Introduction. Comorbidity remains one of the most challenging issues in mental health research and clinical practice. Traditional latent variable models often assume hidden common causes, overlooking the dynamic interplay between symptoms across disorders. Network analysis offers an alternative framework, conceptualizing psychopathology as systems of interacting symptoms and providing tools to identify mechanisms that sustain comorbidity.


Purpose. This presentation introduces methodological guidelines for applying network analysis to the study of comorbidity in mental health. The aim is to provide a structured framework that integrates theoretical, analytical, and interpretive aspects, illustrated through two empirical demonstrations.


Method. The guidelines are organized into four steps: (1) introduction of network analysis as a paradigm shift in psychopathology; (2) network estimation using Pairwise Markov Random Field (PMRF) models, a family that includes the Gaussian Graphical Model as a common specification; (3) interpretation of centrality indices, including both traditional measures and bridge centrality to detect cross-disorder connections; and (4) evaluation of accuracy and stability through bootstrap procedures and correlation stability coefficients. The demonstrations were carried out in two large-scale Peruvian datasets. The first involves depression and alcohol use disorder based on national health survey data (N > 30,000). The second focuses on depression and anxiety symptoms among university students (N > 3,000).


Results. Applying these guidelines allowed the identification of key methodological insights that advance the study of comorbidity. In both empirical demonstrations, the framework facilitated the detection of stable within-disorder structures and highlighted bridge symptoms connecting different mental health conditions. Moreover, accuracy and stability indicators were employed as diagnostics of the quality of information obtained, ensuring robustness and replicability of the findings.


Conclusions. These methodological guidelines demonstrate the utility of network analysis for identifying symptom-level mechanisms of comorbidity. By highlighting bridge symptoms as intervention targets, this framework supports more precise and context-sensitive approaches to mental health research and practice.