Having identified a number of causal associations between seven modifiable risk factors and common diseases, we then sought to test whether there were causative associations between other phenotypes and diseases. We included in the analysis two traits, height48 and years of schooling49 (EduYears), for which there were a large number of instruments owing to the large GWAS sample sizes. We selected 811 and 119 near-independent genome-wide significant(GWS) SNPs for height and EduYears, respectively, using the clumping analysis (Methods). The threshold PGSMR after Bonferroni correction was 7.6 × 10−4 correcting for 66 tests. The large number of instruments for height gave us sufficient power to detect a small effect (Fig. 6; Supplementary Table 12; Supplementary Note 5).
