A comparison of data reduction approaches for the identification of non-cognitive symptom subsyndromes in Alzheimer's disease, based on Neuropsychiatric Inventory scores

Background Within Alzheimer's disease, attempts to perform data reduction on the non-cognitive symptoms comprising the Neuropsychiatric Inventory (NPI) have been inconsistent. Here, most studies used principal component analysis (PCA), despite reports that exploratory factor analysis (EFA) estimates loadings more accurately. Objective We compared how well different data reduction approaches derived NPI models in a discovery dataset (Alzheimer's Disease Neuroimaging Initiative, n = 460), based on confirmatory factor analysis (CFA) in two independent datasets (Brains for Dementia Research, n = 337 and Memento, n = 211). Specifically, we assessed PCA, maximum likelihood (ML) EFA, principal axis (PA) EFA, and a PCA extension that adjusts for zero inflation. For each method, we additionally compared how different approaches to determine the number of factors/components affected CFA fit indices. Methods EFA and standard PCA were performed using PSYCH, the zero-inflation adjusted PCA using zibppca, and CFA was performed in Lavaan (all R packages). Results In both confirmatory datasets, PA EFA yielded the closest fitting model (Tucker Lewis Index > 0.95). This model comprised [agitation and irritability], [depression and anxiety], [delusions and hallucinations], and [elation and disinhibition]. We also demonstrated that selecting the number of latent constructs to retain using parallel analysis based on null factors, as opposed to component eigenvalues, was necessary to identify this best fitting model. Conclusions PA EFA allowed the identification of an acceptable CFA verified NPI subsyndrome structure. Our findings also indicate that previous attempts to identify NPI subsyndrome structure might have been hindered by extracting too few factors/components.

Authors

Institutions

Publication Details

Journal
Journal of Alzheimer s Disease
Published
2026-11-21
DOI
https://doi.org/10.1177/13872877261507355
Primary Topic
Dementia and Cognitive Impairment Research
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

A comparison of data reduction approaches for the identification of non-cognitive symptom subsyndromes in Alzheimer's disease, based on Neuropsychiatric Inventory scores

Patrick G. Kehoe, Peter Holmans, Peter P Henley, Rebecca Sims et al.
Journal of Alzheimer s Disease
Dementia and Cognitive Impairment Research
article

A comparison of data reduction approaches for the identification of non-cognitive symptom subsyndromes in Alzheimer's disease, based on Neuropsychiatric Inventory scores

Patrick G. Kehoe, Peter Holmans, Peter P Henley, Rebecca Sims, Lindsey Isla Sinclair, Alzheimer’s Disease Neuroimaging Initiative
article en

Abstract

Background Within Alzheimer's disease, attempts to perform data reduction on the non-cognitive symptoms comprising the Neuropsychiatric Inventory (NPI) have been inconsistent. Here, most studies used principal component analysis (PCA), despite reports that exploratory factor analysis (EFA) estimates loadings more accurately. Objective We compared how well different data reduction approaches derived NPI models in a discovery dataset (Alzheimer's Disease Neuroimaging Initiative, n = 460), based on confirmatory factor analysis (CFA) in two independent datasets (Brains for Dementia Research, n = 337 and Memento, n = 211). Specifically, we assessed PCA, maximum likelihood (ML) EFA, principal axis (PA) EFA, and a PCA extension that adjusts for zero inflation. For each method, we additionally compared how different approaches to determine the number of factors/components affected CFA fit indices. Methods EFA and standard PCA were performed using PSYCH, the zero-inflation adjusted PCA using zibppca, and CFA was performed in Lavaan (all R packages). Results In both confirmatory datasets, PA EFA yielded the closest fitting model (Tucker Lewis Index > 0.95). This model comprised [agitation and irritability], [depression and anxiety], [delusions and hallucinations], and [elation and disinhibition]. We also demonstrated that selecting the number of latent constructs to retain using parallel analysis based on null factors, as opposed to component eigenvalues, was necessary to identify this best fitting model. Conclusions PA EFA allowed the identification of an acceptable CFA verified NPI subsyndrome structure. Our findings also indicate that previous attempts to identify NPI subsyndrome structure might have been hindered by extracting too few factors/components.

Journal of Alzheimer s Disease
At Bristol (GB), University of Bristol (GB), University of Southampton (GB), Cardiff University (GB)
Dementia and Cognitive Impairment Research
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon