2.3 Mendelian randomization design and instrumental variable (IV) selection
Figure 1 showed a brief description of the overall design of the study for UVMR and MVMR. The potential causal relationships between factors and abortion were deduced from UVMR and the main risk factors for MA were identified through different models of MVMR. In MR, genetic variants are used as IVs to assess the causal effect of an exposure on an outcome. Genetic variants used as IV need to meet the following three pivotal assumptions: First, the SNPs are closely associated with exposure. Second, the SNPs are independent from any confounders of exposure-outcome causality. Third, the SNPs do not directly affect outcome, but only have an impact on outcomes via the exposure. We selected the significant genetic variants that match with the genome-wide statistical significance threshold (P < 5 ×10-8). Using PLINK clustering, the linkage disequilibrium between these SNPs for each exposure was estimated on the basis of the 1000 Genomes reference panel confined to the European population, and SNPs were thresholded at linkage disequilibrium r2 < 0.001 at a 0-10,000 kb window to guarantee the independence of the selected genetic variants. Removing the SNPs for being palindromic with intermediate allele frequencies. For UVMR, we calculated the F-statistic for each exposure to assess the strength of each tool in the abortion GWAS, and SNPs with F statistics <10 were identified as weak instruments(Supplementary table 3-12). In addition, we calculated the conditional F-statistic for assessing instrument strength in two sample summary MVMR
Statistical analyses