ABSTRACTBackground: The aim of this study was to investigate the clustering of symptoms on the Neuropsychiatric Inventory (NPI) and the relative prevalence of symptom clusters in mild and moderate Alzheimer's disease (AD).Methods: Confirmatory factor analysis (CFA) was performed on the polychoric correlation matrix of 12 NPI items (scored yes or no) using the weighted least squares with mean and variance adjustment (WLSMV) estimator in Mplus (N = 224 community-dwelling Chinese persons with mild/moderate AD). Severity of AD was determined by Clinical Dementia Rating scores. The relative model fit of three competing measurement models (also known as factor structures) was tested using a modified χ2 difference test. Prevalence rates across mild and moderate stages were compared using χ2 tests. Furthermore, the measurement model of choice was cross-validated in an independent sample of 181 community-dwelling persons with dementia.Results: CFA supported a four-factor model, namely behavioral problems (agitation/aggressiveness, disinhibition, irritability, and aberrant motor behavior), psychosis (delusions and hallucinations), mood disturbance (depression, anxiety, sleep, appetite, and apathy), and euphoria (a stand-alone item that is equivalent to the factor). The most prevalent symptom clusters were behavioral (72%) and mood (69%) disturbances, followed by psychosis (45%). Euphoria was rare (6%). All syndromes were more prevalent in moderate than in mild stage, except for euphoria. In addition, the four-factor model was replicated in the cross-validation sample.Conclusions: The four syndromes provide a parsimonious conceptualization of neuropsychiatric symptoms corresponding to clinical observations and neurochemical changes of the disease. The independent replication in the second sample supports generalization of the four-factor model in Chinese persons with AD.
The NINCDS-ADRDA neuropsychological criteria for Alzheimer's disease: a confirmatory factor analysis of single versus multi-factor models
