B-182: Evaluation of Updated and Prior Stroop Color and Word Test Norms in a Diverse Veteran Older Adult Sample
Abstract
Objective: Evaluation of norms applied to diverse populations supports ethical decision-making in diagnosis and treatment (Bush et al., 2018). We compared the application of Stroop Color and Word Test (Stroop) Normative Update (Golden et al., 2022) to previous norms (Golden & Freshwater, 2002; MOANS-Ivnik et al., 1996; MOAANS-Lucas et al., 2005) in a diverse veteran older adult sample. Method/Data Selection: Veterans who passed performance validity testing and completed Stroop (n=170; M_age=68.1; 50% with neurocognitive disorder) were included. The sample was diverse (54.1% White/Caucasian, 32.9% Hispanic/Latino, 12.9% African American/Black). Logistic regression (LR) and receiver operator characteristic (ROC) analyses examined diagnostic classification accuracy of Stroop Color-Word (CW) and Interference T-scores overall and within subgroups. Results/Data Synthesis: LR models (overall) for Golden-2022 and MOANS CW norms had the highest classification accuracy (71.8%). MOANS LR model was significant (χ2[1]=39.6, p<.001) and explained 28% of the variance (Nagelkerke-R2) in diagnosis (vs. 27% for Golden-2022 norms). For Hispanic/Latino participants, Golden-2022 and MOANS norms correctly classified 73.2%, accounting for 25% and 24% of the variance, respectively; Golden-2022 better classified those with neurocognitive disorders. For African American/Black participants, Golden-2022 correctly classified 72.7% (24% variance explained). Interference ROC analyses (AUC=.50) and LR found low classification accuracy. Specificity for CW/Interference norms was unacceptably low (highest CW=.32; Golden-2002). At best, CW sensitivity levels for T=40 was .68 (Golden-2002). MOAANS had low sensitivity/specificity. Conclusions: While classification accuracy was best with updated, Golden-2022 Stroop-CW norms, Golden-2002 CW had the best sensitivity/specificity, though none achieved adequate specificity for predicting diagnosis broadly. Use of MOAANS norms is not recommended based on findings.