Purpose: To assess the test-retest reliability of a range of outcome measures in stroke patients.
Method: Twenty-two patients > 1 year post-stroke were tested twice at an interval of 1 week using the Barthel Index (BI); the Rivermead Mobility Index (RMI); the Nottingham Extended Activities of Daily Living Scale (NEADL); and the Frenchay Activities Index (FAI). The mean difference (bias) and reliability coefficient (random error) were calculated for the total scores. Percentage agreement and the kappa coefficient were used to analyse individual items.
Results: The mean differences and reliability coefficients were BI 0.4 +/- 2.0, RMI 0.3 +/- 2.2, the NEADL 0.6 +/- 5.6, FAI -0.6 +/- 7.1. There was little bias between assessments. The performance of the BI and RMI were better with lower random error. The NEADL and FAI did not perform as well having larger random error components. Percentage agreements were generally high especially for the BI (>75%) and RMI (>85%), but there was considerable variation in the kappa coefficients.
Conclusion: Measurement of basic activities of daily living and mobility as measured by the BI and RMI is reliable post-stroke. Measurements used to assess extended activities of daily living were less reliable in this study.