6- and 12-month test-retest score reliability and practice effects of the NIH Toolbox Cognition Battery in older Black adults with subjective cognitive decline

Objective: The NIH Toolbox Cognition Battery (NIHTB-CB) is increasingly used for cognitive assessment. However, its test-retest reliability estimates over clinically meaningful intervals are understudied in older Black adults. This study examined the reliability and practice effects of NIHTB-CB composite and subtest scores over 6- and 12-month intervals in older Black adults with subjective cognitive decline. Methods: Participants were 44 community-dwelling older Black adults with subjective cognitive decline (Mage = 71.4 years, SD = 4.9; Meducation = 16.1 years, SD = 2.1; 93% women) who completed baseline and 6-month testing and N = 29 participants (93% women) completed 12-month testing. Intraclass correlation coefficients evaluated the reliability of uncorrected standard scores and fully corrected T-scores. Paired-samples t tests with Benjamini-Hochberg correction assessed mean-level change, and Bland-Altman plots evaluated agreement. Results: At 6 months, composite-level reliability was good for Total Cognition (ICC = .76–.78) and Crystallized Cognition scores (ICC = .82–.87), and moderate for Fluid Cognition scores (ICC = .69–.72). Subtest-level reliability was more variable, with lower reliability for List Sorting and Picture Sequence Memory scores. At 12 months, significant mean improvements were observed across several composites and subtest fully corrected scores. These changes demonstrated medium paired-samples effect sizes (|dz| = 0.51–0.58). 12-month follow-up reliability estimates remained strong for composite scores, particularly for Crystallized Cognition. Conclusion: NIHTB-CB composite scores showed generally good longitudinal reliability over 6–12 months among older Black adults with subjective cognitive decline. Variable subtest reliability and possible 12-month practice effects warrant cautious interpretation when monitoring change over time.