Efficacy of Group Motivational Interviewing (GMI) for Psychiatric Inpatients With Chemical Dependence

Abstract

Dually diagnosed patients with chemical dependency and a comorbid psychiatric disorder typically show poor compliance with aftercare treatment, which may result in costly and pervasive individual and societal problems. In this study, the authors investigated the effect of adding motivational interviewing in a group format to standard treatment for dually diagnosed psychiatric inpatients. The patients (n = 101) all received standard care and in addition were assigned to either group motivational interviewing (GMI) or a therapist attention activity control group (TAAC). Of patients who attended aftercare and who used alcohol or drugs, those who participated in GMI attended significantly more aftercare treatment sessions, consumed less alcohol, and engaged in less binge drinking at follow-up compared with those in TAAC. Differences between conditions in the overall percentage of participants who achieved complete abstinence or who attended aftercare treatment were not significant, possibly because of a lack of power. These results provide preliminary evidence for the efficacy of GMI when added at the outset to an inpatient program.

Full text

Efficacy of Group Motivational Interviewing (GMI) for Psychiatric
Inpatients With Chemical Dependence
Elizabeth J. Santa Ana and Edelgard Wulfert
University at Albany, State University of New York
Paul J. Nietert
Medical University of South Carolina
Dually diagnosed patients with chemical dependency and a comorbid psychiatric disorder typically show
poor compliance with aftercare treatment, which may result in costly and pervasive individual and
societal problems. In this study, the authors investigated the effect of adding motivational interviewing
in a group format to standard treatment for dually diagnosed psychiatric inpatients. The patients (n ⫽
101) all received standard care and in addition were assigned to either group motivational interviewing
(GMI) or a therapist attention activity control group (TAAC). Of patients who attended aftercare and who
used alcohol or drugs, those who participated in GMI attended significantly more aftercare treatment
sessions, consumed less alcohol, and engaged in less binge drinking at follow-up compared with those
in TAAC. Differences between conditions in the overall percentage of participants who achieved
complete abstinence or who attended aftercare treatment were not significant, possibly because of a lack
of power. These results provide preliminary evidence for the efficacy of GMI when added at the outset
to an inpatient program.
Keywords: motivational interviewing, group therapy, substance abuse, dual diagnoses, psychiatric
inpatients
Individuals with coexisting substance use and psychiatric dis-
orders show high rates of noncompliance and often fail to partic-
ipate in posthospitalization treatment, which is associated with
significantly impaired functioning and poor clinical outcomes
(Wolpe, Gorton, Serota, & Sanford, 1993). Motivational inter-
viewing (MI; Miller & Rollnick, 2002) has shown large effects in
promoting treatment engagement and retention of individuals with
substance use disorders (Hettema, Steele, & Miller, 2005). Recent
studies indicate that dually diagnosed patients who receive indi-
vidual MI participate more actively in treatment and reduce their
substance use compared with patients not receiving MI (Hand-
maker, Packard, & Conforti, 2002).
Despite such positive findings, many inpatient treatment pro-
grams do not incorporate MI into routine practice because the
psychosocial treatments they offer are typically delivered in a
group format rather than one on one. As a consequence, MI goes
largely unused in inpatient settings, although its use might confer
substantial benefits on dually diagnosed patients. Some investiga-
tors have examined the possibility of delivering MI in a group
format (Walters, Ogle, & Martin, 2002) with varying rates of
success, and very few controlled outcome studies exist of this
modality used with substance abusers (e.g., John, Veltrup, Dries-
sen, Wetterling, & Dilling, 2003). Our purpose in the present study
was to develop MI for use in groups of inpatients with a substance
use disorder and a co-occurring psychiatric disorder and to exam-
ine its impact on aftercare attendance and substance use. We
hypothesized that patients receiving group MI (GMI) would be
attending more aftercare treatment sessions, consuming less alco-
hol, and using fewer illicit drugs at follow-up compared with those
who received standard group inpatient treatment without GMI.
Method
Participants
The sample consisted of 101 inpatients consecutively admitted
to a university-based psychiatric hospital that provided crisis sta-
bilization for people requiring medical detoxification from alcohol
or drugs or medical stabilization for a comorbid psychiatric con-
dition (see Figure 1). The average length of inpatient stay was 6.8
days. Sixty patients were hospitalized voluntarily and 41 were
committed for being a risk to themselves or others. To be eligible
for the study, patients had to have at least one current substance
use disorder and a nonsubstance-related major Axis I disorder
(e.g., major depressive disorder, bipolar disorder, anxiety disor-
der), speak English, read and write at least at the fifth grade level,
and not have a diagnosis of dementia or current psychosis.
Elizabeth J. Santa Ana and Edelgard Wulfert, Department of Psychol-
ogy, University at Albany, State University of New York; Paul J. Nietert,
Department of Biostatistics, Bioinformatics, and Epidemiology, Medical
University of South Carolina.
