Responsiveness of five measures of arm function in acute stroke rehabilitation

Abstract

Objective: To determine the responsiveness of five arm function measures in people receiving acute inpatient stroke rehabilitation. Design: Inception cohort study. Setting: Comprehensive stroke unit providing early rehabilitation. Subjects: A total of 64 consecutively admitted stroke survivors with moderately severe disability (Modified Rankin Scale score median (interquartile range (IQR)): 4.0 (1.0)). Main measures: Responsiveness was analyzed by calculating effect size, standardized response mean and median-based effect size. Floor/ceiling effects were calculated as the percentage of participants scoring the lowest/highest possible scores. Results: Average length of stay and number of therapy days were 34 (SD = 27.9) and 12 (SD = 13.1), respectively. Box and Block Test and Functional Independence Measure–Self-Care showed the highest responsiveness with values in the moderate–large range (effect size = 1.09, standardized response mean = 1.07 and median-based effect size = 0.76; effect size = 0.94, standardized response mean = 1.04 and median-based effect size = 1.0). Responsiveness of Action Research Arm Test and Upper Limb–Motor Assessment Scale were moderate (effect size = 0.58, standardized response mean = 0.69 and median-based effect size = 0.59; effect size = 0.62, standardized response mean = 0.75 and median-based effect size = 0.67). For Manual Muscle Test, responsiveness was in the small–moderate range (effect size = 0.42, standardized response mean = 0.59 and median-based effect size = 0.5). Box and Block Test showed the largest floor effect on admission (28%), and Action Research Arm Test and Manual Muscle Test showed the largest ceiling effect on discharge (31%). Conclusion: These five measures varied in their ability to detect change with responsiveness ranging from the small to large range. Box and Block Test and Functional Independence Measure–Self-Care showed a greater ability to detect change; both demonstrated moderate–large responsiveness.