prediction of liability to problem drinking. Furthermore, while we focused on contrasting the genetic relationship of each individual AUDIT subscale with the variety of alcohol-related measures across our different samples, it should be noted that the proportion of variance explained in the alcohol-related outcomes is larger when we consider both the AUDIT-C and AUDIT-P PRS together, suggesting that multiple polygenic scores provide more utility than PRS for the individual sub-scales. The two scores are only modestly correlated and likely, some of this commonality is due to variants in ADH1B that exert a relatively large effect. Overall, our analyses suggest that even though it is far too small to be statistically significant, there is some incremental contribution of variance from inclusion of the AUDIT-C. Despite these caveats, our analyses demonstrate the feasibility of using a fast and simple screening questionnaire, the AUDIT, that assesses both consumption and problem drinking, to understand the course of alcohol use and misuse.
