Learning From the Adoption of a Readmission Clinical Decision Support Tool: Group Model Building Approach.

Background: Computerized clinical decision support (CDS) has the potential to improve patient outcomes by offering evidence-based guidance at the point of care-enhancing guideline adherence and diagnostic accuracy-and supporting system-level outcomes by enabling predictive analytics for more efficient resource planning. Prior work has identified factors that affect adoption, such as clinicians' expectations of usefulness, ease of use, alignment with workflows, and resources to support utilization. However, CDS adoption is not static and changes according to dynamic systems of behaviors and workflows, requiring a deeper understanding of how evolving conditions affect implementation and outcomes. Objective: To explore the dynamic factors influencing CDS adoption, we examined the implementation of the "Unplanned readmission model version 1," developed by Epic Medical Records System, at Duke University Health System, using group model building and system dynamics modeling. Methods: We first conducted group model-building workshops with staff (case managers, physical and occupational therapists, hospitalist faculty physicians, and resident physicians) who participate in decisions about discharging patients. Study team members guided participants to identify and connect variables in causal loop diagrams. We coded workshop transcripts in software designed for system dynamics analysis to identify themes, aggregated them into a causal loop diagram, and reviewed them with participants to converge on a common model. A team member applied equations to the pathways and tested data to simulate conditions leading to full, limited, or no adoption of a tool. Results: We identified key balancing loops driven by external pressure (eg, Centers for Medicare & Medicaid Services penalties) that motivated initial adoption and reinforcing loops based on perceived internal benefits to sustain use. While institutional incentives led to early training and tool use, efforts declined due to staff turnover, competing priorities (eg, COVID-19), and workflow changes. Reinforcing loops emerged when staff described clinical utility, such as improved discharge planning and team communication. However, staff also suggested that these loops were often weak due to difficulty linking the use of the tool to outcomes in real time. Simulation modeling showed that while strong external pressure and rapid training led to initial success, interest in using the tool waned as workflows improved and readmission rates approached Centers for Medicare & Medicaid Services goals. When conflicting priorities were introduced, adoption stalled earlier, and fewer staff were trained. In contrast, when internal motivation was strengthened by reducing the amount of evidence needed to perceive success, individual interest remained high even as institutional attention declined, sustaining tool use and further reducing readmissions. Conclusions: External pressure to improve can be a strong motivator for initial adoption, but in the face of conflicting demands for attention, it may fall short of sustained long-term tool use. Tools are more likely to have extensive and sustained use when those using the tools can perceive internal benefits.