Finally, both the risk factor and outcome likely represent the severely affected end of the spectrum given that parents with EB and young adults with AUD were ascertained through registries. With respect to EB, our approach assumes that, in the absence of a parental registration for AUD, drug abuse, or criminal behavior during a given period, the offspring was not meaningfully exposed to parental EB. This is unlikely to be a reasonable assumption as EB may not entirely remit, and is a limitation of the data. However, the children exposed to the types of externalizing behaviors that can be ascertained via population registries represent those likely to be most in need of targeted services. With respect to AUD, our method primarily identifies young adults whose AUD has resulted in medical problems (more frequently due to acute rather than chronic use). These individuals may be at increased risk for later registrations and are in need of targeted intervention. Selection of severely-affected individuals with EB or AUD may result in an underestimation of effect sizes overall, but it is important to understand the etiology of AUD in these high-risk groups to inform intervention.
