Background Socioeconomic status (SES) is strongly linked to adverse pregnancy outcomes (APOs), but it is unclear whether these associations are causal and involve any mediators. Methods We carried out a two-sample Mendelian randomisation (MR) analysis to evaluate the causal links between SES (pre-tax income, educational level, the Townsend Deprivation Index and unemployment) and 15 APOs, including gestational diabetes, gestational hypertension, spontaneous abortion, ectopic pregnancy, postpartum haemorrhage and pre-eclampsia. A two-step MR approach was used to investigate mediation mechanisms via six selected mediators: female adult body mass index (BMI), 25-hydroxyvitamin D levels, Cigarette-Day, anxiety disorders, folic acid and alcohol intake frequency. Causal estimates were obtained using inverse-variance weighted MR, and multiple sensitivity analyses and false discovery rate corrections (FDR) were performed. Results After FDR correction, we found that higher education level was causally related to gestational diabetes (OR=0.721, 95% CI 0.596 to 0.871), gestational hypertension (OR=0.731, 95% CI 0.620 to 0.862), preterm premature rupture of membranes (OR=0.663, 95% CI 0.524 to 0.838), pre-eclampsia (OR=0.736, 95% CI 0.593 to 0.914), preterm birth (OR=0.759, 95% CI 0.633 to 0.910), spontaneous abortion (OR=0.807, 95% CI 0.703 to 0.927) and ectopic pregnancy (OR=0.675, 95% CI 0.533 to 0.854). BMI mediated the association between education level and gestational diabetes, gestational hypertension and pre-eclampsia, accounting for 60.3%, 50.8% and 54.4% of the total effects, respectively. Cigarette-Day mediated 28.1% of the association with ectopic pregnancy, while alcohol intake frequency explained 42.1% of the association with gestational hypertension. Conclusion Socioeconomic disadvantage has a causal impact on several APOs, particularly through lower educational level. These associations are partly mediated by modifiable behavioural and metabolic factors, underscoring the need for targeted prenatal interventions in populations with low SES.
Zhang et al. (Fri,) studied this question.