Evaluation of major depression symptom networks using clinician-rated and patient-rated data.

Abstract Background Major depressive disorder (MDD) is heterogeneous, but official diagnostic classifications and widely used rating scales are based on the premise that MDD is a single disorder and that symptoms are equally important to assess severity. Also, patients and clinicians frequently diverge in how they evaluate MDD severity. In order to better understand the differences between MDD scales used by clinicians and patients in the context of MDD heterogeneity, we performed a network analysis from an approach that focuses on the interaction of symptoms rather than total score. Methods The Hamilton Depression Rating Scale (HDRS) and the Beck Depression Inventory with 21 items (BDI) scored by the clinician or patient, respectively, were used to estimate the networks based on 794 MDD patients. The networks were estimated using software R 4.0.2 and Graphical Lasso, identifying communities of symptoms by the clique percolation method, and the mixed graphical models were used to evaluate the explained variance of each symptom. Results The networks presented different communities of symptoms and connection structure (M = 0.177, p = 0.0028). The guilt connection strength and its association with suicidal ideation was greater in the BDI network. Limitations Transversal data from severe, chronic, or treatment resistant depression patients. Conclusions The present study suggests that the self-rated scale may perform better when assessing association between guilt and other symptoms, especially suicidal ideation. Communities of symptoms and edges between symptoms suggest that insomnia may be an independent symptom, thus requiring specific interventions. Some similar items are strongly connected and could be collapsed.