Factor analysis validates the internal structure of the Cerebellar Neuropsychiatric Rating Scale Version 2 and the five domains of cerebellar neuropsychiatry.

Abstract

The Cerebellar Neuropsychiatric Rating Scale Version 2 (CNRS-2) measures affective symptoms in cerebellar disease patients across five postulated domains of attentional control, emotional control, autism spectrum, psychosis spectrum, and social skill set, each with overshoot and undershoot symptom subdomains. A data-driven approach is needed to test our <i>a priori</i> model and to further explore the internal structure of the scale. To explore the latent structure of the CNRS-2 and evaluate the five-domain and overshoot-undershoot structure of the scale items. CNRS-2 was administered to 279 cerebellar subjects. Data suitability for factor analysis was assessed. Eigenvalues and parallel analysis scree plots were evaluated for factor extraction. Exploratory factor analysis (EFA) was performed to identify an optimal factor solution. The EFA-derived solution and our <i>a priori</i> five-domain model were examined via confirmatory factor analysis (CFA). Fit indices, item loadings, and conceptual interpretability were compared. Bifactor analysis was performed on each of the five domains of the <i>a priori</i> model to evaluate whether variance in domain scoring was attributable to the construct measured by the domain versus two subfactors. CFA was also applied to each domain to test the validity of the overshoot-undershoot item groupings. We extracted 5, 6, and 7 factors based on eigenvalues and parallel analysis. EFA indicated that a six-factor solution best balanced statistical fit and theoretical interpretability. CFA on this EFA-derived solution and on our <i>a priori</i> five-domain model showed both to have acceptable statistical fit and internal consistency, but the five-domain model demonstrated stronger conceptual coherence. The bifactor analysis revealed variation in the interpretability of two subfactors within each of the domains, while CFA validated the overshoot-undershoot clustering of symptoms. Factor analysis of the items comprising CNRS-2 provided empirical support for the clinically derived conceptual framework that defines the five domains of cerebellar neuropsychiatry. Sub-analyses of the individual domains supported the overshoot-undershoot dichotomy. Convergence between our <i>a priori</i> domain model and the solutions derived from EFA revealed stable and interpretable symptom clusters. These findings underscore the presence of coherent, multidimensional interrelated neuropsychiatric constructs in cerebellar disease that can be identified and measured by CNRS-2.