Fengping shao
Department of Obstetrics and Gynecology, The First Affiliated Hospital, Sun Yat-sen University, Zhongshan Second Road 58, Guangzhou 510080, Guangdong, China
Objective : To explore the potential factors that contribute to the occurrence of medical abortion(MA) through a Mendelian randomization(MR) study.
Design: Univariate MR(UVMR) and multivariate MR(MVMR) analyses.
Setting: Genetic variants from European populations.
Population or Sample: Instrumental variants for MA were obtained from FinnGen with 36,232 cases and 149,622 controls.
Methods: The inverse variance weighting method was adopted as the primary analysis.
Main outcome measures: The associations of MA with household income(HI), education attainment(EA), cognitive performance(CP), risky behaviors: smoking behavior(SB), alcohol consumption(AC), and reproductive traits: age at first sexual intercourse(AFS), lifetime number of sexual partners(LNSP), age at first birth(AFB), age at last birth(ALB).
Results: In the UVMR, increasing HI, EA, AFS and AFB appeared to reduce MA risk(HI, OR=0.569, P = 7.93E-08; EA, OR=0.875, P = 6.02E-21; AFS, OR=0.439, P = 5.17E-25; AFB, OR=0.815, P = 5.46E-12), whereas SB and LNSP appeared to add to MA risk((SB, OR=1.424, P =8.32E-11; LNSP, OR=2.777, P =2.14E-11). In the MVMR, EA, SB, LNSP and AFS seems to be the predominant risk factor for MA risk with the independent effect, while HI had no effect after controlling EA(HI in model 1, OR=0.890, P = 5.78E-01). AFB functioned as mediators in the causal chain of MA risk reduction by EA, with the mediated proportion of AFS and AFB being 57.8%.
Conclusions: Our MR study demonstrated the causal potential of the associations of HI, EA, SB, LNSP, AFS and AFB with medical abortion.
Keywords: medical abortion, risk, causality, Mendelian randomization.
Introduction
Abortion is a common healthcare intervention for unintended pregnancy. Changes in abortion knowledge and attitudes over time that are perceived as safer rather than more dangerous, easier as opposed to more difficult to access, acceptable as opposed to wrong, or self-identifying as pro-choice as opposed to pro-life[1]. It was estimated that there is a global average of 73.3 million abortions per year, which corresponded to 39 abortions for every 1,000 women aged 15-49 years[2]. However, abortion may be associated with a series of immediate complications(e.g. failed attempted abortion, hemorrhage, uterine perforation, cervical trauma, repeat aspiration, disseminated intravascular coagulation), late complications(e.g. pelvic inflammatory diseases, retained products of conception, continuing pregnancy) and poor reproductive outcomes (e.g. asherman syndrome, subfertility, ectopic pregnancy, miscarriage, preterm birth and low birth weight)[3].
It is particularly important to prevent abortions among minors and adults to the maximum extent possible, which requires extensive education, improvement of lifestyles and psychological conditions, and a focus on medical, legal and economic, key direction in health protection strategy not only of adolescents and the youth, but also of population as a whole[4]. Thus, reducing unintended pregnancy and abortion is an urgently need. Consequently, the risk factors for abortion need to be explored. Bearak J et al. estimated the relationship between income, region or legal status of abortion and rate of abortion or unintended pregnancy by a new statistical model[2].
Household income(HI), education attainment(EA), cognition performance(CP), risky behaviors: smoking behavior(SB), alcohol consumption(AC), and reproductive traits: age at first sexual intercourse(AFS), lifetime number of sexual partners(LNSP), age at first birth(AFB), age at last birth(ALB) may contribute to the occurrence of unintended pregnancies and may be responsible for forcing women to choose medical abortion(MA).
Mendelian randomization(MR), using genetic variants as instrumental variables (IV) from genome-wide association studies(GWAS), provides a more robust method for assessing a causal effect between a risk factor and an outcome than does much conventional observational epidemiology[5]. Furthermore, a two-sample MR design utilizing existing genetic data or DNA samples from large-scale genetic association studies increase the feasibility and cost-efficiency of MR studies[6]. MR studies are less susceptible to confounding factors and reverse causality based on the principle of genetic variants being randomly distributed at meiosis and fixed at conception.
Multivariable MR (MVMR) is a recent extension to MR, retaining the benefits of using genetic instruments for causal inference, such as avoiding bias due to confounding. This approach can be used to estimate the effect of multiple exposures on an outcome and/or the genetic variants associated with multiple, potentially related exposures on a single outcome[7].
MR studies have been successfully applied to various causal relationship analyses between individual behaviors, cognition, education, household income with various diseases[8-11]. Hence, we performed univariate MR(UVMR) and MVMR to explore the causal associations between income, education, cognition, risky behaviors, reproductive traits and MA by using newly published summary genetic association statistics from large-scale genome-wide association studies.
Materials and methods