Assessing quality of life in an obesity observational study: A Structural Equation Modeling Approach

peer reviewedPurpose: Quality of life (QOL) of a patient is usually computed as the (weighted) sum of items and analysed by means of multiple regressions to evaluate its relationships with various measured factors. The aim of the present study was to compare results derived under classical statistical method with those obtained under more appropriate statistical techniques for QOL. Methods: Analyses were applied to data from 4155 subjects participated in 2012 in a community based sample study in the French speaking part of Belgium and which completed a web-based questionnaire on their weight-related experience. Confirmatory factor analysis (CFA) and structural equation modeling (SEM) were carried out to derive QOL and to test direct/indirect effects of body mass index (BMI), age, body image discrepancy (BID), latent socio-economic (SOCIO) and latent subjective-norm (SN). Results: No major differences were found under both SEM and the product of coefficients approach using SAS PROCESS macro developed by Hayes. Significant direct and indirect effects on physical and psychological dimensions of QOL were found for age, BMI and SOCIO while significant direct effects were found for BID and SN (p < 0.0001). Factor loadings were found to be significantly different according to gender (p < 0.0001). Conclusions: BID and SN are partially mediators on the relationships between BMI and QOL. The study also confirms the role of SOCIO on the (un)observable variables included in the model. However, the large sample size provided significant tests with small effect size and couldn’t highlight pertinent differences between both methods