Pathways mediating the effect of education on pregnancy outcomes: A Mendelian randomization study

ObjectiveTo investigate the relationship between education and pregnancy outcomes, and the proportion of the effect of education mediated through modifiable cardiometabolic risk factors, using two-sample Mendelian randomization (MR) analyses. Methods and AnalysisWe extracted uncorrelated (R2 <0.01) single-nucleotide polymorphisms strongly associated (p-value <5e-8) with educational attainment, type 2 diabetes mellitus, body mass index, smoking, high-density lipoprotein cholesterol, and systolic blood pressure from the largest genome-wide association studies with available summary data. Genetic associations with ectopic pregnancy, hyperemesis gravidarum, gestational diabetes, preeclampsia, preterm birth, and offspring birth weight were extracted from the largest genome-wide association studies with available summary data. All subjects were of European ancestry. We conducted univariable MR analyses with the inverse-variance weighted method employed in the main analysis, and weighted median, weighted mode and MR Egger regression in the sensitivity analyses to account for potential pleiotropy. In mediation analyses, we compared the direct effect of educational attainment estimated in multivariable MR with the total effect estimated in the main univariable MR analysis. ResultsThe analyses included more than 3 million subjects with data on educational attainment, 270,002 subjects with data on offspring birth weight, and between 2,092 and 15,419 cases with adverse pregnancy outcomes. Each standard deviation increase in genetically-predicted educational attainment (3.4 years) was associated with an increased birth weight (95% confidence interval) of 42 g (28 g to 56 g) and an odds ratio (95% confidence interval) of 0.53 (0.46 to 0.60) for ectopic pregnancy, 0.54 (0.44 to 0.66) for hyperemesis gravidarum, 0.73 (0.67 to 0.80) for gestational diabetes, 0.81 (0.71 to 0.93) for preeclampsia, and 0.72 (0.67 to 0.77) for preterm birth. The combined proportion of the effect (95% confidence interval) of genetically-predicted educational attainment that was mediated by the five cardiometabolic risk factors was 42% (14% to 59%) for ectopic pregnancy, -17% (-46% to 26%) for hyperemesis gravidarum, 48% (19% to 82%) for gestational diabetes, 78% (10% to 208%) for preeclampsia, 28% (0% to 51%) for preterm birth, and 9% (-26% to 24%) for birth weight. Sensitivity analyses accounting for pleiotropy were consistent with the main analyses. ConclusionOur findings support that intervening on type 2 diabetes mellitus, body mass index, smoking, high-density lipoprotein cholesterol, and systolic blood pressure would lead to reductions in several adverse pregnancy outcomes attributable to lower levels of education. Such public health interventions would serve to reduce health disparities attributable to social inequalities. BOX What is Already Known on This TopicLower educational attainment is linked to increased risk of adverse pregnancy outcomes, and cardiometabolic risk factors are suspected to mediate some of this effect. What This Study AddsOur findings from using a two-sample Mendelian randomization approach are in support of a causal relationship between lower educational attainment increasing risk of ectopic pregnancy, hyperemesis gravidarum, gestational diabetes, preeclampsia, preterm birth and offspring low birth weight. A sizeable portion of the effect of educational attainment on ectopic pregnancy, gestational diabetes, preeclampsia and preterm birth is mediated by type 2 diabetes mellitus, body mass index, smoking, high-density lipoprotein cholesterol and systolic blood pressure, while these cardiometabolic risk factors combined explain little of the effect on hyperemesis gravidarum or low birth weight. How This Study Might Affect Research, Practice, or PolicyThe effects of socioeconomic inequalities on risk of ectopic pregnancy, gestational diabetes, preeclampsia and preterm birth can be reduced by intervening on type 2 diabetes mellitus, body mass index, smoking, high-density lipoprotein cholesterol and systolic blood pressure.