Evaluating memory testing to distinguish dementia severity among White, Black, and Spanish-speaking individuals in the Uniform Data Set (UDS).

Abstract

Little work has compared the effectiveness of using multiple types of memory tests alone or in combination to distinguish dementia severity in diverse research cohorts including Black individuals and Spanish speakers. Here we evaluate word list and paragraph recall tests to distinguish cognitively normal, mild cognitively impaired, and those with Alzheimer's disease in diverse cohorts. Using Uniform Data Set (UDS) and site-specific supplemental data, logistic regression models and receiver operating characteristic-area under the curve were used to compare paragraph recall versus word list in differentiating among Clinical Dementia Rating (CDR) scale level. Results reveal high discriminability for all groups and no difference between either test in distinguishing between CDR levels. Combining tests improved discriminability for the whole group but did not for Black individuals or Spanish speakers. Our findings indicate that using multiple memory tests may not improve differentiation between cognitive impairment levels for diverse cohorts. The burden of added testing may be a barrier for maximizing inclusion of under-represented groups in research.