Data were analyzed using the Statistical Analysis System (SAS 9.3, Cary, NC, USA). Cox-Proportional Hazards Regression (SAS PROC PHREG) was used to model drinking milestones and all individuals who did not experience an event in adolescence were censored at their age of last interview or 18. Participants with rs1229984 GA genotype (N=96) and AA genotype (N=2) were collapsed into one group for comparison with the GG genotype participants (N=1,452), as done in previous studies (Bierut et al., 2012). Models were checked for violations of the proportional hazards assumption and Schoenfeld residuals were examined. The option COVSANDWICH (AGGREGATE) was used to statistically adjust for the non-independence of correlated familial data in all analyses, as done in previous studies (Kuperman et al., 2013).
