Intra- and inter-rater reliability of the Sollerman hand function test in patients with chronic stroke

Purpose. To examine whether the Sollerman hand function test is reliable in a test-retest situation in patients with chronic stroke. Method. Three independent examiners observed each patient at three experimental sessions; two days in week I (short-term test-retest) and one day in week 4 (long-term test-retest). A total of 24 patients with chronic stroke (mean age; 59.7 years, mean time since stroke onset 29.6 months) participated. The examiners simultaneously assessed the patients' ability to perform 20 subtests. Both ordinal data (generalized kappa) and total sum scores (Spearman's rank correlation coefficient (Spearman's rho), intra class correlation coefficient (ICC2,1) and mean differences) were used in the statistical analysis. Results. There was agreement (kappa >= 0.4) between the examiners for 15 out of the 20 subtests. Using total sum scores, the agreement within the examiners, both short- and long-term, was higher than 0.96 (for Spearman's rho and ICC, respectively). The mean differences were 0.29-1.0/80 points within each examiner. Agreement between the examiners at each session was higher than 0.96 (Spearman's rho) and 0.92 (ICC), respectively. Systematic differences (p < 0.05) were, however, found between examiners A and B/C for all sessions. Conclusions. The Sollerman hand function test seems to be a reliable test in patients with chronic stroke, but we recommend that the same examiner evaluates a patient's hand function pre- and post-treatment.