Comparing Behavioral and Psychological Symptom Structures on the Neuropsychiatric Inventory Questionnaire

• Published in: Psychological assessment 2023-06, Vol.35 (6), p.522-532
• Main Authors: González, David Andrés, Resch, Zachary J., Obolsky, Maximilian A., Soble, Jason R.
• Format: Article
• Language: English
• Subjects: Aged; Cognitive ability; Dementia; Dementia - diagnosis; Dementia - psychology; Factor Analysis, Statistical; Female; Human; Humans; Male; Measurement; Measurement Invariance; Neuropsychological Tests; Psychiatric Symptoms; Psychiatry; Psychometrics; Questionnaires; Reliability; Reproducibility of Results; Surveys and Questionnaires; Syndromes; Test Reliability; Test Validity
• ISSN: 1040-3590; 1939-134X; 1939-134X
• DOI: 10.1037/pas0001230

The Neuropsychiatric Inventory Questionnaire (NPI-Q) is a collateral-rated measure of behavioral and psychological symptoms commonly found in dementia (BPSD). Several factor structures have been published, but they have not been systematically compared. Furthermore, the possibility of hierarchical models or presence of measurement invariance around cognitive stage or dementia syndrome has not been previously evaluated. This study addressed these gaps with confirmatory factor analyses using a multicenter sample (n = 41,801; Mage = 71.4; 57% women; 79% White, 13% Black, 8% Hispanic; Meducation = 15.1) that was divided into exploratory, derivation, and holdover subsets for cross-validation. We found that a four-factor model had the best fit, with adequate reliability estimates, adequate τ-equivalence, and the least amount of measurement variance. Strict invariance across stage and syndrome was not supported, although there was adequate support for weaker restrictions (e.g., equal forms). Furthermore, all bifactor models had a significant increase in fit. In sum, the present study provides practical guidance on using NPI-Q factor-derived subscales and theoretical elaboration of BPSD's hierarchical and syndrome-variant structure. Public Significance Statement This study compared models for behavioral and psychological symptoms of dementia (BPSD) and found four factors to be recommendable. It provided practical information (i.e., reliability, summability) and assessed hierarchical extensions to discover that joint consideration of broad and narrow contributors best explains BPSD. It also determined that the model variability in literature is, at least partly, due to differences in the dementia stage and type among individuals studied.