2.2.3 SNP Selection
We screened SNPs that had a strong connection with exposure factors (P < 5×10-8) from the GWAS data of exposure. We eliminated linkage disequilibrium (LD) between SNPs using a clustering process (R2 < 0.001 and clumping distance = 10, 000 kb) and SNPs having a minor frequency of alleles were removed (MAF < 0.01) to ensure the results were practicable and undisturbed. The SNPs we selected in exposure GWAS were matched with the GWAS data of outcomes. If SNPs cannot be found in the GWAS data of outcomes, the proxy SNPs with significant LD (r2> 0.8) were employed. Finally, once the palindrome SNPs were excluded, the rest of the SNPs were chosen as IVs.