Clinical and functional determinants of appropriate rehabilitation referrals after stroke: a single-center retrospective cohort study

Background: Rehabilitation is currently the best available treatment for post-stroke disability. There is, however, great variability in the proportion of patients accessing rehabilitation across high-income countries suggesting that factors not explained by facilities availability or guidelines diversity may intervene in decision-making. Objectives: To evaluate which factors are associated with appropriate post-stroke rehabilitation referrals in a tertiary stroke unit setting. Methods: Retrospective single-center cohort study including patients admitted to the Stroke Unit of the "Santa Maria della Misericordia" University Hospital, Udine (IT) from January 1st to December 31st, 2019. Information regarding stroke severity (National Institute of Health Stroke Scale), functional assessment (modified Rankin scale [mRS] and Barthel index [BI]), length of hospital stay, and rehabilitation pathway was collected. Outcome was defined as referral to the appropriate rehabilitation pathway. Appropriateness was assessed comparing patient clinical information at discharge against local criteria for intensive vs. extensive rehabilitation. A mixed-linear effect model was built to explore NIHSS, mRS, and BI variation over time. Multivariable logistic regression was used to estimate the adjusted-odds ratio (OR) and 95% confidence interval (CI 95%) of appropriate assignment to rehabilitation pathways. Results: 288 patients were included in the study (age 73.1 years, males 57.9%) and in 75.7%, the rehabilitation pathway assignment was appropriate. NIHSS at discharge was lower compared to admission but no effect of rehabilitation assignment was evident, while mRS scores at discharge and at three months were 2.6 (CI 95% 2.2; 3.0) and 2.1 (CI 95% 1.8; 2.5) higher compared to admission (p < 0.0001). Rehabilitation assignment effect on mRS was time dependent, resulting in an additional - 0.6 (CI 95% - 1.0; - 0.2) lowering at discharge for those appropriately assigned (p = 0.003), with a trend for significance at three months (p = 0.08). BI score was higher at discharge (p < 0.0001), and appropriate assignment was associated with higher scores (p = 0.01). Multivariate analysis showed that the OR of appropriate rehabilitation pathway assignment were reduced by higher mRS (0.60 [CI 95% 0.48; 0.76], p < 0.0001) and increased by higher NIHSS (1.11 [CI 95% 1.04; 1.19], p = 0.001) scores at discharge. The latter finding might be explained by the rehabilitation assessment focus on post-stroke motor symptoms captured by NIHSS. Conclusions: Higher mRS and lower NIHSS levels at discharge were independent predictors for inappropriate rehabilitation assignment after stroke in our cohort. These findings may reflect a therapeutic bias toward patients with higher post-stroke disability in a rehabilitation framework heavily tilted on post-stroke motor symptoms.