Reliability of gait speed measured by a timed walking test in patients one year after stroke

Objective: To assess the reliability of gait speed in late-stage stroke patients. Design: Test-retest reliability of three timed walks to 10 metres repeated during two assessments one week apart. Setting: The patient's home. Subjects: Twenty-two stroke patients with mobility problems more than one year after stroke. Main outcome measure: Gait speed measured in seconds taken to walk 10 metres. Statistical analysis: Intraclass correlations (ICCs) with 95% confidence interval (Cl) and the Bland and Altman method for assessing agreement by calculating the mean difference between measurements ((d) over bar), the 95% CI for (d) over bar; the standard deviation of the difference (SDdiff); a reliability coefficient and the 95% limits of agreement. Results: There was a trend for decreased times taken to walk 10 metres both within each assessment and between assessments. ICCs for within-assessment reliability were 0.95-0.99. The (d) over bar (SDdiff) for the second and third walks for assessment 1 was -1.00 (2.63) seconds and for assessment 2 was -0.70 (1.58) seconds. The reliability coefficient was 5.26 for assessment 1 and 3.17 for assessment 2. ICCs for between-assessment reliability were 0.87-0.88. The (d) over bar (SDdiff) for the comparison of the third walks at assessment 1 and assessment 2 was -0.90 (5.01) seconds. The reliability coefficient was 10.02 and the 95% limits of agreement were -10.92 to +9.12 seconds. Conclusion: Within-assessment gait speed measured at home is highly reliable. The between-assessment reliability of gait speed measurement is less reliable but comparable with other studies.