Stability of three tests for unilateral neglect

Unilateral visual neglect, an attentional disorder, might show variability on repeated testing. This study investigated test-retest stability in elderly patients post-stroke, 85 with and 83 without neglect. Subjects repeated three common clinical tests for neglect within the hour; the Star Cancellation Test (SCT), Line Bisection (LB) and the Baking Tray Task (BTT). Data analysis indicated good to excellent test repeatability in subjects without neglect. For subjects with neglect, intraclass correlation analysis gave coefficients of .89, .97, and .87 for the SCT, LB, and BTT, respectively, indicating good to excellent agreement. However, analysis of this group, using the 95% limits of agreement, indicated poorer stability, with maximum test-retest differences of: 15 more or 11 less stars cancelled on the SCT, 3 cm for LB, and five "buns" on one side of the "tray" for the BTT. Limits of agreement analysis of sub-groups demonstrated better test-retest stability for the SCT, and a trend for this in the LB, in subjects with more severe neglect. Clinically, limits of agreement analysis is useful, providing indication of the maximum difference, in the units of the test, which may be expected on retest. We suggest that, if SCT, LB, or BTT are used as outcome measures, subjects with severe unilateral visual neglect are likely to show better stability on repeated testing. This may be especially relevant for single-case design when stability of baseline measurement is of particular importance.