Optimizing care transitions through the post-acute care (PAC) continuum is a leading priority for many healthcare systems. Even highly integrated systems experience challenges with ensuring patients move seamlessly through the care continuum. UPMC Centers for Rehabilitation Services (CRS) found that <15% of all patients discharged from UPMC’s inpatient rehabilitation network (i.e., a patient population with a presumably high therapy-need) transition to CRS outpatient services. CRS leadership wanted a better understanding of which patients were vulnerable to poor care transitions, where breakdowns occurred in the continuum, and which factors potentially influenced transitions. 

CRS leadership identified individuals with stroke as their highest priority for examining PAC pathways for two reasons. First, deficits in physical and/or cognitive function are ubiquitous and often undetected after stroke. Second, early and intense therapy is crucial for reducing poor outcomes (e.g., readmission). Thus, it was imperative that CRS focus on optimizing rehab care pathways for these individuals. Our LeaRRn team used a mixed-methods approach to characterize these PAC pathways.

To begin this exploration, we employed qualitative methods. We assembled an interdisciplinary advisory team to help understand shared priorities, setting-specific characteristics, inter-related processes (i.e. processes in one setting that influence processes in another setting), and patient characteristics perceived to be influential to care transitions. Advisory team members were selected from the various settings across the PAC continuum. Members included health system administrators (i.e., medical and rehabilitation directors), physicians, and therapists who had expertise in the stroke population. Direct observations and front-line staff interviews were also conducted in UPMC’s various PAC settings, which helped the research team understand setting-specific workflows and perspectives.

Guided by our qualitative findings, we then identified health system data sources that were necessary to construct episodes of care (e.g., admission dates/discharge dates, readmissions) and to examine potentially influential factors for care transitions (e.g., patient health characteristics, patient demographics, care processes, setting-specific outcomes). Data sources generally fell in three categories: billing, quality reporting, and electronic health records. Two different data stewardship groups managed access and sharing for inpatient and outpatient settings, respectively. Thus, coordination between each settings’ administrators, data stewards, and regulatory officials was critical.

This case illustrates how systems science helped direct and refine our approach to understanding care transitions. Data analysis is ongoing; we are currently constructing episodes of care from hospitalization data through 6-months post-hospital discharge. This work will inform future research focused on improving care pathways.