Open Access

Johanna Klinger-König, Anke Hannemann, Nele Friedrich, Matthias Nauck, Henry Völzke, Hans J. Grabe

• Background Lower cortisol concentrations in adulthood were repeatedly associated with more severe childhood maltreatment. Additionally, childhood maltreatment was reported to promote health risk behavior, such as smoking or alcohol consumption, and to increase the risk of mental and somatic diseases during adulthood, such as major depressive disorders or obesity. The present study investigated if health risk behavior and disease symptoms in adults mediate the associations between past childhood maltreatment and present basal serum cortisol concentrations. Methods Data from two independent adult cohorts of the general population-based Study of Health in Pomerania (SHIP-TREND-0: N = 3,517; SHIP-START-2: N = 1,640) was used. Childhood maltreatment was assessed via the Childhood Trauma Questionnaire (CTQ). Cortisol concentrations were measured in single-point serum samples. Health risk behavior and mental and physical symptoms were used as mediators. Mediation analyses were calculated separately for both cohorts; results were integrated via meta-analyses. Results In mediator-separated analyses, associations between childhood maltreatment and basal serum cortisol concentrations were partly mediated by depressive symptoms (BDI-II: βindirect effect = -.011, pFDR = .017, 21.0% mediated) and subjective somatic health complaints (somatic complaints: βindirect effect = -.010, pFDR = .005, 19.4% mediated). In the second step, both mediators were simultaneously integrated into one mediation model. The model replicated the mediation effects of the subjective somatic health complaints (whole model: βindirect effect = -.014, p = .001, 27.6% mediated; BDI-II: βindirect effect = -.006, p = .163, 11.4% mediated, somatic complaints: βindirect effect = -.020, p = .020, 15.5% mediated). Conclusion The results support the hypothesis that the long-lasting effects of childhood maltreatment on the stress response system are partly mediated through self-perceived disease symptoms. However, no mediation was found for health risk behavior or physically measured mediators. Mediation models with multiple simultaneous mediators pointed to a relevant overlap between the potential mediators. This overlap should be focused on in future studies.