Pietro Giorgio Lovaglio1, Emiliano Monzani. 1. CRISP and Department of Quantitative Methods, Faculty of Statistics, University of Bicocca-Milan, V. Sarca, 202, 20143, Milan, Italy. piergiorgio.lovaglio@unimib.it
Abstract
PURPOSE: To explore the internal structure of the health of the nation outcome scales (HoNOS-12), proposing a shorter one-dimensional version for routine use in community-oriented mental heath services. METHODS: A validation study involving four mental health departments, located in the province of Milan (Italy). Eligible patients were outpatients and residential inpatients rated on three occasions during the year 2009, with a range of mental illnesses and diagnoses. Methodologically, we use both exploratory factor analysis (EFA) with holdout validation and Rasch approaches and parallel analysis. RESULTS: EFA, Rasch analysis and parallel analysis demonstrate a large violation of unidimensionality. Both EFA (training sample) and Rasch analyses yield convergent results, generating the same unidimensional abbreviated version of the HoNOS-12, resulting in a six-item scale (HoNOS-6) which demonstrates unidimensionality, good item fit, a solid factor structure (strong loadings and communalities) and acceptable model fit, evaluated using confirmatory factor analysis on a validation sample. CONCLUSIONS: The HoNOS-12 does not measure a single, underlying construct of mental health status. Nevertheless, the instrument can be utilized in a reduced version (HoNOS-6), as a clinically acceptable outcome scale (measuring self-perceived clinical and social needs for community support, rather than global mental disorder) for routine use in a community setting population.
PURPOSE: To explore the internal structure of the health of the nation outcome scales (HoNOS-12), proposing a shorter one-dimensional version for routine use in community-oriented mental heath services. METHODS: A validation study involving four mental health departments, located in the province of Milan (Italy). Eligible patients were outpatients and residential inpatients rated on three occasions during the year 2009, with a range of mental illnesses and diagnoses. Methodologically, we use both exploratory factor analysis (EFA) with holdout validation and Rasch approaches and parallel analysis. RESULTS:EFA, Rasch analysis and parallel analysis demonstrate a large violation of unidimensionality. Both EFA (training sample) and Rasch analyses yield convergent results, generating the same unidimensional abbreviated version of the HoNOS-12, resulting in a six-item scale (HoNOS-6) which demonstrates unidimensionality, good item fit, a solid factor structure (strong loadings and communalities) and acceptable model fit, evaluated using confirmatory factor analysis on a validation sample. CONCLUSIONS: The HoNOS-12 does not measure a single, underlying construct of mental health status. Nevertheless, the instrument can be utilized in a reduced version (HoNOS-6), as a clinically acceptable outcome scale (measuring self-perceived clinical and social needs for community support, rather than global mental disorder) for routine use in a community setting population.
Authors: Conal Twomey; A Matthew Prina; David S Baldwin; Jayati Das-Munshi; David Kingdon; Leonardo Koeser; Martin J Prince; Robert Stewart; Alex D Tulloch; Alarcos Cieza Journal: PLoS One Date: 2016-11-30 Impact factor: 3.240