Our findings should be viewed in the context of several limitations. First, the AUDIT is a past-year screener for alcohol consumption and problems, and thus may not capture individuals who had alcohol use problems in the past. Second, the AUDIT discovery GWAS was performed in a subset of the UKB that consisted of individuals who responded to an email request to complete a mental health follow-up survey. A recent study of the genomic profiles of these individuals has found them to represent lower polygenic risk for psychopathology and greater polygenic load for educational achievement, suggesting selection bias in the discovery sample itself (Adams et al., 2018). In addition, these individuals had relatively low endorsement of AUDIT-P items, which might have limited the statistical power of the discovery GWAS to capture the full spectrum of problem drinking and, in turn, limited our ability to compare across the AUDIT-C and AUDIT-P. Third, as the discovery GWAS was conducted on individuals of primarily European descent, these PRS could not be applied to other ancestral groups, reflecting a research gap in the dearth of
