Comparison of the psychometric characteristics of the functional independence measure, 5 item Barthel index, and 10 item Barthel index in patients with stroke.

Abstract

To compare the reliability, validity, and responsiveness of the motor subscale of the functional independence measure (FIM), the original 10 item Barthel index (BI), and the 5 item short form BI (BI-5) in inpatients with stroke receiving rehabilitation. 118 inpatients with stroke at a rehabilitation unit participated in the study. The patients were tested with the FIM motor subscale and original BI at admission to the rehabilitation ward and before discharge from the hospital. The distribution, internal consistency, concurrent validity, and responsiveness of each measure were examined. The BI and FIM motor subscale showed acceptable distribution, high internal consistency (alpha coefficient > or = 0.84), high concurrent validity (Spearman's correlation coefficient, r(s) > or = 0.92, intraclass correlation coefficient (ICC) > or = 0.83), and high responsiveness (standardised response mean > or = 1.2, p < 0.001). The BI-5 exhibited a notable floor effect at admission but this was not found at discharge. The BI-5 showed acceptable internal consistency at admission and discharge (alpha coefficient > or = 0.71). The concurrent validity of the BI-5 was poor to fair at admission (r(s) = 0.74, ICC < or = 0.55) but was good at discharge (r(s) > or = 0.92, ICC > or = 0.74). It is noted that the responsiveness of the BI-5 was as high as that of the BI and the FIM motor subscale. The results showed that the BI and FIM motor subscale had very acceptable and similar psychometric characteristics. The BI-5 appeared to have limited discriminative ability at admission, particularly for patients with severe disability; otherwise the BI-5 had very adequate psychometric properties. These results may provide information useful in the selection of activities of daily living measures for both clinicians and researchers.