Post-hoc testing was then performed to further understand the nature of our effects. First, while only a few participants exhibited a substance use disorder other than AUD and CUD, we added a binary substance use disorder (SUD) covariate into the ANCOVA models to investigate whether any effects were driven by other SUDs. Then we interrogated whether our effects were driven by AUD/CUD severity, versus the amount/frequency of alcohol/cannabis use. To index use disorder severity, we used number of use disorder symptoms for AUD and CUD from the SCID interview. To index amount/frequency of use, we utilized scores from the first three questions of the AUDIT and CUDIT questionnaires, which is often known as the AUDIT-C/CUDIT-C62. These questions inquire about frequency of use (Q1), amount of use (Q2), and binge (Q3) use, which importantly is not assessed in the SCID interview. That is, use disorder symptoms and diagnosis do not take into account amount or frequency of use. Therefore, we used these four variables (AUD/CUD number of symptoms based on the SCID and A/CUDIT-C score) in a regression to determine whether our effects were related to amount of alcohol/cannabis use or alcohol/cannabis addiction symptomatology.