Elizabeth J. Santa Ana is now at the Yale University School of Medicine
and the Department of Psychiatry, VA Connecticut Healthcare System.
This research was conducted in partial fulfillment of the requirements
for Elizabeth J. Santa Ana’s doctoral dissertation and with support from
Grant R03 DA016747 from the National Institute on Drug Abuse, Behav-
ioral Therapies Development Program, awarded to Elizabeth J. Santa Ana.
We acknowledge the assistance of John C. Roitzsch in the completion of
this study. We thank the staff of the Medical University of South Carolina
Center for Drug and Alcohol Programs for their assistance and support of
this research.
Correspondence concerning this article should be addressed to Elizabeth
J. Santa Ana, 950 Campbell Avenue, 151-D, West Haven, CT 06516.
E-mail: elizabeth.santaana@yale.edu
Journal of Consulting and Clinical Psychology Copyright 2007 by the American Psychological Association
2007, Vol. 75, No. 5, 816 – 822 0022-006X/07/$12.00 DOI: 10.1037/0022-006X.75.5.816
816

Procedures
A treatment team chaired by a psychiatrist evaluated patients on
admission. Diagnostic interviews were based on the Diagnostic
and Statistical Manual of Mental Disorders (4th ed.; DSM–IV;
American Psychiatric Association, 1994), and diagnoses were de-
termined by consensus. Trained research staff blind to conditions
screened inpatients for study eligibility (from November 2003 to
June 2004) and obtained written informed consent before the
assessments were administered. All participants received standard
inpatient care and, in addition, were assigned to either GMI or a
therapist attention activity control group (TAAC). The first group
was randomly determined; thereafter, assignment to GMI or
TAAC (average n ⫽ 4 per group) alternated weekly. Elizabeth J.
Santa Ana delivered the GMI and TAAC interventions. A research
assistant blind to condition conducted in-person follow-up inter-
views with patients 1 and 3 months after discharge. Patients
received an honorarium for completing the follow-up assessments.
Assessment
At baseline, we collected patient demographics, administered
the Brief Symptom Inventory 18 (Derogatis, 2000) to measure
psychological distress, and obtained the patients’ psychiatric and
substance abuse treatment history for the 12 months preceding
hospitalization.
Substance use patterns were assessed with the time line follow-
back, a calendar-based method with good psychometric properties
that yields reliable retrospective data on addictive behaviors (So-
bell & Sobell, 1992). Participants provided retrospective estimates
of their daily alcohol and illicit drug consumption for each day in
a 31-day baseline and for the 1- and 3-month follow-up periods.
This information yielded the number of days of drinking alcohol,
the number of days of binge drinking (defined as 4 consecutive
drinks per occasion for women or 5 consecutive drinks per occa-
sion for men; Wechsler, Davenport, Dowdall, Moeykens, &
Castillo, 1994), and the number of days of using illicit drugs.
Assessed for eligibility (n = 211)
Excluded (n = 110)
Not meeting inclusion criteria (n =35)
Refused to participate (n = 40)
Other reasons (n = 35)
Assigned (n = 101)
Allocated to TAAC (n = 51)
Received allocated intervention (n = 51)
Allocated to GMI (n = 50)
Received allocated intervention (n = 50)
Lost to follow-up:
Month 1 (n = 2)
Month 3 (n = 8)
Lost to follow-up:
Month 1 (n = 2)
Month 3 (n = 6)
Analyzed:
Month 1 (n = 48)
Month 3 (n = 44)
Analyzed:
Month 1 (n = 49)
Month 3 (n = 43)
Figure 1. Flow of participants through the study. GMI ⫽ group motivational interviewing; TAAC ⫽ therapist
attention activity control group.
817
BRIEF REPORTS

Alcohol consumption was converted to standard ethanol content
units (SECs) equivalent to 0.5 oz of ethanol (Miller, Heather, &
Hall, 1991). We also assessed the number of days in outpatient or
residential treatment, 12-step or other self-help meetings attended,
and visits with a physician or other professional for psychiatric or
substance use treatment during baseline and follow-up.
With the patients’ permission, a research assistant blind to
condition interviewed a significant other at the same time we
conducted the patient follow-up interviews. Significant others
were interviewed in person or by telephone using the Collateral
Interview Form (Miller & Marlatt, 1987) to corroborate patient
reports on alcohol and drug use and aftercare attendance.
Interventions
Standard care. A multidisciplinary team developed compre-
hensive treatment plans for all inpatients. All patients attended
standard treatment groups
1
and received discharge summaries that
provided a detailed listing of their continuing aftercare appoint-
ments.
