Cox proportional hazards survival models for onset of hazardous drinking (AUDIT total score ≥8) in ALSPAC and DSM-IV alcohol dependence in COGA were fit to the data with identical covariates as the regression analyses. Those who did not meet criteria for hazardous drinking or alcohol dependence were censored at their age at last interview (since age at birth). For ease of comparison across the two samples, the same PRS threshold (pT < 0.5) was used in both COGA and ALSPAC and for both the AUDIT-C and AUDIT-P PRS. Violations of the proportional hazards assumption for the PRS were tested using scaled Schoenfeld residuals. All analyses were conducted in R (R Core Team, 2017). In COGA, family identifier and recruitment site were included in the survival models as a nested frailty term (i.e. `(1 | site/family)`), using the “coxme” package for mixed-effect Cox proportional hazards models containing Gaussian random effects (Therneau, 2018). ALSPAC data were analyzed using the “survival” and “survminer” packages in R (Kassambara et al., 2017; Therneau & Lumley, 2015)
