Item hierarchy-based analysis of the Rivermead Mobility Index resulted in improved interpretation and enabled faster scoring in patients undergoing rehabilitation after stroke

Abstract

OBJECTIVE: To enable improved interpretation of the total score and faster scoring of the Rivermead Mobility Index (RMI) by studying item ordering or hierarchy and formulating start-and-stop rules in patients after stroke. DESIGN: Cohort study. SETTING: Rehabilitation center in the Netherlands; stroke rehabilitation units and the community in the United Kingdom. PARTICIPANTS: Item hierarchy of the RMI was studied in an initial group of patients (n=620; mean age +/- SD, 69.2+/-12.5y; 297 [48%] men; 304 [49%] left hemisphere lesion, and 269 [43%] right hemisphere lesion), and the adequacy of the item hierarchy-based start-and-stop rules was checked in a second group of patients (n=237; mean age +/- SD, 60.0+/-11.3y; 139 [59%] men; 103 [44%] left hemisphere lesion, and 93 [39%] right hemisphere lesion) undergoing rehabilitation after stroke. INTERVENTIONS: Not applicable. MAIN OUTCOME MEASURES: Mokken scale analysis was used to investigate the fit of the double monotonicity model, indicating hierarchical item ordering. The percentages of patients with a difference between the RMI total score and the scores based on the start-and-stop rules were calculated to check the adequacy of these rules. RESULTS: The RMI had good fit of the double monotonicity model (coefficient H(T)=.87). The interpretation of the total score improved. Item hierarchy-based start-and-stop rules were formulated. The percentages of patients with a difference between the RMI total score and the score based on the recommended start-and-stop rules were 3% and 5%, respectively. Ten of the original 15 items had to be scored after applying the start-and-stop rules. CONCLUSIONS: Item hierarchy was established, enabling improved interpretation and faster scoring of the RMI.