Abstract
OBJECTIVE: With the increased use of computer-based tests in clinical and research settings, assessing retest reliability and reliable change of NIH Toolbox-Cognition Battery (NIHTB-CB) and Cogstate Brief Battery (Cogstate) is essential. Previous studies used mostly White samples, but Black/African Americans (B/AAs) must be included in this research to ensure reliability. METHOD: Participants were B/AA consensus-confirmed healthy controls (HCs) (n = 49) or mild cognitive impairment (MCI) (n = 34) adults 60-85 years that completed NIHTB-CB and Cogstate for laptop at two timepoints within 4 months. Intraclass correlations, the Bland-Altman method, t-tests, and the Pearson correlation coefficient were used. Cut scores indicating reliable change provided. RESULTS: NIHTB-CB composite reliability ranged from .81 to .93 (95% CIs [.37-.96]). The Fluid Composite demonstrated a significant difference between timepoints and was less consistent than the Crystallized Composite. Subtests were less consistent for MCIs (ICCs = .01-.89, CIs [-1.00-.95]) than for HCs (ICCs = .69-.93, CIs [.46-.92]). A moderate correlation was found for MCIs between timepoints and performance on the Total Composite (r = -.40, p = .03), Fluid Composite (r = -.38, p = .03), and Pattern Comparison Processing Speed (r = -.47, p = .006).On Cogstate, HCs had lower reliability (ICCs = .47-.76, CIs [.05-.86]) than MCIs (ICCs = .65-.89, CIs [.29-.95]). Identification reaction time significantly improved between testing timepoints across samples. CONCLUSIONS: The NIHTB-CB and Cogstate for laptop show promise for use in research with B/AAs and were reasonably stable up to 4 months. Still, differences were found between those with MCI and HCs. It is recommended that race and cognitive status be considered when using these measures.