The rs9829896 association with drug abuse was replicated at P=0.0019 in AAs from GAIN, and its OR of 0.53 was somewhat stronger than the discovery cohort (Table 2). Meta-analysis of rs9829896 across the AA cohorts resulted in P=3.93×10−10. Rs9829896 was genotyped in the UHS cohort and imputed with high quality in the GAIN cohort (info≥0.99, Table 2). Sensitivity analysis, in which we increased the required number of symptoms from 1 to 3 in the GAIN cohort to reach the threshold for a dependence diagnosis, maintained statistical significance (p=0.015) and a nearly identical OR (0.55) despite the reduced number of cases and statistical power (Table S12).