TAAC. Patients assigned to TAAC met with Elizabeth J. Santa
Ana for two 120-min group sessions, with the first session taking
place the day after their baseline assessments. TAAC consisted of
a “box activity” where participants wrote anonymous questions,
comments, or topics (e.g., substance abuse, relationship and family
issues, financial problems) on slips of paper that were placed in a
box and then read aloud for group discussion.
2
GMI. Patients assigned to GMI met with Elizabeth J. Santa
Ana for two structured 120-min sessions based on a GMI protocol
(manual available on request), with the first session taking place
the day after their baseline assessments. The therapist introduced
GMI “culture building,” which involves identifying common emo-
tions (Foote et al., 1999) experienced in treatment (e.g., fear,
anger), discussing ambivalence as normal, asking open-ended
questions that encourage members to share their personal views on
their hospitalization, showing empathy, using reflections, validat-
ing personal choice and autonomy, and presenting MI-consistent
group ground rules (e.g., behave collaboratively, support member
confidence, avoid argumentation). Participants who strayed from
the ground rules were redirected and reinforced for GMI-
consistent behaviors. So that key therapeutic components of indi-
vidual MI were effectively translated into a group format, GMI
was designed to operate in a workshop format, providing elements
of group therapy that serve as advantages over individual treatment
(i.e., peer support, opportunities for sharing information, role mod-
eling, feedback from peers, altruism, and instilling hope; Foote et
al., 1999; Yalom,1995) while remaining consistent with the central
principles of MI (Miller & Rollnick, 2002). To enhance discrep-
ancy building, participants, who were asked not to share results
with other members, received a personalized graphic feedback
report
3
adapted from the feedback delivered in Project MATCH
(Miller, Zweben, DiClemente, & Rychtarik, 1992). The therapist
reviewed the various categories of the report with the group while
facilitating change talk among patients and having them share their
reactions. The therapist encouraged the group to evaluate the pros
and cons of changing substance use and attending aftercare treat-
ment, their personal values and how alcohol and drug use might be
discrepant with them, and how their personal strengths could be
used to facilitate change. Finally, group members discussed obsta-
cles to aftercare treatment participation and brainstormed solutions
when members requested advice.
Therapist Training and Adherence to Protocol
Elizabeth J. Santa Ana was a doctoral-level candidate in psy-
chology who had completed intensive MI workshop training con-
sisting of skill building, practice, and skill assessment. For GMI,
she followed a manual and was closely supervised by a licensed
psychologist with formal MI training who listened to therapy tapes
and provided feedback and coaching. All 26 sessions were audio-
taped and a random sample of 5 GMI and 5 TAAC tapes was
independently rated by four raters using a 16-item GMI therapist
adherence measure, detailing strategies for therapist adherence
only, that was adapted from the Yale Adherence and Competence
Scale–II (Nuro et al., 2005).
4
Data-Analytic Approach
Because many participants (⬎20%) either attended no aftercare
sessions or did not drink or use drugs during the follow-up period,
zeros in the data set were inflated, which rendered the data non-
normal. Therefore, the data were analyzed using zero-inflated
negative binomial (ZINB) models (Long, 1997). Baseline sub-
stance use was standardized to a 31-day value. All ZINB analyses
were controlled for baseline values and conducted using Stata
(Stata Statistical Software, 2003). Analyses only included patients
who completed the 1- (n ⫽ 97) and 3-month (n ⫽ 84) follow-up
interviews.
Results
Table 1 presents participants’ baseline demographic, substance
use, and clinical characteristics. Of 211 patients assessed, 110 were
excluded, leaving 101 patients (63 men, 38 women) who were
assigned to GMI (n ⫽ 50) or TAAC (n ⫽ 51). Table 2 presents
their DSM–IV diagnoses. There were no baseline differences be-
tween groups on any of the relevant demographic, diagnostic, or
substance use variables. Alcohol (66.4%), cocaine (43.5%), and
1
Standard groups conducted on the unit included a morning community
meeting and pharmacology, relapse prevention, skills training, leisure, and
wrap-up groups. An evening group emphasized the principles espoused by
Alcoholics Anonymous.
2
Group members were asked to join the discussion and were told that
they would receive education and answers to their questions. The therapist
provided high-frequency directive but supportive advice and candidly
answered questions using a psychoeducational style not typical of MI.
3
Feedback reports summarized drinking and drug use patterns, typical
and peak blood alcohol levels, alcohol and drug use percentiles, alcohol
and drug use consequences, and blood liver enzyme functioning, among 12
other feedback components.
4
Interclass correlation coefficients for therapist adherence, using a ran
-
dom effects model (Shrout & Fleiss, 1979), ranged from .58 to .99, with a
mean of .85 (SD ⫽ .13). The t tests, with treatment group as the indepen-
dent variable and respective mean fidelity scores as the dependent vari-
ables, showed that the GMI intervention was characterized by significantly
higher rates of GMI adherence for all items, t(8) ⫽ 13.04, p ⬍ .001,
whereas the TAAC intervention was characterized by significantly higher
rates of TAAC adherence, t(8) ⫽ 15.92, p ⬍ .001.
