PrI recurrence for veterans with SCI.