quasi-continuous, and family income was log-transformed owing to its positive skew. To control for prior psychiatric disorders, we created a prior internalizing disorder variable (priorto- past-year, including lifetime at W1) and a prior AUD variable (lifetime DSM-IV AUD at W1). For descriptive purposes, we also created a variable to indicate internalizing disorder status including persistent (prior-to-past-year [lifetime W1 and prior-to-past year at W2] and past-year disorder [W2]), remitted (prior-to-past year but not past-year disorder), incident (past-year disorder only), and unaffected (no lifetime disorder [no disorder at W1 or W2]). We also controlled for lifetime DSMIV drug use disorder (DUD), including sedative, tranquilizer, opioid, amphetamine, cannabis, hallucinogen, cocaine, inhalant, heroin, and other drug use disorder. Last, we derived a prior-to-past-year alcohol problem severity measure by summing all DSM-IV abuse and dependence criteria (max score of 11) using W1 and W2 data. We omitted the alcohol craving criterion of DSM-5 AUD from this measure because it was not assessed at W1. The AUDADIS-IV assessed DSM-IV alcohol abuse and dependence symptoms with ICCs of 0.86 and 0.89, respectively (Grant et al., 2003).
