This was a secondary investigation using data from a longitudinal study on emerging adult SUD treatment outcomes and Alcoholics Anonymous (AA) engagement. All participants from the larger study were included in the current study. Study participants were recruited from the Hazelden Center for Youth and Families (HCYF) in Plymouth, MN between October 2006 and April 2008.
HCYF is an addictions treatment program and part of the Hazelden Foundation, a non-profit organization with substance use disorder treatment facilities nationwide. Treatment at HCYF is grounded in the 12-step based Minnesota model of treatment and supplemented with other methods (e.g., cognitive behavioral therapy, motivational interviewing). Separated by gender, key components of the residential program include individual and group therapy, integrated mental health assessment and therapy, individual assignments related to treatment goals and the 12 steps of AA, education through bibliotherapy and daily lectures, and recreational and spiritual care. Patients at HCYF range in age from 14 to 25 years, with a mean age of 18.
The sample included 302 participants (73.8% male), reflecting the male to female ratio at the treatment center. Participants were, on average, 20.35 years old (Range = 18–24; SD = 1.58), and spent 25 days in residential treatment (SD = 6.45). Consistent with the racial and ethnic distribution of the surrounding region, 95.3% of the sample was Caucasian. The majority had a H.S. diploma (81.7%), while only 1.7% (n = 5) had completed a bachelor’s or associate’s degree (see ). At treatment intake, 46.1% of the sample was not in school (n = 137), 31.0% were attending college part- or full-time (n = 92), 12.8% were seeking a college degree but not enrolled (n = 38), and 10.1% were in high school (n = 30). The justice system had suggested or mandated treatment for 34 of the participants (11.3%), and in the year before treatment participants had spent an average of 1.56 days (SD = 7.50; range 0–120) in residential SUD treatment or in-patient detoxification programs.
Participant Characteristics at Intake as a Function of Past-Month Depression Status
Potential participants were required to be 18 or older, and were approached for recruitment during their first three days of residential treatment. Research assistants spoke with 367 individuals about the study. Of these, 47 (12.8%) declined to participate, citing a lack of interest (n = 15), a desire to avoid follow-ups (n = 8), the study taking too much time (n = 8), preferring to focus on treatment (n = 6), not expecting to finish treatment (n = 5), and other miscellaneous reasons (n = 5). While there was no difference in gender between those who participated versus those who declined, there was a difference in age, t(360) = p <.05; those who declined to participate were younger. The actual age difference between the two groups was slight (mean of 19.79 for non-participants vs. 20.37 for participants).
Individuals who gave informed consent were scheduled for a baseline interview within their first seven days of treatment. An independent review board, Schulmann Associates IRB, approved the larger study. Out of the 320 individuals who agreed to participate in the study, 302 (94.4%) completed the baseline interview and were included in the analyses. There were no significant differences in age, gender, or length of stay between those who completed the baseline interview and those who did not.
Participants completed a mid-treatment assessment 14 days into treatment, an end-of-treatment session prior to discharge, and a follow-up interview three months after discharge. The per-session compensation was scaled according to session duration (e.g., $30 for intake, $10 for mid-treatment). Ninety-one percent completed the mid-treatment assessment and 86% completed the end-of-treatment assessment. Participants who completed the end of treatment assessment did not differ from non-completers in gender, length of stay, presence of MDD, intake ADUSE score, or pre-treatment use. Completers were significantly older, t(296) = −2.07, p = .039 (M = 20.40 versus 19.84 for non-completers). Attrition analyses were also conducted for those who completed the three-month follow-up (80.5%; n = 243) versus those who did not, and no significant differences were found.
Treatment attendees provided information about their gender, age, ethnicity, and education as part of the routine intake process.
Recent depressive symptoms
The Brief Symptom Inventory 18 (BSI 18; Derogatis, 2000
) was administered at each time point and the six-item Depression subscale provided a continuous measure of depressive symptoms during the previous seven days. The BSI 18 has been normed in community and oncology samples and has acceptable test-retest reliability and internal consistency (Derogatis, 2000
Substance use consequences Major depressive disorder
The Structured Clinical Interview for the DSM-IV (SCID-I/P) is a semi-structured interview for the primary DSM-IV Axis I diagnoses (First, Spitzer, Gibbon, & Williams, 2002
). The study had three SCID raters: two had bachelor’s degrees in psychology and one had a PhD in experimental psychology. To prepare for SCID-I/P administration, each of the raters studied the SCID User’s Guide, thoroughly reviewed the SCID, practiced administering the SCID with each other, watched the diagnosis-relevant SCID-101 Didactic Training Series videos, and completed SCID ratings while watching the pre-recorded SCID interviews. SCID interviews were audiotaped during the study, and a random subset of tapes was listened to by a doctoral-level research scientist to ensure adequate diagnostic reliability. No formal inter-rater reliability coefficients were computed.
