A-49 Construct Validity and Predictive Utility of Cognitive Intraindividual Variability Varies Based on Ethnic Group

Abstract

Abstract Objective Compare the construct validity and predictive utility of cognitive intraindividual variability (IIV) in a sample of community dwelling Hispanic and non-Hispanic older adults. Methods The present study included annual data from 651 older adult control participants (Hispanic = 293; Non-Hispanic = 358) enrolled in the Texas Alzheimer’s Research and Care Consortium for at least 5 years. Mean composite z-scores were calculated for attention, language, memory and executive function domains. IIV-across domains was calculated as the standard deviation between these composite scores, while IIV-within domain scores were computed as the absolute difference between tests in each domain. Results At baseline, Non-Hispanic individuals obtained significantly higher composite scores in each domain than their Hispanic counterparts. They also showed significantly greater variability across and within domains, with the exception of IIV-within the language domain which was significantly larger in the Hispanic group. IIV-across domains was driven primarily by IIV-within the executive function domain in the non-Hispanic cohort and by IIV-within the language domain in the Hispanic cohort. In both groups, the addition of IIV-within and -across cognitive domains at baseline significantly improved prediction of cognitive status (MMSE) after 5 years above and beyond demographic characteristics, genetic risk and vascular integrity. However, IIV-across domains was the strongest predictor in the non-Hispanic group, while IIV-within the attention domain was the strongest predictor in the Hispanic group. Conclusions Findings suggest that, while IIV-across domains is a promising preclinical biomarker of subsequent cognitive decline, its construct validity and predictive utility varies based on ethnic group.