Frequency and network analysis of depressive symptoms in patients with cancer compared to the general population

BACKGROUND: The use of sum scores of depressive symptoms has been increasingly criticized and may be particularly problematic in oncological settings. Frameworks analyzing individual symptoms and their interrelationships such as network analysis represent an emerging alternative.

METHODS: We aimed to assess frequencies and interrelationships of 9 DSM-5 symptom criteria of major depression reported in the PHQ-9 questionnaire by 4020 patients with cancer and 4020 controls from the general population. We estimated unregularized Gaussian graphical models for both samples and compared network structures as well as predictability and centrality of individual symptoms.

RESULTS: Depressive symptoms were more frequent, but less strongly intercorrelated in patients with cancer than in the general population. The overall network structure differed significantly between samples (correlation of adjacency matrices: rho=0.73, largest between-group difference in any edge weight: 0.20, p < 0.0001). Post-hoc tests showed significant differences in interrelationships for four symptom pairs. The mean variance of symptoms explained by all other symptoms in the same network was lower among cancer patients than in the general population (29% vs. 43%).

LIMITATIONS: Cross-sectional data do not allow for temporal or causal inferences about the directions of associations and results from population-based samples may not apply to clinical psychiatric populations.

CONCLUSIONS: In patients with cancer, both somatic and cognitive/affective depression symptoms are less likely to be explained by other depressive symptoms than in the general population. Rather than assuming a consistent depression construct, future research should study individual depressive symptom patterns and their potential causes in patients with cancer.