A-178 Characterizing Memory Measurement in Cognitively Normal Individuals Using Latent State Trait Modeling of Neuropsychological Tests. (28th August 2020)

• Record Type: Journal Article
• Authors: W, Goette; J, Schaffert; A, Carlew; H, Rossetti; L, Lacritz
• Is Part Of: Archives of clinical neuropsychology. Volume 35:Number 6(2020)
• Page Start: 972
• Page End: 973
• Publication Date: 2020-08-28
• Subjects: Clinical neuropsychology -- Periodicals; 616.805
• DOI: 10.1093/arclin/acaa068.178
• Languages: English
• ISSNs: 0887-6177

Abstract

Abstract: Objective: Characterize trait and state measurement of memory in a longitudinal, cognitively normal (CN) sample. Method: Data were taken from 515 participants (81.17% white; Mage = 64.78, SD = 8.20; Medu = 13.04, SD = 4.39; 73.40% female) deemed CN for 5 consecutive visits in the Texas Alzheimer's Research and Care Consortium. The memory trait was measured by the CERAD (total learning and delayed recall) and WMS-III Logical Memory (LM) and Visual Reproduction (VR) I and II subtests. Multiple-indicator growth models were used as the primary analysis. Results: Assumption screening revealed the memory trait was characterized by three domain factors: list learning (CERAD learning and recall), story memory (LM I and II), and visual memory (VR I and II). Generalized second-order growth models demonstrated best overall fit. Scores demonstrated a positive latent growth process such that trait memory scores actually increased over time. Consistency and specificity coefficients suggest less situation-specific error occurs as individuals are tested over time with the first evaluation having the largest portion of situation-specific measurement error (Supporting Figure). Conclusions: Measurement accuracy of memory in CN individuals improved over repeated administrations. Accurate identification of cases with normal memory in memory disorders research may require serial assessments to account for (a) improved measurement accuracy of the trait after multiple assessments and (b) reducing the influence of state-specific sources of error that are greatest at the first assessment time. Further research is needed to determine whether the observed positive growth in trait scores is unique to CN samples and whether measurement accuracy changes in cognitive domains other than memory.