risk to develop depressive symptoms (M) (OR 3.57; 95%; CI 1.99;6.40, adjusted for smoking). Highly educated women have decreased risks for depressive symptoms (OR 0.36; 95%; CI 0.21;0.61, adjusted for smoking). Smoking, adjusted for educational level, was not associated with depressive symptoms during pregnancy.
Path B shows that EDS score (M) is no longer associated with PTB (Y) afteradjusting for the initial risk factors (X) (OR: 1.00; 95%; CI 1.00;1.01). Conform the mediation protocol, this implies that depressive symptoms do not mediate the association between initial risks and PTB. In addition, adjusting for EDS scores inpath C0 barelychanges the associations between the initial risk factors and preterm birth of path C (path C0: low educational level OR: 1.06; 95%; CI 1.02;1.10, high educational level OR: 0.97; 95%; CI 0.94;1.01).
Secondary mediation analysis with the initially excluded risk factors (past) psychotropic medication use and past psychiatric history confirms the absence of EDS score as a mediator in the pathway to preterm birth. Both risk factors decrease the impact of educational level and smoking on depression, but do not affect the impact of educational level and smoking on preterm birth (data available upon request).