Our results should be considered in light of several limitations. First, COGA is a high-risk sample, with most participants from extended families enriched for AUD. In addition, parental separation was more prevalent in our sample (relative to 28% in the general population5), which may limit generalizability to other populations with different risk profiles. Second, our measure of parental relationship discord was based on offspring reports and may not fully reflect the actual quality of marital relationship between parents, but it does indicate how adolescents perceive it. This may be what matters for capturing risk, although it is not clear whether perceived or actual parental discord drives the association. Incorporating information from additional reporters or methods could address this possibility. Third, the timing of parental separation was not assessed, although it is important given that younger children typically face more challenges following parental separation.50 Fourth, because young adult participants were asked to retrospectively report on parental relationship discord, whereas adolescent participants provided concurrent reports, it was not feasible to treat parental relationship discord as a time-varying predictor. Future research that takes