SUD diagnoses were used to describe the sample, and diagnoses of past-month MDD were used as a dichotomous predictor variable indicating depression present or absent. Those diagnosed with non-current lifetime MDD were not included with this group because lifetime MDD has not been a reliable predictor of SUD treatment outcomes (Bradizza, Stasiewicz, & Paas, 2006
; Ramo & Brown, 2008
When diagnosing MDD using the SCID, studies have found excellent Kappas for joint reliability (0.80 to 0.93), while Kappas for test-retest reliability were fair, ranging from 0.61 to 0.64 (First, Spitzer, Gibbon, & Williams, 1995
; Zanarini & Frankenburg, 2001
). Several studies have found somewhat lower reliability when using the SCID for comorbid diagnoses among those with substance use disorders (Kranzler et al., 1996
), which may result from a discrepancy in determining whether MDD is independent or substance-induced (Grant et al., 2004
). Because of this, we utilized the BSI 18 Depression scale as an additional alternative measure of depressive symptoms.
Substance use frequency and AA attendance
The Form-90 instruments were developed for Project MATCH to assess alcohol consumption (Miller & Del Boca, 1994
). This family of instruments has shown good test-retest reliability and validity across samples, including adolescents (Rice, 2007
; Slesnick & Tonigan, 2004
; Tonigan, Miller, & Brown, 1997
). The Form-90D is an alternate version of the Form-90 that has demonstrated excellent reliability in measuring illicit drug use (ICCs = 0.75 – 0.82; Westerberg, Tonigan, & Miller, 1998
). A modified version of the Form-90D was used to assess days using substances other than nicotine in the 90 days before intake, mandated treatment status, pre-treatment SUD treatment and AA attendance, and abstinence status at follow-up.
The Alcohol and Drug Use Self-Efficacy (ADUSE) scale (Brown et al., 2002
) was modified from the Alcohol Abstinence Self-Efficacy scale (DiClemente, Carbonari, Montgomery, & Hughes, 1994
) to inquire about both alcohol and drug use. Participants completed this measure at each assessment point. The ADUSE consists of 20 questions rated on a Likert scale from 0 (not at all) to 4 (extremely) that inquire about confidence to avoid drinking or using in different situations. The sum of all 20 items provides an index of overall abstinence self-efficacy. DiClemente and colleagues constructed the Alcohol Abstinence Self-Efficacy Scale from Marlatt and Gordon’s relapse categories (1985)
. Subscales include Negative Affect, Social/Positive, Physical and Other Concerns, and Craving/Urges situations.
The ADUSE demonstrated excellent reliability at intake, with an overall alpha of 0.94 (20 items, N
= 295). Subscales also demonstrated good reliability (α = 0.86 – 0.91). These values are similar to reliability estimates for the Alcohol Abstinence Self-Efficacy Scale (DiClemente et al., 1994
). For the ADUSE, we used item means for missing values because individual items varied in their correlation with the total ADUSE score.
All analyses were conducted using SPSS 17.0. Using the three in-treatment assessment points, random coefficient regression was used (RCR; Cohen, Cohen, West, & Aiken, 2003) to model within-subject linear changes in abstinence self-efficacy over the course of treatment, and to assess for variability in those linear changes (Hypothesis 2). Two models were computed for these analyses: one focused on total ADUSE as the dependent variable, the other focused on ADUSE Negative Affect as the dependent variable.
Time was centered at end-of-treatment and coded in weeks. MDD at baseline (coded as 0 = no depression, 1 = depression), and its interaction with time were then added to each model to determine whether there was a main effect of depression on ADUSE scores, and whether depression moderated changes in ADUSE scores over time. A second series of models was estimated, this time with recent depressive symptoms (BSI 18 Depression) as a time-varying covariate. Logistic regression was used to predict substance use (abstinent yes/no) at the three-month follow-up (Hypothesis 3).