Psychometric properties of the Rivermead Motor Assessment: its utility in stroke.

OBJECTIVE To investigate the psychometric properties of the Rivermead Motor Assessment by Rasch analysis and conventional statistics to improve its clinical utility. METHODS A total of 107 patients after stroke were evaluated using the Rivermead Motor Assessment and Functional Independence Measure (FIM). Scaling properties were assessed using Mokken scaling, internal construct validity using Rasch analysis, reliability using Cronbach's alpha and intra-class correlation coefficients, external construct validity through convergent validity with FIM, and responsiveness using the effect size and standardized response mean. RESULTS Cronbach's alpha and intra-class correlation coefficients for 3 sections of the Rivermead Motor Assessment were between 0.88 and 0.95. Mokken scaling showed appropriate Guttman patterns, but the hierarchical ordering of the items differed from that of the original. After removing 4 items of gross function, 1 of leg-trunk, and 4 of arm, all sections met Rasch model expectations. External construct validity was confirmed. Mean values of effect size and standardized response were 0.38-0.51 and 0.60-0.89, respectively. CONCLUSION The Rivermead Motor Assessment has been shown to be reliable and responsive. Guttman scaling is apparent, but not as originally defined. After removing some items, the scale satisfies the most stringent Rasch measurement criteria and can produce interval scaling for the assessment of motor function in stroke.