818
BRIEF REPORTS

cannabis (20.8%) dependence were the most prevalent substance
use disorders, whereas major depression (58.1%), the anxiety
disorders (30.8%), and bipolar disorder (13.9%) were the most
prevalent non–substance use disorders. Ninety-seven (96%) of the
101 patients (GMI ⫽ 48, TAAC ⫽ 49) completed the 1-month
follow-up; 87 patients (86%; GMI ⫽ 44, TAAC ⫽ 43) completed
the 3-month follow-up.
Aftercare Attendance at 1-Month Follow-Up
The results of the multivariate ZINB model for the 1-month
follow-up outcomes are summarized in Table 3. After multivariate
adjustment, 18.8% (n ⫽ 9) of GMI and 20.4% (n ⫽ 10) of TAAC
participants reported that they had not attended any aftercare
sessions; this proportional difference was not significant ( p ⫽
.93). Of participants who had attended aftercare (n ⫽ 78), those in
GMI averaged 25.4 (SD ⫽ 22.7) sessions compared with 16.8
(SD ⫽ 17.5) sessions for TAAC participants, a difference that
approached statistical significance ( p ⫽ .06).
Aftercare Attendance at 3-Month Follow-Up
The results of the multivariate ZINB models for the 3-month
follow-up outcomes are summarized in Table 4. After multi-
variate adjustment, 15.9% (n ⫽ 7) of GMI participants and
34.9% (n ⫽ 15) of TAAC participants reported that they had not
attended any aftercare sessions; this difference was not statis-
tically significant ( p ⫽ .30). Of patients who had attended
aftercare (n ⫽ 65), those in the GMI condition (n ⫽ 37)
attended a greater number of sessions than did those in the
TAAC condition (n ⫽ 28; GMI M ⫽ 21.1, SD ⫽ 21.3, vs.
TAAC M ⫽ 10.7, SD ⫽ 12.1, p ⫽ .02).
Table 2
DSM–IV Diagnoses for All Subjects
Diagnosis
GMI
(n ⫽ 50)
TAAC
(n ⫽ 51) p
Odds ratio
(95% CI)
Dually diagnosed 49 (98.0%) 51 (100%) .31 —
Alcohol dependent 35 (70.0%) 32 (62.7%) .44 1.4 (0.6–3.2)
Opioid dependent 8 (16.0%) 11 (21.6%) .48 0.7 (0.3–1.9)
Sedative dependent 9 (18.0%) 6 (11.8%) .38 1.6 (0.5–5.0)
Cocaine dependent 19 (38.0%) 25 (49.0%) .27 0.6 (0.3–1.4)
Marijuana dependent 12 (24.0%) 9 (17.6%) .43 1.5 (0.6–3.9)
Stimulant dependent 0 (0%) 1 (2.0%) .32 —
Inhalant dependent 1 (2.0%) 1 (2.0%) .99 —
Major depressive
disorder 29 (59.2%) 29 (56.9%) .91 1.0 (0.5–2.3)
Substance-induced mood
disorder 9 (18.4%) 8 (15.7%) .96 1.0 (0.4–2.8)
Bipolar affective
disorder 5 (10.2%) 9 (17.6%) .27 0.5 (0.2–1.7)
Anxiety disorder 18 (36.0%) 13 (25.5%) .26 1.6 (0.7–3.9)
Posttraumatic stress
disorder 8 (16.0%) 11 (22.0%) .47 0.7 (0.2–1.9)
Panic disorder 4 (8.0%) 1 (2.0%) .16 —
Social phobia 1 (2.0%) 1 (2.0%) .99 —
Generalized anxiety
disorder 11 (22.0%) 3 (6.0%) .02 4.5 (1.2–17.3)
Obsessive–compulsive
disorder 1 (2.0%) 0 (0%) .31 —
Anxiety disorder NOS 2 (4.0%) 1 (2.0%) .55 —
Schizoaffective disorder 2 (4.0%) 1 (2.0%) .55 —
Borderline personality
disorder 1 (2.0%) 1 (2.0) .57 —
Depression NOS 0 (0%) 3 (6.0%) .08 —
Mood disorder NOS 1 (2.0%) 0 (0%) .31 —
Mean substance use
diagnoses (SD) 1.68 (0.84) 1.67 (0.79) .94
Mean Axis I
nonsubstance use
diagnoses (SD) 1.33 (0.47) 1.29 (0.50) .74
Note. Totals exceed 100% because some participants have more than one
substance use and comorbid psychiatric disorder. Psychiatric diagnoses
were provided by staff psychiatrists using the Diagnostic and Statistical
Manual of Mental Disorders (4th ed.; DSM–IV; American Psychiatric
Association, 1994). Dashes indicate that values could not be estimated
because of small cell size. GMI ⫽ group motivational interviewing;
TAAC ⫽ therapist attention activity control; NOS ⫽ not otherwise spec-
ified.
