Responsiveness of the rehabilitation activities profile and the Barthel index

The goal of this study was to compare the responsiveness for clinically meaningful change over time of a newly designed functional status scale, the Rehabilitation Activities Profile (RAP), with more frequently used Barthel Index (BI). Four techniques for the quantification of responsiveness were utilized: effect sizes, p-values, t-statistics and ROC curves. The patient's return home was chosen as external criterion. An inception cohort of stroke patients was followed during 26 weeks. All patients still hospitalized on the 14th day after the stroke were included. The functional assessments took place at 2, 3, 4, 8, 12, and 26 weeks after stroke. The patients were visited at the hospital, home, nursing home, or rehabilitation center. Of the 125 patients included in the study, 18 patients died during the observation period, 2 patients were lost to follow-up, and 1 patient refused to cooperate after 12 weeks. After 26 weeks, 104 patients remained for analysis. Three time periods were discerned: 2 to 12 weeks (early response), 12 to 26 weeks (late response), and 2 to 26 weeks after stroke (overall response). The effect sizes of the RAP were consistently higher on all three time periods than those of the BI. The p-value of the overall response mean change score of the RAP appeared to discriminate between patients returning home and those not returning home, whereas the BI failed on this point (p = 0.004 vs. 0.496). Using t-statistics, the RAP showed a higher efficiency in expressing change on all time periods (relative efficiency = 1.42, 1.77, and 1.43, respectively). The receiver operating characteristic surface area of the RAP score was higher than the area of the BI score (0.74 and 0.59, respectively for the early response period). In conclusion, all results seemed to indicate that the RAP is more responsive than the BI when returning home is chosen as an external criterion