Measurement and description of underlying dimensions of comorbid mental disorders using Factor Mixture Models: results of the ESEMeD project

作者：Almansa Josue; Vermunt Jeroen K; Forero Carlos G; Vilagut Gemma; De Graaf Ron; De Girolamo Giovanni; Alonso Jordi^*

来源：International Journal of Methods in Psychiatric Research, 2011, 20(2): 116-133.

DOI：10.1002/mpr.334

• internalizing disorders
• externalizing disorders
• measurement invariance
• mental comorbidity
• latent classes

Epidemiological studies on mental health and mental comorbidity are usually based on prevalences and correlations between disorders, or some other form of bivariate clustering of disorders. In this paper, we propose a Factor Mixture Model (FMM) methodology based on conceptual models aiming to measure and summarize distinctive disorder information in the internalizing and externalizing dimensions. This methodology includes explicit modelling of subpopulations with and without 12 month disorders ("ill" and "healthy") by means of latent classes, as well as assessment of model invariance and estimation of dimensional scores. We applied this methodology with an internalizing/externalizing two-factor model, to a representative sample gathered in the European Study of the Epidemiology of Mental Disorders (ESEMeD) study - which includes 8796 individuals from six countries, and used the CIDI 3.0 instrument for disorder assessment. Results revealed that southern European countries have significantly higher mental health levels concerning internalizing/externalizing disorders than central countries; males suffered more externalizing disorders than women did, and conversely, internalizing disorders were more frequent in women. Differences in mental-health level between socio-demographic groups were due to different proportions of healthy and ill individuals and, noticeably, to the ameliorating influence of marital status on severity. An advantage of latent model-based scores is that the inclusion of additional mental-health dimensional information - other than diagnostic data - allows for greater precision within a target range of scores.

• 出版日期2011-6