NEW RESEARCH

Predictors of Abstinence: National

Institute of Drug Abuse Multisite
Buprenorphine/Naloxone Treatment
Trial in Opioid-Dependent Youth
Geetha A. Subramaniam, M.D., Diane Warden, Ph.D., M.B.A.,
Abu Minhajuddin, Ph.D., Marc J. Fishman, M.D., Maxine L. Stitzer, Ph.D.,
Bryon Adinoff, M.D., Madhukar Trivedi, M.D., Roger Weiss, M.D.,
Jennifer Potter, Ph.D., Sabrina A. Poole, M.S., George E. Woody, M.D.

Objective: To examine predictors of opioid abstinence in buprenorphine/naloxone (Bup/

Nal)-assisted psychosocial treatment for opioid-dependent youth. Method: Secondary anal-
yses were performed of data from 152 youth (15–21 years old) randomly assigned to 12 weeks
of extended Bup/Nal therapy or up to 2 weeks of Bup/Nal detoxification with weekly
individual and group drug counseling. Logistic regression models were constructed to identify
baseline and during-treatment predictors of opioid-positive urine (OPU) at week 12. Predictors
were selected based on significance or trend toward significance (i.e., p ⬍ .1), and backward
stepwise selection was used, controlling for treatment group, to produce final independent
predictors at p ⱕ .05. Results: Youth presenting to treatment with previous 30-day injection
drug use and more active medical/psychiatric problems were less likely to have a week-12 OPU.
Those with early treatment opioid abstinence (i.e., weeks 1 and 2) and those who received
additional nonstudy treatments during the study were less likely to have a week-12 OPU and those
not completing 12 weeks of treatment were more likely to have an OPU. Conclusions: Youth
with advanced illness (i.e., reporting injection drug use and additional health problems) and
those receiving ancillary treatments to augment study treatment were more likely to have lower
opioid use. Treatment success in the first 2 weeks and completion of 12 weeks of treatment were
associated with lower rates of OPU. These findings suggest that youth with advanced illness
respond well to Bup/Nal treatment and identify options for tailoring treatment for opioid-
dependent youth presenting at community-based settings. Clinical trial registration information—
Buprenorphine/Naloxone-Facilitated Rehabilitation for Opioid Dependent Adolescents; http://
www.clinicaltrials.gov; NCT00078130. J. Am. Acad. Child Adolesc. Psychiatry, 2011;50(11):
1120 –1128. Key Words: treatment predictors, opioid dependent youth, buprenorphine
treatment

P
rescription opioids are second only to mar-
ijuana as the most commonly used illicit
substances among high school seniors.1
During the previous 10 years, annual use preva-
lence of non-heroin opioids among 12th graders
has increased from 6% to 9%, whereas heroin use
has hovered around 1%.1 Treatment-seeking
youth with opioid-use disorders2 and opioid
problem use (added to cannabis/alcohol prob-
lem use)3 have complex needs, presenting with
higher rates of psychiatric and medical (risk of
human immunodeficiency virus [HIV] and hepati-
tis C infection) conditions, polysubstance use, legal
problems, and greater risk for school dropout com-
pared with youth with marijuana and/or alcohol-
use disorders/problem use. Opioid analgesic-
related emergency department visits among
youth younger than 21 years have increased from
17,267 in 2004 to 53,668 in 2008, reflecting the
enormity of this problem.4 Despite these trends,
effective treatments for opioid-dependent youth
are just emerging.
Methadone and buprenorphine, medications
with opioid-agonist actions, are highly effective
in treating opioid-dependent adults (see Co-
chrane reviews5,6), but only a few uncontrolled

