Using network analysis to examine links between individual depressive symptoms, inflammatory markers, and covariates

Background. Studies investigating the link between depressive symptoms and inflammation have yielded inconsistent results, which may be due to two factors. First, studies differed regarding the specific inflammatory markers studied and covariates accounted for. Second, specific depressive symptoms may be differentially related to inflammation. We address both challenges using network psychometrics. Methods. We estimated seven regularized Mixed Graphical Models in the Netherlands Study of Depression and Anxiety (NESDA) data (N = 2321) to explore shared variances among (1) depression severity, modeled via depression sum-score, nine DSM-5 symptoms, or 28 individual depressive symptoms; (2) inflammatory markers C-reactive protein (CRP), interleukin 6 (IL-6), and tumor necrosis factor alpha (TNF-alpha); (3) before and after adjusting for sex, age, body mass index (BMI), exercise, smoking, alcohol, and chronic diseases. Results. The depression sum-score was related to both IL-6 and CRP before, and only to IL-6 after covariate adjustment. When modeling the DSM-5 symptoms and CRP in a conceptual replication of Jokela et al., CRP was associated with 'sleep problems', 'energy level', and 'weight/appetite changes'; only the first two links survived covariate adjustment. In a conservative model with all 38 variables, symptoms and markers were unrelated. Following recent psychometric work, we re-estimated the full model without regularization: the depressive symptoms 'insomnia', 'hypersomnia', and 'aches and pain' showed unique positive relations to all inflammatory markers. Conclusions. We found evidence for differential relations between markers, depressive symptoms, and covariates. Associations between symptoms and markers were attenuated after covariate adjustment; BMI and sex consistently showed strong relations with inflammatory markers.