Six‐Month Consistency of NIH Toolbox Cognition Battery Scores in Healthy Older Black Women with Subjective Cognitive Decline

Background Interest in the NIH Toolbox Cognition Battery (NIHTB‐CB) for the clinical assessment of older adults is burgeoning. Longitudinal neuropsychological assessments are the most sensitive way to establish cognitive decline and dementia risk of older adults. The psychometric property of measurement invariance is essential to valid interpretations of change. However, few studies have evaluated the stability of NIHTB‐CB scores in a sample of cognitively normal older adults with subjective cognitive decline (SCD). This study aims to analyze measurement consistency of NIHTB‐CB performance over 6 months. Method N = 29 cognitively normal, older Black adult women (65‐81 years of age) with SCD were administered the NIHTB‐CB at baseline and six‐month follow‐up. Diagnostic status was determined via consensus conference following NACC UDS guidelines. We evaluated the mean change by matched‐pairs t‐tests, and measurement consistency by two‐way mixed effects models of intraclass correlations coefficients and Bland‐Altman plots. Uncorrected standard scores and fully corrected T scores were analyzed from NIHTB‐CB Cognition Composites of Total Cognition, Fluid Cognition, and Crystallized Cognition. Result There were no significant mean differences from baseline to follow‐up for corrected and uncorrected Total Cognition Scores. The uncorrected Fluid Score significantly increased by 2.5 points at follow‐up (t(28) = 2.18, p = .04), and the corrected Crystallized Score increased by 1.8 points at follow‐up (t(28) = ‐2.25, p = .03). The uncorrected Crystallized Score and the corrected Fluid Score did not differ across assessments. Consistency for the uncorrected and corrected Total scores was good, (ICC(3,1) = 0.83 and 0.80, respectively). Similarly, consistency for uncorrected and corrected Fluid Scores was also good (ICC(3,1) = 0.77 and 0.78). Consistency was excellent for uncorrected Crystallized Scores (ICC(3,1) = .92), but good for corrected scores (ICC(3,1) = .87). Discrepancy scores (baseline vs follow‐up) derived from uncorrected scores for the three composites were unrelated to age and education. Bland‐Altman plots of uncorrected scores for the Composite Scores were consistent with the ICC analyses, showing good agreement between measures. Conclusion NIHTB‐CB corrected and uncorrected composite scores have good longitudinal psychometric properties in a sample of older Black adult women with SCD, a relatively under‐investigated examinee sample. These findings also