4.2 Strengths and limitations
Our work has several important strengths. The MR approach can largely reduce the risk of reverse causality bias and the effect of confounders by using genetic variation as instrumental variation to analyse the potential risks of medical abortion. In addition, complementary methods and MVMR methods were used for pleiotropy and sensitivity analysis, resulting in reliable conclusions that EA, SB, LNSP, AFB, AFS were associated with MA. The causal chains of EA-AFB-MA causality were elaborated through mediated MR. EA was elaborated to impact MA more than HI by MVMR. Finally, we limited the bias due to population stratification by using only genetic variants from European populations in all datasets.
There are several limitations to our study. Firstly, the strength of the test for multivariate MR was modest, leading to the possibility of false positives, which could be explained by the insufficient sample size. Secondly, there were significant heterogeneity in the analyses addressing the exposure factors: EA, AFS, and AFB, but the conclusions were plausible because of the consistency of the results from MVMR and mediated MR. Thirdly, there is a lack of published GWAS articles on outcome-medical abortion. A bidirectional study is essential if complete GWAS statistics for medical abortion become available in the future. Finally, abortion is a worldwide issue, but we only used summary data from European populations, resulting in data deficits, and this finding should be validated in other populations in further study.