Intervention for Veterans With Depression, Substance Disorder, and Trauma

Primary outcomes

1. Depression Symptoms Were Assessed With the Hamilton Depression Rating Scale.

   Time frame: Assessed quarterly; trajectories analyzed total scores from randomization through end of study (6 timepoints covering approximately 15 months)

   Depression symptoms were assessed using a structured clinical interview assessment consisting of 21 items. Depression symptoms experienced in the past week are rated on a 0 (none) to 4 (most severe) scale. The total score is summed from the item scores and range from 0 (none) to 84 (most severe). We tested whether treatment group interacted with time in order to examine whether trajectories of symptom scores differed across treatment conditions. Linear mixed effects models were used to ascertain trajectories of total scores from randomization through end of study (6 timepoints across approximately 15 months). This analytic approach was advantageous in that it provided examination of change in depression across all quarterly assessments. Models were estimated with maximum likelihood methods. Total score at the time of randomization is reported in Data Table and coefficient estimates for trajectories are reported in Statistical Analysis.

2. Posttraumatic Stress Disorder (PTSD) Symptoms

   Time frame: Assessed quarterly; trajectories analyzed total scores from randomization through end of study (6 timepoints covering approximately 15 months)

   The Posttraumatic Stress Disorder Checklist - Civilian version (PCL-C) is a 17 item self-report checklist of PTSD symptoms experienced in the past month rated on a 1 (not at all) to 5 (extremely) scale; total score is summed from the item scores and range from 17 (none) to 85 (most severe). . Civilian version was selected as it allows for a variety of trauma types. Scores above 50 are considered clinical levels. We tested whether treatment group interacted with time to examine whether trajectories of symptom scores differed across treatment conditions. Linear mixed effects models were used to ascertain trajectories of total scores. This analytic approach was advantageous in that it provided examination of change in PTSD across all quarterly assessments. Models were estimated with maximum likelihood methods. Total score at the time of randomization is reported in Data Table and coefficient estimates for trajectories are reported in Statistical Analysis.

3. Timeline Followback

   Time frame: Assessed quarterly; trajectories analyzed from randomization through end of study (covering approximately 15 months)

   The Timeline Followback is a calendar-assisted structured interview that assesses the frequency of alcohol and drug use on a daily basis and the quantity of alcohol use. Summary proportion of days abstinent were calculated at each time point. Trajectory analyses examine two substance use outcomes: probability of any alcohol or drug use on a given day and probability of heavy drinking (5 or more drinks consumed in a day) on a given day. Trajectories of substance use (any alcohol or drug use on a particular day) and heavy drinking (>5 drinks on a particular day) were modeled as dichotomous outcomes, using logit links to predict the probability of substance use or heavy drinking on a particular day. Data Table reports starting proportion of days abstinent at time of randomization.