JOURNAL OF THE AMERICAN ACADEMY OF CHILD & ADOLESCENT PSYCHIATRY

1120 www.jaacap.org VOLUME 50 NUMBER 11 NOVEMBER 2011

studies from the 1970s have been published,
providing limited evidence for the efficacy of
methadone maintenance in youth. Further, meth-
adone is often not a feasible option for youth
younger than 18 years because of regulatory
restrictions and the need to dispense it only at
specialized opioid treatment programs.7 Bu-
prenorphine, a partial ␮-agonist, has shown
promise in treating opioid-dependent youth in
two recent National Institute of Drug Abuse
(NIDA)-funded randomized controlled outpa-
tient trials. The first study with opioid-dependent
youth (n ⫽ 36, 13–18 years old), conducted at a
single site, showed that a 28-day treatment epi-
sode with buprenorphine was associated with
significantly less opioid use and better treatment
retention compared with clonidine.8 A more re-
cent multisite trial funded by the NIDA Clinical
Trials Network randomized opioid-dependent
youth (15–21 years old) to a 14-day buprenor-
phine/naloxone (Bup/Nal) detoxification (DETOX)
or 12-week extended Bup/Nal therapy (BUP)
with a dose taper beginning at week 9 and
ending during week 12.9 The BUP group fared
significantly better than the DETOX group on the
primary outcome of opioid use, and almost all
secondary outcomes, including better treatment
retention, less injection drug use (IDU), and less
cocaine and marijuana use. At week 12, 53 in the
DETOX group had opioid-positive urine (OPU)
results (51%; 95% confidence interval [CI] 35%–
67%) versus 49 in the BUP group (43%; 95% CI
29%–57%). At week 12, 16 of 78 DETOX patients
(20.5%) remained in treatment versus 52 of 74
BUP patients (70%; ␹2 ⫽ 32.90, p ⱕ .001).
These efficacy studies are important because
they provide mean results for the overall sample,
but they do not illustrate the heterogeneity of the
sample or identify specific subgroups of patients
who may have greater benefits. An exploration of
baseline and during-treatment factors of better
outcomes may provide additional information to
help physicians tailor these interventions to indi-
vidual patients.10,11 In studies with adult opioid-
dependent patients treated with Bup/Nal, so-
ciodemographic factors such as female gender,
fewer days of paid employment, illness seve-
rity (as indicated by high depressive symptoms
and greater opioid withdrawal), and during-
treatment factors such as higher medication dose
and better adherence were associated with
greater abstinence.12-15 Concomitant use of co-
caine or marijuana was not related to outcome,
JOURNAL OF THE AMERICAN ACADEMY OF CHILD & ADOLESCENT PSYCHIATRY
VOLUME 50 NUMBER 11 NOVEMBER 2011
PREDICTORS OF YOUTH OPIOID TREATMENT OUTCOMES
whereas tobacco use was associated with poorer
outcome15 or unrelated to outcome.16 In the only
related study, Motamed et al.17 conducted sec-
ondary analyses of the study by Marsch et al.8
to determine if outcomes were different for
prescription opioid-dependent (n ⫽ 17) versus
heroin-dependent (n ⫽ 19) youth; they found no
differences in opioid abstinence or treatment
retention.
Given the lack of information on predictors
of outcome for Bup/Nal treatment of opioid-
dependent youth, secondary analyses of the
larger (N ⫽ 152) multisite Bup/Nal study9 of
opioid-dependent youth were conducted to
identify baseline and during-treatment predic-
tors associated with lower opioid use that
might provide insights into potential mecha-
nisms of improvement and/or information to
help guide patient management and clinical
decision-making in this population.
METHOD
Participants and Study Procedures
The study methods were approved by the institutional
review boards of all participating institutions and
reported in the primary outcome article.9 Participants
were treatment-seeking youth 15 to 21 years old who
met DSM-IV criteria for opioid dependence with phys-
iologic features and were recruited from six community-
based treatment sites from July 2003 to December
2005. Participants (N ⫽ 152) were randomized to 12
weeks of Bup/Nal (BUP) with a dose taper beginning
in week 9 and ending in week 12 or up to 2 weeks of
Bup/Nal (DETOX). The two groups were offered one
weekly individual and one group counseling session
during the 12-week active study phase guided by the
individual and group drug counseling manuals, which
encouraged making positive relationships and stop-
ping drug use, taking medication as prescribed,
tolerating stressful events without using drugs,
keeping appointments, teaching ways to avoid drug-
using situations, educating about addiction, giving
positive feedback for achieving goals, referring for
treatment of associated problems, and participating
in age-appropriate self-help groups.18 The partici-
pants were assessed at baseline and weeks 4, 8, and
12 (active treatment phase).
Study Treatment
Details of medication dosing and administration and
counseling are described elsewhere.9,19,20 Participants
were asked to abstain from opioids for at least 6 hours
and present with opioid withdrawal before their first
dose of Bup/Nal. Dosing was given under direct
www.jaacap.org 1121
SUBRAMANIAM et al.
observation 5 to 7 days per week, depending on the
site. Participants were inducted on day 1 with a
maximum dose of 8 mg. On days 2 and 3, patients
received the total dose from the previous day unless
they were overmedicated and as clinically needed
received additional doses, as needed. The maximum
dose was determined a priori as 14 mg for DETOX and