Table 1
Demographic and Clinical Characteristics at Baseline
Characteristic GMI TAAC p
n 50 51
Mean age in years (SD) 38.2 (11.3) 36.8 (10.8) .46
Gender .53
Male 33 (66%) 30 (58.8%)
Female 17 (34%) 21 (41.2%)
Race .97
Caucasian American 40 (80%) 38 (74.5%)
African American 9 (18%) 11 (21.6%)
Hispanic 1 (2.0%) 1 (2.0%)
Native American 0 (0%) 1 (2.0%)
Education .54
⬍12 years 9 (18%) 9 (17.7%)
12 years 19 (38%) 16 (31.3%)
⬎12 years 22 (44%) 26 (51%)
Income .10
⬍$20,000 23 (46%) 30 (58.8%)
$20,000–$39,999 11 (22%) 11 (21.5%)
ⱖ$40,000 16 (32%) 9 (17.6%)
Did not answer 0 (0%) 1 (2%)
Marital status .45
Single 21 (42%) 19 (37.3%)
Married 10 (20%) 8 (15.7%)
Divorced or separated 18 (36%) 22 (43.1%)
Widowed 1 (2%) 2 (3.9%)
Involuntarily committed 18 (36%) 23 (45%) .34
BSI Somatic Scale, M (SD) 63.4 (14.2) 65.4 (10.3) .43
BSI Depression Scale, M (SD) 71.0 (10.5) 71.4 (9.1) .86
BSI Anxiety Scale, M (SD) 68.5 (11.0) 70.1 (10.1) .47
BSI Global Scale, M (SD) 70.6 (9.7) 71.5 (8.8) .63
Age in years at first
intoxication (SD) 14.9 (4.2) 13.5 (3.3) .07
Number of treatment sessions
12 months prior to
hospitalization (SD) 70.7 (133.2) 75.9 (132.4) .85
Weekly alcohol consumption
in SECs (SD) 64.7 (80.1) 52.4 (84.6) .17
Number of days spent
drinking alcohol (SD) 26.0 (20.7) 24.0 (20.8) .63
Number of days spent binge
drinking (SD) 22.1 (19.6) 17.7 (19.3) .26
Number of days using illicit
drugs (SD) 22.9 (23.3) 25.8 (22.8) .52
Note. GMI ⫽ group motivational interviewing; TAAC ⫽ therapist atten-
tion activity control; SEC ⫽ standard ethanol content unit.
819
BRIEF REPORTS

Alcohol and Illicit Drug Consumption at 1-Month
Follow-Up
At 1-month follow-up, 60.4% (n ⫽ 29) of GMI participants and
44.9% (n ⫽ 22) of TAAC participants reported being abstinent
from alcohol; this difference did not achieve statistical significance
( p ⫽ .12). Of participants who had used alcohol (n ⫽ 46), those in
GMI (n ⫽ 19) averaged fewer SECs (GMI M ⫽ 60.6, SD ⫽ 49.7,
vs. TAAC M ⫽ 256.3, SD ⫽ 422.9, p ⫽ .01) and reported fewer
binge-drinking days than did those in TAAC (n ⫽ 27; GMI M ⫽
4.6, SD ⫽ 3.2, vs. TAAC M ⫽ 13.9, SD ⫽ 10.5, p ⫽ .001). Group
differences on the number of drinking days approached signifi-
cance (GMI M ⫽ 5.7, SD ⫽ 4.6, vs. TAAC M ⫽ 10.9, SD ⫽ 10.3,
p ⫽ .06).
At 1-month follow-up, 68.8% (n ⫽ 33) of GMI and 51.0% (n ⫽
25) of TAAC participants reported being abstinent from illicit
drugs. This difference was not statistically significant after multi-
variate adjustment using ZINB models ( p ⫽ .70). Among partic-
ipants who reported using illicit drugs (n ⫽ 39), those in GMI (n ⫽
15), compared with those in TAAC (n ⫽ 24), averaged fewer drug
use days (GMI M ⫽ 5.2, SD ⫽ 7.2, vs. TAAC M ⫽ 13.0, SD ⫽
12.0, p ⫽ .01).
