Health Disparities
Assessing mediation of sexual orientation disparities in miscarriage by age at pregnancy Payal Chakraborty* Payal Chakraborty Chakraborty Chakraborty Chakraborty University of North Carolina at Chapel Hill
Background: Existing studies show sexual orientation-related disparities in miscarriage. Because age is strongly associated with miscarriage and may lie on the causal pathway (e.g., higher rates of both teen and later-life pregnancies among sexual minority people), conventional age adjustment fails to produce direct effect estimates. To address this, we used formal mediation methods to evaluate the role of age at pregnancy in sexual orientation disparities in miscarriage.
Methods: We used pregnancy data from the Nurses’ Health Study 2, randomly selecting one pregnancy per person as pregnancy outcomes within a person are correlated. Because exposure induced mediator-outcome confounding was likely present, we estimated interventional mediated effects (interventional pure natural direct effect [iPNDE] and total natural indirect effect [iTNIE]). Our outcome was miscarriage (yes vs. no). We treated age as multinomial (pregnancies at age <20 or ≥35 vs. 20–34 years). We examined sexual orientation dichotomously (sexual minority vs. heterosexual) and with sexual minority groups disaggregated (heterosexual with same-sex experience, mostly heterosexual, bisexual, and lesbian/gay). Baseline covariates included measures of social origin and mediator-outcome confounders included a variety of demographic and health factors.
Results: Among 64,713 participants, 12% had a miscarriage. About 3% of pregnancies occurred at age <20 and 16% at age ≥35 years. We found some evidence of mediation by age comparing sexual minority groups combined to heterosexual participants (iPNDE: 1.17; 95% CI: 1.00–1.37; iTNIE: 1.02; 95% CI: 1.01–1.04; proportion mediated: 12%). Results varied for sexual orientation subgroups (Figure).
Conclusions: Age at pregnancy may contribute to a small portion of sexual orientation disparities in miscarriage. Analyses of other clinically relevant pregnancy characteristics, such as use of medically assisted reproduction, are ongoing.