24 mg for BUP. A dose taper was begun in week 1 or
2 in the DETOX group and completed by day 14 and
was begun in week 10 and completed by the end of
week 12 in the BUP group. In addition, all participants
were asked to attend weekly manual-guided individ-
ual and group drug counseling for 12 weeks.18
Baseline Demographic and Clinical Predictors
The NIDA Clinical Trials Network Baseline Demo-
graphics Form was used to collect information on
gender, race (white/non-white), age, years of educa-
tion, employment, and use of substances (alcohol,
cannabis, cocaine, and cigarettes) in the previous
month (days) and lifetime (years). At baseline, the
Substance Dependence Severity Scale Lite21 docu-
mented opioid dependence with physiologic features
and provided information on the type of opioid iden-
tified as the main problem (i.e., heroin, prescription
opioids, or both). The Risk Behavior Survey22,23
provided information on previous 30-day HIV risk
behaviors (e.g., IDU of opioids, cocaine, and/or am-
phetamine, sexual risk behaviors, etc.). The Youth
Self-Report24 and the Young Adult Self-Report25 pro-
vided baseline information on internalizing and exter-
nalizing problems in the previous 90 days. The partic-
ipant’s health status was assessed at baseline and at
week 12: the medical history provided information on
lifetime and current medical and psychiatric condi-
tions; liver enzyme levels (i.e., aspartate aminotrans-
ferase, alanine aminotransferase, gamma-glutamyl
transferase, and lactate dehydrogenase) and serum
hepatitis B and C titers were obtained at baseline and
week 12. Having active (current) medical and/or psy-
chiatric problems was defined as reporting such
problems at the time of baseline medical history.
Nonstudy treatments (e.g., additional outpatient vis-
its, hospitalizations, etc.) and medications taken
before study entry and during the study were doc-
umented, including names, doses, and duration of
administration.
During-Treatment Predictors
Daily logs recorded the Bup/Nal dose prescribed and
taken and medication adherence for each day. In the
present analyses, medication adherence was defined as
taking at least five of seven doses per week. Doses in
the BUP group were categorized into low (⬍12 mg),
moderate (12–16 mg), and high (17–24 mg), the latter
including one participant who received a maximum
dose of 32 mg. Doses in the DETOX group, where the
JOURNAL
1122 www.jaacap.org
maximum recommended dose was 14 mg, were cate-
gorized into low (⬍10 mg), moderate (10 to ⬍14 mg),
and high (ⱖ14 mg), the latter including one participant
whose maximum dose was 20 mg. Different dosing
categories were used for the two groups because the
recommended maximum doses were different. The
number and type of counseling appointments kept
were documented. Treatment completion was defined
as continuing in study treatment for at least 77 days
(i.e., 12 weeks). Information on medical, psychiatric,
and addiction services that the participant received
outside the assigned treatment condition was collected
at baseline (prior 30 days) and weekly during the
study. Signs and symptoms of withdrawal severity
during the first 2 dosing weeks were measured by the
Short Opiate Withdrawal Scale.26 Detailed information
on adverse events and serious adverse events was
collected weekly during the first 12 weeks. Urinalyses
for drugs of abuse were performed onsite using the
SureStep (Alere Inc, Orlando, FL) drug screen card
(which tests all opioids except oxycodone) and the
Rapid One OXY (RDP RapidDetect Inc, Poteau, OK)
(which tests for oxycodone).
Data Analyses
As reported previously,10 the primary outcome for
these secondary analyses was the presence of OPU at
week 12. Preliminary bivariate analyses were con-
ducted from a list of baseline and during-treatment
variables to determine those associated with an OPU.
Comparisons between those with OPU on baseline and
during-treatment factors were performed using Stu-
dent t tests for continuous measures and ␹2 tests for
categorical variables. If the cell frequencies were too
low for the ␹2 approximation to be valid, the Fisher
exact test was used. Analyses included all participants
randomized to treatment, consistent with an intent-to-
treat approach. Two separate logistic regression mod-
els were constructed to identify independent predic-
tors of opioid use while controlling for treatment
group assignment. In the first model, five baseline
factors were entered that were significant or approach-
ing significance (p ⱕ .10) in preliminary bivariate
analyses; the second model was constructed using six
during-treatment factors with similar significance (p ⱕ
.10). The number of active medical/psychiatric prob-
lems was excluded from the second model because
these data were collected only at the end of treatment
(i.e., the week-12 visit), and the small sample limited
the number of variables that could be included in the
model. Backward stepwise selection was used to refine
the model with a threshold p value of .05 for including
variables in the final predictive model.
Urine drug screen results were available for 59% of
participants at week 12. An intent-to-treat sample was
used, with missing urine tests imputed as positive.
Because there were no significant differences between
BUP and DETOX in percent samples available or in the
OF THE AMERICAN ACADEMY OF CHILD & ADOLESCENT PSYCHIATRY
VOLUME 50 NUMBER 11 NOVEMBER 2011
 PREDICTORS OF YOUTH OPIOID TREATMENT OUTCOMES

