We conducted several secondary analyses: first, we tested whether the PRS for schizophrenia was associated with the severity of CUD diagnosis and whether CUD severity predicted greater endorsement of cannabis-related experiences. We also tested whether age at first cannabis use influenced the associations between the schizophrenia PRS and cannabis-related experiences. Given the ascertained nature of the COGA sample, we also tested whether controlling for AUD diagnosis influenced associations. Given prior evidence of genetic overlap between schizophrenia and CUD,11,36 we also tested whether including a PRS for CUD in the model attenuated associations between the schizophrenia PRS and cannabis-related experiences. For cannabis-related experiences that were nominally associated (P < .05) with the schizophrenia PRS, we tested whether there were significant interactions between the schizophrenia PRS and duration of daily cannabis use, age of first cannabis use, or CUD. We controlled for all moderator-by-covariate and PRS-by-covariate cross-terms37 but did not control for CUD in interaction analyses of duration or age of first use. Finally, we also tested whether the schizophrenia PRS was associated with the use of illicit substances, or with the number of cannabis-related experiences endorsed.
