Depression phenotypes informed by the RDoC framework using the Patient Health Questionnaire-9: A retrospective study in Peru

Background Traditional approaches treat depression as unidimensional constructs rather than neurobiologically informed symptom clusters. The objective to investigate how Research Domain of Criteria-informed depression assessment enhances phenotype identification compared to traditional factor structures using nationally representative Peruvian data. Methods A cross-sectional analysis of Peru's National Health Survey (2015–2024) included confirmatory factor analysis, comparing five competing models ( n = 318,681), and latent profile analysis to identify depression phenotypes among clinically significant cases ( n = 22,339; PHQ-9 score ≥ 10). Models were estimated using maximum likelihood with robust standard errors and entropy-based class selection criteria. Results The four-factor model demonstrated a superior fit (RMSEA = 0.019) compared to traditional approaches (RMSEA = 0.050–0.055), representing an 89% improvement in model fit. Seven distinct phenotypes emerged with unique factor score patterns: Mild Cross-Domain Depression (39.4%), High Negative Valence Internalizing Depression (15.1%), Primary Negative Valence Externalizing Depression (12.5%), High Negative Valence with Preserved Cognitive Systems Depression (12.1%), High Arousal-Regulatory Systems Depression (9.1%), Severe Cross-Domain Depression (8.7%), and Severe Dual Negative Valence Depression (3.1%). Conclusions Depression functions as a multidimensional neurobiological system with distinct phenotypic presentations across Peruvian populations. The High Negative Valence Internalizing Depression—characterized by suicide risk independent of mood severity—necessitates phenotype-specific rather than symptom-severity-based clinical protocols.