The Problem
Excess days in the hospital are multi-factorial. One factor affecting the length of stay is care team design. In 2021, the combinations of physicians, case managers, physical therapists, occupational therapists, and nurses were formed by chance. The assignment of care members to teams that worked optimally together was rarely considered when these teams were established. There were also too many distinct team members that each staff member had to work for to care for their group of patients. Consequently, care communication and handoffs suffered. Inefficiency lengthened patients’ stay in the hospital and decreased staff satisfaction.
To improve care coordination at Duke Raleigh, Hospital Medicine leaders (including lean transformation coaches) structured patient care teams and created a manual Microsoft Excel process supporting documentation of recent care teams, information about incoming faculty/staff, and formatting to support staff-to-patient and staff-to-team matching. While this advanced spreadsheet improved care coordination, aligning these departments’ patient care assignments remained a complex puzzle that required significant manual effort, especially to create it for new units and scale across the health system. They sought a more streamlined approach through automation.
Our Solution
In 2022, Hospital Medicine leaders at Duke Raleigh, including lean transformation coaches, structured patient care teams and created a manual Excel-based process to document care teams, staff availability, and new floor patients. This approach aimed to support optimal staff-to-patient and staff-to-team matching. However, the manual process was labor-intensive and needed to be more scalable across the health system. The Duke Institute for Health Innovation (DIHI) aimed to automate data entry for the Excel-based Scheduling Assistant Tool to streamline this process. DIHI used its Bed Watch solution to identify patient unit and bed locations in real time. By integrating real-time data pipelines, DIHI extracted physician and nurse scheduling data from QGenda and Symplr and treatment team relationships data from Epic APIs. This streamlined approach aimed to reduce time and errors in data entry and validation.
Impact
Next Steps
Across a health system, identifying real-time care team members and matching them to their patients is crucial for clinical decision support. For example, early warnings of maternal sepsis or hemorrhage should be sent to the first-call provider and nurse best able to respond. Now, DIHI can automatically send those alerts quickly and specifically. DIHI filled this critical need by successfully developing a real-time pipeline for extracting physician and nurse-patient assignments, scheduling data from QGenda and Symplr, and treatment team relationships from Epic APIs. Today, DIHI can target information signals to the appropriate care team members.
