Fig. 3
Effects of the genetic variants for increased educational attainment (EA) on alcohol dependence (AD) and AUDIT. Fifty-three genome-wide significantly associated (P < 5 × 10-8) independent (linkage disequilibrium (LD) R2 = 0.001, clumping distance = 10,000 kb) single nucleotide polymorphisms (SNPs) were used as instruments for EA. AUDIT outcomes were assessed on sub-cohort of UKB participants in the UKB AUDIT module. Results from inverse variance weighted (IVW) and three complementary two-sample MR methods, following removal of variants identified as outliers (MR-PRESSO P < 0.10) are shown. Effect (ß) measures the change per unit increase in outcome per standard deviation (SD = 3.61 years) increase in EA; with regard to binary outcome AD, ß is equal to the ln (odds ratio) (OR) of AD per SD unit increase in EA. Error bars indicate 95% confidence intervals at the nominal threshold 0.05. With 20 comparisons overall in the UKB cohort, the Bonferroni corrected threshold for comparisons would be 0.0025, given a nominal threshold 0.05. With only one comparison in the PGC AD study, the threshold for AD is 0.05. EA = Educational attainment; AUDIT = Alcohol Use Disorder Identification Test: MR = Mendelian randomization; IVW = inverse-variance weighted
