Comparison of word list recall and paragraph recall to distinguish dementia severity in the Uniform Data Set (UDS): Results from a diverse cohort

Abstract Background Prior research reported word list is more sensitive to changes early in the Alzheimer’s disease process. However, there is limited investigation into ethnic level differences between memory tests. This project examines: ethnic level (Black/African American) and testing language differences (Spanish) between word list and paragraph recall to distinguish dementia severity, and added benefit of using both tests. Method Baseline UDS and site‐specific supplemental participant data collected at the Icahn School of Medicine at Mount Sinai’s Alzheimer’s Disease Research Center (ISMMS ADRC) were utilized to compare the efficacy of paragraph recall (Logical Memory and Craft Story) versus a word list test (CERAD) both together and alone in differentiating dementia severity as measured by the Clinical Dementia Rating (CDR) Scale. Composite scores for each memory test were established by calculating Z‐scores and taking the average. Logistic regression models were employed to distinguish between CDR 0 versus 0.5 and 0.5 versus 1. Receiver Operating Characteristic‐Area Under the Curves (ROC‐AUCs) were used as an indicator for overall performance in differentiating CDR. Subsample analyses were performed for White, Black/African American (AA), English, and Spanish language testing. Result 1,319 participants who self‐identified as White (n = 788, 60%), Black/AA (n = 225, 17%), or Hispanic (n = 306, 23%). Results reveal high discriminability and no statistically significant difference between word list and paragraph recall in distinguishing between CDR 0 vs 0.5 (range AUC = 0.79 to 0.84; p>.05) or 0.5 vs 1 (range AUC = 0.81 to 0.87; p>.05). Discriminability was greater when combining tests in the overall group (0 vs 0.5 range AUC = 0.82 to 0.87; p<.01, 0.5 vs 1 range AUC = 0.81 to 0.93; p<.05) but this not always the case for the Black/AA group (0 vs 0.5 range AUC = 0.80 to 0.91; p>.05, 0.5 vs 1 range AUC = 0.83 to 0.93; p>.05) or Spanish speakers (0 vs 0.5 range AUC = 0.74 to 0.86; p>.05, 0.5 vs 1 range AUC = 0.85 to 0.90; p>.05). Conclusion Our findings confirm that there is utility in employing multiple memory tests to differentiate between CDR levels. However, these findings do not extend to all ethnically diverse or Spanish speaking participants. Researchers may want to consider burden of extra testing time when assessing the cognition of diverse participants.