Distinct behavioral and psychological symptoms (BPS; e.g., disinhibition, delusions) frequently manifest in many forms of dementia and may hold significance for improved diagnostic accuracy and effective intervention/management. BPS are commonly measured by an informant-based questionnaire, the Neuropsychiatric Inventory Questionnaire (NPI-Q). A recent study (Gonzalez et al., 2023) established a robust 4-factor structure for the NPI-Q among a large sample of individuals with heterogenous cognitive impairment but did not investigate the degree to which the NPI-Q factors may vary across different severities of cognitive impairment/dementia. Quantifying subscales differences across among dementia stages is important to inform criterion validity and possible clinical utility, as some symptoms may become more or less prominent at different levels of impairment. Thus, the primary aim of this study was to evaluate the four NPI-Q factors across a full range of cognitive stages (i.e., unimpaired to severe dementia).

The study analyzed data from the National Alzheimer’s Coordinating Center (NACC) database and included 45,033 participants who had collateral informants, who completed the NPI-Q during their first visit to an Alzheimer’s Disease Research Center (ADRC). All-cause dementia was diagnosed and staged with the Clinical Dementia Rating Scale (CDR) Dementia Staging Instrument using the O’Bryant staging system derived from the CDR sum of boxes, which resulted in five study groups: 1) Cognitively Unimpaired; 2) Mild Cognitive Impairment; 3) Mild Dementia; 4) Moderate Dementia, and 5) Severe Dementia. The components of the 4-factor NPI-Q model (i.e., psychosis, mood, behavioral activation, and behavioral suppression/somatic disturbance) were then compared between CDR-established cognitive impairment severity groups via Analyses of Variance (ANOVAs) to determine if differences in BPS emerged based on severity of dementia.

The ANOVA revealed that the behavioral suppression/somatic disturbance factor had the largest differences across cognitive impairment stages (p < .001, η_p² = .217), followed by the behavioral activation (p < .001, η_p² = .162), psychosis (p < .001, η_p² = .136), and mood p < .001, η_p² = .129) factors. Across all four factors, a similar pattern emerged whereby the cognitively unimpaired group demonstrated the lowest mean and variability, indicating more NPI-Q symptom variability as dementia severity increased.

The NPI-Q subscales appear to function somewhat differently across stages of dementia severity, with the factor capturing somatic disturbance symptoms being the most responsive to impairment stages, and the factor capturing mood symptoms being the least specific and responsive to impairment stage. Future studies should further implement these subscales correlate with neuropsychological testing that can assist in patient conceptualization across racial/ethnically diverse populations to further understand cultural differences and similarities given that culture may play a role in how collateral informants may interpret/respond to NPI-Q items.