ABSTRACT
Introduction
The high comorbidity between depression and anxiety challenges traditional nosological models. In order to better reflect this overlap, dimensional approaches seek to clarify whether these symptoms reflect a single underlying construct of general distress or a more complex multidimensionality. The aim was to examine the underlying structure of depression and anxiety symptoms, measured by the PHQ-9 and GAD-7, by estimating and comparing three factor models: one-factor, two-factor correlated, and bifactor.
Methods
Data from 1704 primary care (PC) patients from the PsicAP clinical trial were analysed. Dimensionality was assessed using hierarchical Exploratory Graph Analysis (EGA), and models were estimated using Exploratory Structural Equation Modelling (ESEM). Model fit was compared using the χ
2, CFI, TLI, RMSEA, and SRMR indices.
Results
The bifactor model offered the most acceptable comparative fit to the data (CFI = 0.956; TLI = 0.928; RMSEA = 0.076; SRMR = 0.028). Bifactor indices revealed a relevant but not dominant general factor (ECV = 0.40), indicating it does not account for sufficient variance to justify an essentially unidimensional interpretation. Furthermore, the specific factors of depression and anxiety emerged as well-defined constructs (H > 0.75) with modest reliable specific variance (ωHS: 0.22 and 0.27, respectively).
Discussion
The findings suggest a hierarchical structure where a general factor of distress coexists with specific factors. This suggests the potential utility of considering a multi-level perspective in clinical assessment, accounting for both shared distress and specific symptom profiles.