Reliability of Cognitive Measures in Individuals With a Chronic Spinal Cord Injury

Following spinal cord injury (SCI), up to 64% of individuals experience cognitive deficit. However, the reliability of commonly used neuropsychological tests is currently unknown in this population.

To evaluate the test-retest reliability of cognitive measures in individuals with SCI.

Cross-sectional study.

Vancouver General Hospital.

Individuals with a chronic (>2 years) SCI (n = 22).

Across three visits (separated by ~16 days), 22 participants with chronic SCI completed a neuropsychological battery evaluating memory (Rey Auditory-Verbal Learning Test [RAVLT]), attention/concentration/psychomotor speed (Digit Span Task, Stroop Test), and executive function (Trail Making Test A&B, Symbol Digit Modalities Test, Controlled Oral Word Association Test). Coefficients of variation (CV_intra ) and intraclass correlation coefficients (ICCs) were calculated to determine the reliability of each test between visits. Linear regressions were performed to assess the associations between variability (CV_intra ) and participant characteristics, such as age or highest education level attained. Repeated-measures, one-way analysis of variance (ANOVA) was conducted to determine any significant practice effects, and smallest real differences (SRDs) were calculated.

Repeated scores on aforementioned neuropsychological tests.

ICCs ranged from 0.77 to 0.93, with the exception of RAVLT recognition score (ICC = 0.27). Age showed a moderate association with CV_intra in RAVLT interference recall scores (r = 0.43, P = .047), but was not a confounding factor for other measures. Education was not associated with CV_intra . Significant practice effects were noted for most of the cognitive tests assessed.

Other than the RAVLT recognition score, these cognitive measures demonstrated good-to-excellent reliability. Although this is encouraging, test-retest variability should be considered when interpreting the efficacy of various cognitive training strategies to mitigate cognitive decline in this population. Thus, the SRD values presented herein will allow researchers and clinicians to identify "true" changes in cognitive function with repeated testing.

III.