Marijn Mulder1, Rinske H Nijland2, Ingrid G van de Port3, Erwin E van Wegen4, Gert Kwakkel1. 1. Department of Rehabilitation Medicine, Amsterdam Movement Sciences, Amsterdam UMC, Location VU University Medical Center, Amsterdam, Netherlands; Department of Neurorehabilitation, Amsterdam Rehabilitation Research Center, Reade, Amsterdam, Netherlands. 2. Department of Neurorehabilitation, Amsterdam Rehabilitation Research Center, Reade, Amsterdam, Netherlands. Electronic address: r.nijland@reade.nl. 3. Revant Rehabilitation Centers, Breda, Netherlands. 4. Department of Rehabilitation Medicine, Amsterdam Movement Sciences, Amsterdam UMC, Location VU University Medical Center, Amsterdam, Netherlands.
Abstract
OBJECTIVE: To classify patients with stroke into subgroups based on their characteristics at the moment of discharge from inpatient rehabilitation in order to predict community ambulation outcome 6 months later. DESIGN: Prospective cohort study with a baseline measurement at discharge from inpatient care and final outcome determined after 6 months. SETTING: Community. PARTICIPANTS: A cohort of patients (N=243) with stroke, referred for outpatient physical therapy, after completing inpatient rehabilitation in The Netherlands. INTERVENTIONS: Not applicable. MAIN OUTCOME MEASURES: A classification model was developed using Classification And Regression Tree (CART) analysis. Final outcome was determined using the community ambulation questionnaire. Potential baseline predictors included patient demographics, stroke characteristics, use of assistive devices, comfortable gait speed, balance, strength, motivation, falls efficacy, anxiety, and depression. RESULTS: The CART model accurately predicted independent community ambulation in 181 of 193 patients with stroke, based on a comfortable gait speed at discharge of 0.5 meters per second or faster. In contrast, 27 of 50 patients with gait speeds below 0.5 meters per second were correctly predicted to become noncommunity walkers. CONCLUSIONS: We show that comfortable gait speed is a key factor in the prognosis of community ambulation outcome. The CART model may support clinicians in organizing community services at the moment of discharge from inpatient care.
OBJECTIVE: To classify patients with stroke into subgroups based on their characteristics at the moment of discharge from inpatient rehabilitation in order to predict community ambulation outcome 6 months later. DESIGN: Prospective cohort study with a baseline measurement at discharge from inpatient care and final outcome determined after 6 months. SETTING: Community. PARTICIPANTS: A cohort of patients (N=243) with stroke, referred for outpatient physical therapy, after completing inpatient rehabilitation in The Netherlands. INTERVENTIONS: Not applicable. MAIN OUTCOME MEASURES: A classification model was developed using Classification And Regression Tree (CART) analysis. Final outcome was determined using the community ambulation questionnaire. Potential baseline predictors included patient demographics, stroke characteristics, use of assistive devices, comfortable gait speed, balance, strength, motivation, falls efficacy, anxiety, and depression. RESULTS: The CART model accurately predicted independent community ambulation in 181 of 193 patients with stroke, based on a comfortable gait speed at discharge of 0.5 meters per second or faster. In contrast, 27 of 50 patients with gait speeds below 0.5 meters per second were correctly predicted to become noncommunity walkers. CONCLUSIONS: We show that comfortable gait speed is a key factor in the prognosis of community ambulation outcome. The CART model may support clinicians in organizing community services at the moment of discharge from inpatient care.