Third, as noted above, confounding of the association between PTEs and drug use could be due to parental substance misuse, which we adjust for in final models. However, parental substance misuse could also be a direct outcome of a PTE, if the event was experienced by both the parent and the child (e.g. death of a loved one, car accidents). Including this variable in statistical models may therefore induce bias in these estimates. As we are unable to tease out the temporality of these variables in the NCS-A dataset, we present models both unadjusted and adjusted for parental substance misuse, for comparison. We also conducted interaction tests to assess moderation by parental substance misuse, and mostly found no significant differences, as noted above. Future studies should utilize longitudinal data to further explore these associations.