Alcohol and Illicit Drug Consumption at 3-Month
Follow-Up
Fifty percent (n ⫽ 22) of GMI and 34.9% (n ⫽ 15) of TAAC
participants reported that they had remained abstinent from alco-
hol, a proportional difference that was not statistically significant
( p ⫽ .29). Of patients who used alcohol (n ⫽ 50), those in GMI
(n ⫽ 22) averaged significantly fewer SECs than did those in
TAAC (n ⫽ 28; GMI M ⫽ 117.3, SD ⫽ 182.8, vs. TAAC M ⫽
262.3, SD ⫽ 312.9, p ⫽ .04) and fewer GMI participants reported
binge drinking (34.1% of GMI participants vs. 55.8% of TAAC
Table 3
Zero-Inflated Negative Binomial (ZINB) Outcomes for Aftercare and Substance Use at 1-Month Follow-Up
Outcome
GMI TAAC
Odds ratio
(95% CI)
Ratio of adjusted
means (95% CI)
%
abstinent
Nonzero
MSD
%
abstinent
Nonzero
MSD
SECs 60.4 60.6
*
49.7 44.9 256.3
*
422.9 0.5 (0.2–1.2) 2.9
*
(1.3–6.4)
Days spent drinking alcohol 60.4 5.7 4.6 44.9 10.9 10.3 0.6 (0.2–1.5) 1.9 (1.0–3.6)
Binge drinking days 68.8 4.6
**
3.2 61.2 13.9
**
10.5 0.7 (0.3–1.8) 2.8
**
(1.6–5.1)
Days using all illicit drugs 68.8 10.3
*
11.6 51.0 14.8
*
11.7 0.8 (0.2–2.9) 2.4
*
(1.1–5.2)
Number of all treatment
sessions attended 18.8 25.4 22.7 20.4 16.8 17.5 1.1 (0.3–4.5) 0.6 (0.4–1.0)
Note. Odds ratios are adjusted for baseline values of the outcome of interest, and they reflect the comparison between the likelihood of having a nonzero
response between the GMI and TAAC groups. Odds ratios less than one indicate that subjects in the GMI group were less likely than subjects in the TAAC
group to have a nonzero response, whereas odds ratios greater than one indicate that subjects in the GMI group were more likely than subjects in the TAAC
group to have a nonzero response. Number of all treatment sessions attended presents the percent of participants who did not attend aftercare treatment.
The ratios of means were obtained from the ZINB models and reflect the ratio between the nonzero means in the outcome of interest in the TAAC group
compared with the GMI group. As with the odds ratios reported, the ratios of means are adjusted for baseline values of the outcomes of interest. GMI ⫽
group motivational interviewing; TAAC ⫽ therapist attention activity control; SECs ⫽ standard ethanol content units.
*
p ⬍ .05 compared with TAAC.
**
p ⬍ .01 compared with TAAC.
Table 4
Zero-Inflated Negative Binomial (ZINB) Outcomes for Aftercare and Substance Use at 3-Month Follow-Up
Outcome
GMI TAAC
Odds ratio
(95% CI)
Ratio of adjusted
means (95% CI)
%
abstinent
Nonzero
MSD
%
abstinent
Nonzero
MSD
SECs 50.0 117.3
*
182.8 34.9 262.3
*
312.9 0.6 (0.2–1.6) 2.5
*
(1.0–6.1)
Days spent drinking alcohol 50.0 10.2 10.7 34.9 15.0 10.7 0.6 (0.2–1.9) 1.6 (0.8–3.1)
Binge drinking days 65.9
*
11.3 11.2 44.2
*
14.9 10.4 0.2 (0.1–0.7) 1.5 (0.8–3.1)
Days using all illicit drugs 70.5 9.4 11.5 46.5 12.3 10.9 0.0 (0.0–34.8) 1.9 (0.6–5.7)
Number of all treatment
sessions attended 15.9 21.1
*
21.3 34.9 10.7
*
12.1 4.4 (0.3–65.1) 0.5
*
(0.3–0.9)
Note. Odds ratios are adjusted for baseline values of the outcome of interest, and they reflect the comparison between the likelihood of having a nonzero
response between the GMI and TAAC groups. Odds ratios that are less than one indicate that subjects in the GMI group were less likely than subjects in
the TAAC group to have a nonzero response, whereas odds ratios greater than one indicate that subjects in the GMI group were more likely than subjects
in the TAAC group to have a nonzero response. The ratios of adjusted means were obtained from the ZINB models and reflect the ratio between the nonzero
means in the outcome of interest in the TAAC group compared with the GMI group. As with the odds ratios reported, the ratios of adjusted means are
adjusted for baseline values of the outcomes of interest. Number of all treatment sessions attended presents the proportion of participants who did not attend
aftercare treatment. GMI ⫽ group motivational interviewing; TAAC ⫽ therapist attention activity control; SECs ⫽ standard ethanol content units.
*
p ⬍ .05 compared with TAAC.
**
p ⬍ .01 compared with TAAC.
820
BRIEF REPORTS

participants, p ⫽ .02). The group difference on the number of
drinking days (GMI M ⫽ 10.2, SD ⫽ 10.7, vs. TAAC M ⫽ 15.0,
SD ⫽ 10.1) was not significant ( p ⫽ .19).