baseline characteristics (described earlier) between attending at least one study therapy session
those who submitted a urine sample (i.e., the com- (␹21 ⫽ 8.45, p ⫽ .004), and adhering to medication
pleter sample) and those who did not (except for mean (Fisher exact test, p ⬍ .0001) were associated with
scores on internalizing symptoms), the two samples lower rates of OPU at week 12. However, mark-
were merged for analyses. Data were also reanalyzed
ers related to distress, such as the number of
including only those participants who provided a
urine sample at week 12. Analyses were performed
moderate to severe study-related serious adverse
using SAS 9.2.27 events or the number of withdrawal symptoms
in the first 2 weeks, were not associated with
treatment outcome (Table 2).
RESULTS Across 12 Weeks of Treatment Markers. Treatment
characteristics that were significantly associated
Study Sample Characteristics
with lower rates of week-12 OPUs consisted of
The age range was 15 to 21 years with a sample
having an elevated liver enzyme (␹21 ⫽ 11.1, p ⫽
mean of 19.2 years (standard deviation [SD], 1.5);
.001), receiving any nonstudy treatment services
17% were younger than 18 years; 74% were
(␹21 ⫽ 17.5, p ⬍ .0001) or concomitant medica-
Caucasian, 24% were Hispanic, and 26% of an-
tions (␹21 ⫽ 12.06, p ⫽ .0005), attending more
other race; 40% reported previous 30-day heroin-
study counseling sessions (t148 ⫽ 4.93, p ⬍ .0001),
only use, 24% opioid analgesics/other opioids
having more active medical and/or psychiatric
only, and 36% both types of opioids; and 16%
problems at the week-12 visit (t150 ⫽ 5.43, p ⬍
reported previous 30-day IDU. Mean years of
.0001), and completing 12 weeks of treatment
education was 11.2 (SD, 1.6), and 75% reported
(␹21 ⫽ 9.62, p ⫽ .002). Maximum dose of
being employed in the previous 3 years. Mean
Bup/Nal, high/intermediate/low Bup/Nal
maximum doses of Bup/Nal were 15.1 mg (SD,
dosing ranges, and reports of serious adverse
4.9) in the BUP group and 11.5 mg (SD, 2.9) in the
events (study medication-related or not) were
DETOX group. Only 44% of the study sample
unrelated to outcome (Table 2).
completed 12 weeks of treatment.

