A-1 Comparison of Consensus and Actuarial Diagnosis of MCI in Hispanic and non-Hispanic Samples

Abstract

Abstract Objective: The Jak/Bondi actuarial method has been found to be superior to consensus diagnosis in identifying MCI in some but not all populations and may not perform as well in diverse samples. In this study, diagnostic agreement between actuarial and clinical consensus methods was examined by ethnic group in the heavily Hispanic/Latinx Texas Alzheimer’s Research and Care Consortium (TARCC) cohort. Method: TARCC participants with consensus diagnoses of NC and MCI at baseline (n = 1011; 66.47% female; 49.75% Hispanic; Mage = 67.28; Medu = 13.21) were reclassified via actuarial criteria of two failed tests (>1 SD below normative [MOANS in TARCC] mean) in one cognitive domain OR one failed test in 3+ cognitive domains. Chi square analyses investigated diagnostic agreement and test fail rates by ethnicity. Results: Diagnostic agreement significantly differed between ethnic groups (Cramer’s V = 0.42). 57% of the Hispanic consensus-NC group were relabeled as actuarial-MCI while just 12% of the consensus-NC non-Hispanic sample were classified as actuarial-MCI. The non-Hispanic consensus-MCI sample was more likely to be reclassified as actuarial-NC (47% reclassified). Failure rates were significantly higher for the Hispanic group on 8 of 10 neuropsychological tests (Cramer’s V = 0.106–0.542). Conclusions: The Jak/Bondi MCI actuarial method could be prone to false-positive errors in Hispanic/Latinx samples, based on these findings in a significantly more diverse cohort than previously studied. Because the actuarial method relies on normed scores, its application may be more difficult in populations for which adequate norms are still being developed. Further research exploring use of “robust” norms or updated Hispanic/Latinx norms (e.g., NP-NUMBRS) when applying the Jak/Bondi method is needed.