In order to evaluate differences in characteristics across the mutually exclusive diagnostic groups—neither DSM-IV nor DSM-5 AUD, DSM-5 AUD alone, and both DSM-IV and DSM-5 AUD—three post hoc statistical tests were conducted using 3 aggregate outcome measures to avoid multiple statistical comparisons. The three outcomes for these post hoc analyses were: 1) number of mental health or drug use conditions (out of 5 possible); 2) number of negative alcohol-related consequences in the past-3 months on the SIP (out of 15 possible); and 3) an indicator of whether or not patients reported feeling “somewhat” to “ready” to change (versus not ready). Three generalized linear models were constructed, one for each of the 3 aggregate outcomes, with the categorical variable for the remaining DSM-IV and/or DSM-5 AUD groups as the independent variable, using the group with neither DSM-IV nor DSM-5 AUD as the referent. The outcomes (number of events; binary indicator) were assumed to have a binomial distribution, with the denominator equal to the number of possible events, and a logistic link was used. An overall post-estimation Wald test then evaluated whether
