“…The four remaining methods give more reliable results in the presence of horizontal pleiotropy, although at the cost of reduced statistical power; 2) To detect heterogeneity and possible horizontal pleiotropy, Cochran’s Q statistic, the intercept test, leave-one-out analysis, MR pleiotropy residual sum and outlier analysis (MR-PRESSO) ( Verbanck et al, 2018 ) and funnel plot analysis were performed. 3) PhenoScanner ( Kamat et al, 2019 ) resulted in the removal of 2 SNPs from the original 20 SNPs of Li et al (18 SNP set) ( Supplementary Table S3 ), because the 2 SNPs were previously reported associations with three plausible confounders (BMI, smoking, and alcoholic drinks) at p < 5 × 10 –8 , which have been reported to impact the risk of shorter LTL and COVID-19 severity ( Valdes et al, 2005 ; Zhou et al, 2020 ; Maugeri et al, 2021 ). 4) We restricted our analysis to a subset of the 20 SNPs from Li et al used in primary analyses (12 SNP set) encoding components of the SHELTERIN complex, regulating telomere structure, or regulating the formation and activity of telomerase ( Li et al, 2020 ) ( Supplementary Table S3 ); 5) To avoid the possibility of insufficiently powered instruments, we included more instruments up to 52 SNPs associated with LTL ( Li et al, 2020 ) (52 SNP set) ( Supplementary Table S3 ) at a liberal significance level (ranging from p = 5 × 10 –8 to p = 1.03 × 10 –5 , equivalent to a false discovery rate of <0.05).…”