Regarding illicit drug use, 70.5% (n ⫽ 31) of participants in
GMI and 46.5% (n ⫽ 20) in TAAC were abstinent from illicit
drugs, a difference that was not statistically significant ( p ⫽ .21).
GMI (n ⫽ 13) participants averaged 9.0 (SD ⫽11.5) drug-use days
compared with 10.0 (SD ⫽ 9.7) drug-use days for TAAC (n ⫽ 23)
participants, a difference that was also not significant ( p ⫽ .22).
Convergence of Participant and Collateral Reports
Eighty-five and 72 significant others (i.e., 84.2% and 71.3% of
significant others) were available for interview by phone at 1- and
3-month follow-up, respectively. Significant other and patient re-
ports were correlated (Spearman’s ␳ coefficients ranged from .62
to .83; see Table 5), supporting the overall validity of patients’
self-reports.
Discussion
Adding GMI as opposed to TAAC to standard care did not result
in a greater proportion of patients attending aftercare or being
completely abstinent. However, as these numbers were quite small,
the lack of observed differences between conditions might reflect
low power rather than the absence of true differences. For exam-
ple, the power to detect medium effect sizes (e.g., a Cohen’s d of
0.50) among the dichotomous outcomes in our sample was only
68% at Month 1 and 63% at Month 3 (Cohen, 1977). GMI may
have increased some patients’ motivation to attend aftercare,
thereby promoting greater treatment attendance. This, in turn, was
associated with better drinking outcomes,
5
a finding consistent
with other studies (e.g., Moggi, Ouimette, Moos, & Finney, 1999).
However, some patients who received this rather modest dose of
GMI were obviously not sufficiently motivated to attend aftercare,
and failure to follow through with aftercare was associated with
unfavorable outcomes regardless of the intervention patients had
received.
Nevertheless, adding GMI to standard care yielded several pos-
itive treatment effects in this very difficult patient population.
Specifically, participants in GMI had more favorable drinking
outcomes and also showed a short-term positive effect on illicit
drug use, even though this effect was not maintained at 3-month
follow-up. As most of the drug-dependent patients in our study
were young, less educated, cocaine-abusing, minority men, we
speculate that they returned to a high-stress environment that may
have eroded treatment effects.
The present findings, although preliminary, are nevertheless
promising and suggest that GMI might be a viable treatment
modality in inpatient settings to enhance treatment outcome. Our
design provided a rigorous test of the efficacy of GMI because, by
comparing GMI with TAAC, we controlled for therapist attention
effects. The findings therefore indicate that the positive effect that
GMI added to standard care was due to specific treatment effects
rather than to nonspecific factors.
Limitations
One weakness of the present study was the use of only one
therapist in both the GMI and the TAAC conditions. This might
have led to the accidental contamination of the treatments. How-
ever, independent fidelity assessments found that the interventions
were indeed different and could be discriminated and that the
therapist adhered to the GMI manual. It is also highly unlikely that
treatment contamination occurred by staff because the study took
place before the inpatient staff received training in MI.
Another limitation was that patients were not randomized to
GMI or TAAC but assigned to weekly alternating groups. Unfor-
tunately, this type of assignment is often necessary because of the
constraints of an inpatient setting. As the conditions alternated
weekly over a long period of time, it seems implausible that any
systematic effects could have influenced the patient assignment
and thus account for the differences between GMI and TAAC we
found.
We note that the positive effects on treatment outcome cannot
be solely attributed to GMI, given that we used an incremental
design, adding GMI to standard treatment. Thus, the reported
positive effects on aftercare attendance and substance use out-
comes may have resulted from a combination of the two interven-
tions rather than from GMI alone. Also, these preliminary results
cannot be generalized to other populations (e.g., outpatients) or
larger GMI groups (here the average was 4 –5 patients per group).
Future studies should replicate the findings with larger samples
and in different settings and populations.
Conclusions
The present study provides preliminary evidence that GMI,
when added to standard treatment on an inpatient ward for dually
diagnosed patients with chemical dependency, leads to improved
treatment outcomes. This intervention may enhance aftercare at-
tendance and reduce harmful substance use in this difficult-to-treat
patient population.
5
Spearman’s correlations between aftercare treatment attendance and
substance use outcomes, for example, ranged from ⫺0.47 to ⫺0.25, p ⬍
.05, indicating that number of treatment sessions attended by the 1- and
3-month follow-ups was moderately associated with alcohol and drug use
outcomes.
References
American Psychiatric Association. (1994). Diagnostic and statistical man-
ual of mental disorders (4th ed.). Washington, DC: Author.
