Hierarchical Dimensions of Psychiatric Comorbidity: An Integrated Latent Class–Network Analysis of U.S. Mental Health Client-Level Data

Understanding the complex architecture of comorbid mental health disorders is essential for improving diagnostic frameworks and guiding targeted interventions. This study employs an integrated analytical framework that combines Latent Class Analysis (LCA) and network modeling to examine the 2021 U.S. Mental Health Client-Level Dataset (MHCLD). Analysis of a nationally representative, treatment-seeking sample (N=100,000) identified a reproducible six-class latent structure, capturing the principal dimensions of psychopathology. Network community detection revealed three higher-order disorder clusters, psychotic–affective, internalizing, and externalizing, that closely mirrored the hierarchical organization proposed in contemporary psychiatric theory. Mapping each latent class onto DSM-5 diagnostic domains demonstrated strong clinical interpretability: classes aligned with bipolar–psychotic, internalizing, and neurodevelopmental spectra, while also exposing transdiagnostic bridges such as trauma–depression and bipolar–substance overlap. Validation through cross-validation, bootstrap, and entropy-based perturbation confirmed the robustness and reliability of the model, and demographic analyses further supported the distinctiveness of the identified profiles. Collectively, these findings support a hierarchical and transdiagnostic model of comorbidity, bridging data-driven discovery with DSM-5 nosology and advancing the empirical foundations of dimensional mental-health classification.