Association between hospital-level factors and early rehabilitation initiation in patients with acute ischemic stroke: a multilevel analysis using a Japanese administrative database from a rural prefecture

BACKGROUND: Early rehabilitation after acute ischemic stroke improves functional outcomes and is recommended by clinical guidelines. However, initiation of early rehabilitation varies substantially across hospitals, and the contribution of hospital-level factors remains unclear. OBJECTIVE: To examine the association between hospital-level factors and early rehabilitation initiation, adjusting for patient-level characteristics. METHODS: We conducted a multicenter cross-sectional study using the Japanese Diagnosis Procedure Combination database from 43 hospitals in Yamagata Prefecture between 2017 and 2022. Patients admitted within 3 days of ischemic stroke onset who received inpatient rehabilitation were included. Early rehabilitation was defined as initiation of physical, occupational, or speech-language-hearing therapy within 2 days of admission. Multilevel Poisson regression models were constructed with patients nested within hospitals. Median rate ratio (MRR) was calculated to quantify hospital-level variation. RESULTS: A total of 17,194 patients from 19 hospitals were analyzed; 9,533 patients (55.4%) received early rehabilitation. At the hospital-level factors, provision of weekend rehabilitation services was significantly associated with early initiation (incidence rate ratio, 1.65; 95% credible interval, 1.21-2.25). Therapist staffing levels, annual stroke volume, and availability of thrombolytic therapy were not associated with early rehabilitation. The hospital-level MRR decreased from 1.47 in the null model to 1.35 after inclusion of weekend rehabilitation services. CONCLUSIONS: The provision of weekend rehabilitation was associated with the initiation of early rehabilitation for ischemic stroke patients. Our findings suggest that variations in the initiation of early rehabilitation for acute ischemic stroke are largely driven by organizational systems rather than patient characteristics.