Advanced Psychometric Interpretation of the Spanish-language NIH Toolbox Cognition Battery: Developing Diagnostic Algorithms for Cognitive Impairment and Interpreting Reliable Change

The American Latinx community represents roughly a fifth of the U.S. population and is rapidly growing. This community is often underserved by the U.S. medical system and has equivalent or higher risk compared to non-Latinx adults for medical conditions that can adversely affect cognitive functioning (e.g., diabetes, HIV/AIDS, dementia). This creates a population-specific need for neuropsychological assessment, but clinicians have a dearth of representative normative data for Latinx patients, especially those who communicate primarily in Spanish. The use of inappropriate normative data with Spanish-speaking Latinx patients can lead to higher risk of misdiagnosing cognitive impairment. To improve assessment with Latinx patients, clinicians require representative normative samples, demographic corrections for normative scores, methods for identifying cognitive impairment that consider premorbid functioning, and methods for interpreting clinically meaningful change at retest. The National Institutes of Health Toolbox Cognition Battery (NIHTB-CB) is a newly developed test battery designed to assess neuropsychological functioning across the lifespan, with available demographic corrections (e.g., gender, education, race/ethnicity) for scores based on nationally representative normative data for both English- and Spanish-speakers. With research supporting its reliability, construct validity, and clinical utility, the NIHTB-CB could enhance cultural neuropsychological research and clinical practice, but there are no resources for operationalizing cognitive impairment or interpreting reliable change on the Spanish-language NIHTB-CB. Our research program has focused on refining neuropsychological assessment practices by preparing resources to (a) inform the diagnosis of cognitive impairment based on multivariate base rates of low and high scores, and (b) define clinically meaningful change at retest based on reliable change indices. Multivariate base rates quantify how frequently scores below or above certain percentiles (e.g., .16th or .84th) occur among healthy individuals administered multiple neuropsychological tests. Both low and high scores are common and change in frequency with estimated intelligence and educational attainment. For average functioning individuals, obtaining one or more low scores may be normal; and for high functioning individuals, obtaining few or no high scores may be abnormal. We have begun to combine low and high score base rates to identify patterns of test performances that occur rarely within normative data, producing flexible diagnostic algorithms that operationalize cognitive impairment differently based on estimated premorbid functioning or education. For example, on the English-language NIHTB-CB, 16% of adults with superior crystallized ability obtain no scores .75th percentile and 9% of college-educated adults obtain 3 scores .16th percentile. Both the presence of low scores and the absence of high scores may indicate cognitive impairment in high-functioning people. Further, the preparation of reliable change indices, which define cutoffs for interpreting change on neuropsychological tests, can be used in clinical practice to identify whether scores have declined, improved, or remained stable over time. By leveraging pre-existing publicly available normative data for the Spanish-language NIHTB-CB (N=408 adults), our proposed research project expands on our previous work to improve neuropsychological practice with Spanish-speaking Latinx patients by pursuing the aims listed below. Specific Aim 1. To develop flexible diagnostic algorithms for cognitive impairment based on multivariate base rates of low and high scores, identifying patterns of performances that occur rarely among healthy adults with different levels of premorbid functioning and educational backgrounds. Specific Aim 2. To prepare reliable change indices for individual and composite cognitive scores from the Spanish-language NIHTB-CB, identifying the change required in demographic-corrected scores to indicate statistically reliable improvement or decline in cognitive performance. Accomplishing these aims will result in clinically translatable outcomes that can refine the psychometric interpretation of the Spanish-language NIHTB-CB in clinical practice, including (a) published tables on multivariate base rates of low and high scores for the test battery, (b) flexible diagnostic algorithms for operationalizing cognitive impairment in Spanish-speaking patients, and (c) reliable change indices for detecting meaningful change at re-evaluation of Spanish-speaking patients.