Table 5
Spearman’s ␳ Correlations Between Participants’ Self-Reports
and Collateral Reports by Significant Others
Outcome
1-month
follow-up
3-month
follow-up
SECs .72
***
.83
***
Days spent drinking alcohol .76
****
.82
****
Days using all illicit drugs .65
***
.62
***
Number of all treatment sessions attended .79
****
.79
***
Note. SECs ⫽ standard ethanol content units.
***
p ⬍ .001.
****
p ⬍ .0001.
821
BRIEF REPORTS

Cohen, J. (1977). Statistical power analysis for the behavioral sciences.
New York: Academic Press.
Derogatis, L. (2000). Brief Symptom Inventory 18: Administration, scoring
and procedures manual. Minneapolis, MN: Pearson.
Foote, J., DeLuca, A., Magura, S., Warner, A., Grand, A., Rosenblum, A.,
& Stahl, S. (1999). A group motivational treatment for chemical depen-
dency. Journal of Substance Abuse Treatment, 17, 181–192.
Handmaker, N. S., Packard, M., & Conforti, K. (2002). Motivational
interviewing in the treatment of dual disorders. In W. R. Miller & S.
Rollnick (Eds.), Motivational interviewing: Preparing people for change
(2nd ed., Vol. 2, pp. 362–376). New York: Guilford Press.
Hettema, J., Steele, J., & Miller, W. R. (2005). Motivational interviewing.
Annual Review of Clinical Psychology, 1, 91–111.
John, U., Veltrup, C., Driessen, M., Wetterling, T., & Dilling, H. (2003).
Motivational intervention: An individual counseling vs a group treat-
ment approach for alcohol-dependent in-patients. Alcohol and Alcohol-
ism, 38, 263–269.
Long, J. S. (1997). Regression models for categorical and limited depen-
dent variables. Thousand Oaks, CA: Sage.
Miller, W. R., Heather, N., & Hall, W. (1991). Calculating standard drink
units: International comparisons. British Journal of Addiction, 86, 43–
47.
Miller, W. R., & Marlatt, G. A. (1987). Manual supplement for the Brief
Drinker Profile, Follow-Up Drinker Profile, and Collateral Information
Form. Odessa, FL: Psychological Assessment Resources.
Miller, W. R., & Rollnick, S. (2002). Motivational interviewing: Preparing
people for change (Vol. 2). New York: Guilford Press.
Miller, W. R., Zweben, A., DiClemente, C. C., & Rychtarik, R. G. (1992).
Motivational enhancement therapy manual: A clinical research guide
for therapists treating individuals with alcohol abuse and dependence
(Vol. 2). Rockville, MD: National Institute on Alcohol Abuse and
Alcoholism.
Moggi, F., Ouimette, P. C., Moos, R. H., & Finney, J. W. (1999). Dual
diagnosis patients in substance abuse treatment: Relationship of general
coping and substance-specific coping to 1-year outcomes. Addiction, 94,
1805–1816.
Nuro, K. F., Macarelli, L., Martino, S., Ball, S. A., Baker, S. M., Roun-
saville, B. J., & Carroll, K. M. (2005). Yale Adherence and Competence
Scale (YACS-II) Guidelines (2nd ed.). West Haven, CT: Yale University
Psychotherapy Development Center.
Shrout, P. E., & Fleiss, J. L. (1979). Intraclass correlations: Uses in
assessing rater reliability. Psychological Bulletin, 86, 420– 428.
Sobell, L. C., & Sobell, M. B. (1992). Timeline follow-back: A technique
for assessing self-reported alcohol consumption. In R. Z. Litten & J. P.
Allen (Eds.), Measuring alcohol consumption: Psychosocial and bio-
chemical methods (pp. 41–72). Totowa, NJ: Humana Press.
Stata Statistical Software. (2003). Stata (Version 9.1) [Computer software].
College Station, TX: Stata Corporation.
Walters, S. T., Ogle, R., & Martin, J. E. (2002). Perils and possibilities of
group-based motivational interviewing. In W. R. Miller & S. Rollnick
(Eds.), Motivational interviewing: Preparing people for change (Vol. 2,
pp. 377–390). New York: Guilford Press.
Wechsler, H., Davenport, A., Dowdall, G., Moeykens, B., & Castillo, S.
(1994). Health and behavioral consequences of binge drinking in col-
lege: A national survey of students at 140 campuses. Journal of the
American Medical Association, 272, 1672–1677.
Wolpe, P. R., Gorton, G., Serota, R., & Sanford, B. (1993). Predicting
compliance of dual diagnosis inpatients with aftercare treatment. Hos-
pital and Community Psychiatry, 44, 45– 49.
Yalom, I. D. (1995). The theory and practice of group psychotherapy (4th
ed.). New York: Basic Books.
Received May 25, 2006
Revision received June 18, 2007
Accepted June 26, 2007 䡲
822
BRIEF REPORTS