Ethnic Differences Between Hispanics and Non-Hispanic Whites in Neuropsychiatric Symptoms Predict Conversion to Mild Cognitive Impairment

Abstract

The aim of the study is to ascertain the neuropsychiatric symptoms (NPS) subtypes significantly influencing progression to mild cognitive impairment (MCI) by ethnicity. In this retrospective cohort study, we included 386 cognitively normal individuals participating in the longitudinal Texas Alzheimer’s Research and Care Consortium between February 2007 and August 2014. The primary outcome was time to incident MCI. Data driven NPS subtypes at baseline were identified and the effects of these subtypes on the outcome were obtained for Hispanic and non-Hispanic ethnic cohorts and summarized with a hazard ratio (HR). Three NPS subtypes were identified and internally validated: psychomotor apathy factor (including agitation, irritability, apathy), affective mood factor (including depression, anxiety), and physical behavior factor (including nighttime behavior, eating/appetite disturbances). In adjusted analysis, a psychomotor apathy score of NPS was the best predictor for MCI (HR = 2.19, p = 0.037) among non-Hispanics whereas physical behavior score was the most predictive of MCI (HR = 2.55, p = 0.029) among Hispanics. A high score of affective mood factor also tended to increase the risk of MCI (HR = 2.09, p = 0.06) in Hispanics. Progression from normal cognition to MCI was differentially predicted by NPS subtypes in Hispanics and non-Hispanic whites. These data may inform the allocation of efforts for monitoring individuals at-risk of MCI.

Gender Differences in Neuropsychiatric Symptoms Among Community-Dwelling Mexican Americans Aged 80 and Older

“…Among non-Hispanic White individuals, apathy was the most predictive NPS of conversion to MCI, while for Hispanic individuals, physical behaviors were most predictive. 16 Another study found that among individuals with Alzheimer's disease, Hispanics more often exhibited apathy and anxiety, while non-Hispanic Whites more often exhibited apathy and depression. 17 Hispanics also had more NPS compared to non-Hispanic Whites.…”

Comparing behavioral and psychological symptom structures on the Neuropsychiatric Inventory Questionnaire.

“…In addition to statistical heterogeneity, many studies used variants in NPI-Q scoring procedures, with the most common being a dichotomized presence/absence (de Vito et al, 2018; Sayegh & Knight, 2014; Siafarikas et al, 2018; Thakur et al, 2021) and 4-point ordinal severity (Feghali et al, 2021; Seidl & Massman, 2016; Siafarikas et al, 2021; Trzepacz et al, 2013). Many studies also eliminated items because they had low rates of endorsement in their samples, high cross-loadings, and/or failed to load adequately on a single factor (de Vito et al, 2018; Sayegh & Knight, 2014; Seidl & Massman, 2016; Thakur et al, 2021; Trzepacz et al, 2013). These modifications to the test item structure further obscure cross-study comparison.…”

“…Prior EFA studies have used different extraction, selection, and rotation techniques that can contribute to poor replicability and misapproximation of the NPI-Q's true structure. For extraction/estimation, many studies used principal component analysis (PCA; Feghali et al, 2021;Siafarikas et al, 2018Siafarikas et al, , 2021Thakur et al, 2021;Trzepacz et al, 2013), which may overestimate factor loadings and is less likely to reproduce in new samples. PCA also relies on assumptions about normal indicator distribution that are often not met and may, thereby, contribute to misestimation and limited reproducibility (Brown, 2015).…”

“…In addition to statistical heterogeneity, many studies used variants in NPI-Q scoring procedures, with the most common being a dichotomized presence/absence (de Vito et al, 2018;Sayegh & Knight, 2014;Siafarikas et al, 2018;Thakur et al, 2021) and 4point ordinal severity (Feghali et al, 2021;Seidl & Massman, 2016;Siafarikas et al, 2021;Trzepacz et al, 2013). Many studies also This document is copyrighted by the American Psychological Association or one of its allied publishers.…”

“…Despite its wide dissemination, there is disarray in the literature regarding what subscales or symptom domains the NPI-Q captures, which impedes consistent and accurate use of the scale by clinicians and researchers. Most concerning, extant studies examining principal component and factor analyses of the NPI-Q have revealed a strikingly wide range of single-level models, ranging from 2 to 10 factors (de Vito et al, 2018;Feghali et al, 2021;Sayegh & Knight, 2014;Seidl & Massman, 2016;Siafarikas et al, 2018Siafarikas et al, , 2021Thakur et al, 2021;Trzepacz et al, 2013). The wide variability in factor structure may be related to differences in statistical methodology, measure modifications, and the sampled populations.…”