Quantifying Cognitive Bias in Post-acute Decision Making After Stroke

After a stroke, nearly 80% of individuals live with a degree of chronic functional impairment. Advances in acute stroke treatment have resulted in a significant decline in mortality, with more people requiring post-acute rehabilitation after surviving a stroke. Indeed, individuals with stroke are among some of the highest, most costly users of post-acute care. The incidence of stroke and its associated costs is projected to surge as the population ages, creating an urgent need to maximize care transitions. Rehabilitation is a critical component of post-stroke care and is associated with functional gains, reduced hospital readmission, and decreased long-term morbidity. The recommendation for post-acute rehabilitation setting is initiated by the physical and occupational therapist during acute hospitalization. This initial recommendation then sets a larger process in motion that includes a Physiatry consultation and insurance approval. Prior research has examined factors associated with discharge to post-acute setting after stroke (e.g., inpatient rehabilitation, skilled nursing). What is less understood, however, is how the decision for initial post-acute rehabilitation setting is made and to what clinical or sociodemographic factors therapists anchor their decision. There are no standardized criteria to guide this decision, introducing significant noise and potential for decision bias among therapists. Additionally, it is unknown to what extent this subjectivity introduces disparities in access to post-acute rehabilitation post-stroke. The purpose of this application is to evaluate the decision making process for post-acute stroke rehabilitation among acute therapists. This retrospective cohort study will include approximately 30,000 individuals, admitted for a stroke at a Penn Medicine hospital (January 1, 2017 - June 1, 2024), who discharged to either an inpatient rehabilitation or a skilled nursing facility. Using electronic health record data, we will evaluate therapist decision making, the potential for cognitive bias in this decision making, and whether or not disparities exists in access to post-acute rehabilitation among clinically similar subgroups. The specific aims are to (1) describe the recommendation for post-acute rehabilitation post-stroke; (2) quantify the presence of availability and time of day bias in post-acute recommendation post-stroke; and (3) to evaluate disparities in post-acute rehabilitation recommendation. Our long-term goal is to improve the quality of post-acute care decision making by developing clinical decision support tools that help ensure equitable care for all adults post-stroke. This study will also establish methodology for translating this work to other populations who are high users of post-acute rehabilitation in future grant applications.