Designing and Implementing an Electronic Health Record Based Intervention to Reduce Hospital-Acquired Infections at BWH: Considerations and Lessons Learned

Abstract

Purpose: Nosocomial C. difficile infection is the most common hospital-acquired infection in the United States, and has implications for clinical care, quality assurance, and healthcare expenditure. The increase in molecular testing for C. difficile infection, which cannot distinguish true infection from asymptomatic colonization, may be partly responsible for overdiagnosis and overtreatment of C. difficile. Consequently, health care providers are responsible for deciding which patients should be tested for C. difficile infection. We sought to design and pilot a clinical decision support tool that would make use of the electronic health record to assist clinicians in making those decisions at Brigham and Women’s Hospital. Methods: We designed and implemented an intervention with the goal of reducing inappropriate C. difficile testing (based on nationally accepted testing guidelines). Once per day, the “operator” (an individual from the Department of Quality and Safety) runs an EPIC report that locates all admitted patients who have had C. difficile testing ordered but not collected. The operator then manually ascertains certain data from the patient chart (laxative use, recent testing, formed stool) that would define the test as inappropriate. If a contraindication is found, the operator pages the responding clinician to discuss C. difficile testing. The operator then follows up on the patient (via EPIC) to determine the diagnosis and clinical outcome. Results: This project was piloted from 1/2/2017 to 1/27/2017 at BWH. In total, 8 patients were identified and intervened upon. Reasons for intervening included recent laxative use (3), recently documented test result (1), and lack of diarrhea per provider notes (4). Of the 8 patients, 2 had C. difficile testing cancelled and 6 proceeded with testing; of the 6 tests conducted, 2 were positive and 4 negative. No patients experienced complications of C. difficile infection at 4-week follow- up. The number of patient cases was too low for any significant statistical analysis, and the pilot was subsequently discontinued. Discussion: We successfully designed and piloted an EHR-based clinical decision support tool aimed at reducing inappropriate C. difficile testing. Though the project was not continued beyond the 4-week pilot, there were many operational and quality lessons gleamed from the experience. In particular, this project highlighted the importance of operational thoughtfulness and robust data measures when designing a quality improvement initiative, while balancing the priorities of various project stakeholders, including the institution, the department, and those directly affected by the intervention.