Method: Data from two waves of the Midlife Development in the United States study were used in this study (N = 1,917). Four subscales of the Center for Epidemiological Studies Depression (CES-D) scale (depressed affect, positive affect, somatic complaints, interpersonal difficulties), five subscales of the Childhood Trauma Questionnaire (emotional abuse, emotional neglect, physical abuse, physical neglect, sexual abuse), and six immune markers (C-reactive protein [CRP], interlukin-6 [IL6], fibrinogen, E-selectin, Intercellular adhesion molecule-1, and tumor necrosis factor- α) served as nodes in an undirected psychometric network analysis. Age, body mass index (BMI), income, and alcohol consumption served as model covariates. Edges between nodes were calculated using partial Spearman’s correlation. Strength centrality (SC) was used to understand the relevance of nodes in the overall network. Separate networks were then modeled for males and females and tested for differences using network comparison testing.
Results: The sample had a mean age of 53 years. Most identified as non-Hispanic White racial/ethnic identity (76.9%), and 46.3% as female. Mean CES-D scores were 15.29 out of 60, while mean CTQ scores were 38.09 out of 125.The full sample network showed emotional abuse (SC=1.52), depressed affect (SC=1.12), IL6 (SC=1.63), and CRP (SC=1.11) had the greatest strength centrality, indicating strong influence in the model. Bootstrapped correlation found excellent stability across all metrics. Model comparison by gender showed excellent stability for male and female networks across all metrics. IL6 (male SC= 1.37; female SC=1.45), emotional abuse (male SC=1.23; female SC=1.57), and depressed affect (male SC=1.12; female SC=1.08) were found to have high strength centrality for both networks. However, somatic complaints (SC= 1.42) and CRP(SC=1.30) had greater centrality for males, while BMI (SC=0.93) had greater centrality for females. Network comparison testing revealed significant model invariance (M=0.16, p<0.01), but insignificant differences in global network strength (P=0.90).
Conclusion: Dimensions of immune system dysregulation, childhood maltreatment, and depressed affect were significant nodes across models and may represent important targets for treatment. However, network models disaggregated by gender indicated that men and women have differing experiences related to somatic complaints, CRP, and BMI. Intervention efforts that consider gender differences for survivors of childhood maltreatment who experience depressive symptoms in adulthood may be an important step for improving treatment outcomes. Additional research examining inclusive racial/ethnic groups and gender identities will further our understanding of how networks may best inform the treatment for persons with depression.