ADH1B rs1229984 variant was not associated with self-reported peer drinking, supporting our interpretation that peer drinking acts as an environmental modifier, but other gene-environment correlations may still contribute to the observed effects. Fourth, the temporal ordering of peer drinking and the onset of drinking behaviors could not be assessed in this study (Table 1). It is possible that other risk factors correlated with peer drinking, such as parental monitoring or genetic risk for anti-social behavior, may account for the observed associations. Fifth, peer drinking was assessed by respondent report and may not reflect the actual proportion of best friends drinking. Finally, the majority of participants were from high-risk families, which may limit the generalizability of the findings. It is possible that only individuals at high-risk for alcohol use disorders lose the protective effect of the ADH1B rs1229984 variant under environments that encourage drinking. Replication of these findings in independent samples is a critical next step.