Independent Baseline and During-Treatment

Baseline Predictors
Characteristics and Opioid Abstinence
Among the sociodemographic characteristics ex-
In the first logistic regression model (controlling
amined (Table 1), age, race, previous 30-day
for treatment group assignment), the following
employment, and years of education were not
five baseline factors that were significant or ap-
related to outcome, but there was a trend for
proached significance were included: previous
female subjects (␹21 ⫽ 3.92, p ⫽ 0.06) to have
30-day IDU (yes/no), number of active comorbid
lower rates of OPU at week 12. Neither previous
medical or psychiatric problems, gender (female/
30-day nor lifetime use (latter results not shown)
male), elevated liver enzymes (yes/no), and
of any substance (including type of opioid) was
number of internalizing problems. In the final
related to outcome. Among the clinical factors
model (Table 3), those reporting previous 30-day
examined, reporting previous 30-day IDU of opi-
IDU (odds ratio [OR] 0.32, 95% CI 0.13– 0.80) and
oids, cocaine, or amphetamines (␹21 ⫽ 7.27, p ⫽
having more active medical/psychiatric prob-
.007), developing more active medical and/or
lems (OR 0.77, 95% CI 0.60 – 0.98) were less likely
psychiatric problems (t150 ⫽ 2.22, p ⫽ .028), and
to have an OPU.
higher mean scores on the Internalizing Problem
In the second logistic regression model (also
subscale (t141 ⫽ 2.24, p ⫽ .027) were linked to
controlling for treatment group assignment), the
lower rates of OPU. There was a trend toward
following six during-treatment factors that were
significance for having an elevated liver enzyme
significant or approached significance were in-
and lower rates of OPU (␹21 ⫽ 3.37, p ⫽ .067).
cluded: opioid-negative urine at weeks 1 and 2
Neither nonstudy medications nor treatment ser-
(yes/no), any study therapy attendance at weeks
vices received in the 30 days before baseline (results
1 and 2 (yes/no), medication compliance during
not shown) were associated with outcome.
weeks 1 and 2, receiving any nonstudy medica-
tion (yes/no), dropping out before the 12th week
During-Treatment Characteristics of treatment (yes/no), and receiving any non-
Early Treatment Markers (Weeks 1 and 2). Having study treatment services (yes/no). In the final
opioid-negative urine (␹21 ⫽ 6.85, p ⫽ .009), model (Table 3), those having a negative urine

JOURNAL OF THE AMERICAN ACADEMY OF CHILD & ADOLESCENT PSYCHIATRY

VOLUME 50 NUMBER 11 NOVEMBER 2011 www.jaacap.org 1123
SUBRAMANIAM et al.

TABLE 1 Bivariate Comparison of Baseline Factors With Week-12 Opioid-Positive Urine Tests
Intent-to-Treat Sample

Urine Negative at 12 wk Urine Positive at 12 wk

Variable n (%)/Mean (SD) n (%)/Mean (SD) Test Statistic p

Treatment group 3.92 .048

BUP 28 (60.9) 46 (43.4)
DETOX 18 (39.1) 60 (56.6)
Baseline: sociodemographics
Age ⱖ18 y 0.282 .596
Yes 37 (80.4) 89 (84.0)
No 9 (19.6) 17 (16.0)
Gender 3.540 .060
Female 24 (52.2) 38 (35.8)
Male 22 (47.8) 68 (64.2)
Race 1.550 .213
Caucasian 37 (80.4) 75 (70.8)
Non-Caucasian 9 (19.6) 31 (29.2)
Employment at 30 days 0.287 .592
Yes 23 (50.0) 58 (54.7)
No 23 (50.0) 48 (45.3)
Years of education 11.24 (1.5) 11.11 (1.6) 0.46 .647
Baseline: substance abuse characteristics
Opioid use at 30 days 2.515 .284
Heroin only 15 (32.6) 45 (42.9)
Non-heroin opioids 10 (21.7) 26 (24.8)
Both 21 (45.7) 34 (32.4)
Opioid use during lifetime 0.437 .804
Heroin only 17 (37.0) 43 (42.6)
Non-heroin opioids 13 (28.3) 27 (26.7)
Both 16 (34.8) 31 (30.7)
Cocaine use at 30 days 0.287 .592
Yes 23 (50.0) 48 (45.3)
No 23 (50.0) 58 (54.7)
Marijuana use at 30 days 0.829 .363
Yes 33 (71.7) 68 (64.2)
No 13 (28.3) 38 (35.8)
Tobacco use at 30 days 1.493 .222
Yes 5 (10.9) 20 (18.9)
No 41 (89.1) 86 (81.1)
Polysubstance use at 30 days N/A .152
Yes 44 (95.7) 92 (87.6)
No 2 (4.3) 13 (12.4)
Baseline: comorbid conditions
No. of active medical or psychiatric problems 1.52 (1.6) 1.0 (1.3) 2.22 .028
Liver enzyme 3.367 .067
Any enzyme elevated 16 (34.8) 22 (20.8)
None 30 (65.2) 84 (79.2)
Hepatitis C infected 0.046 .831
Yes 9 (19.6) 19 (18.1)
No 37 (80.4) 86 (81.9)
Previous 30-day injection drug use 7.274 .007
Yes 13 (28.3) 11 (10.7)
No 33 (71.7) 92 (89.3)
Mean scores on internalizing problems 65.78 (9.1) 61.08 (12.7) 2.24 .027
Mean scores on externalizing problems 59.59 (8.5) 57.66 (9.8) 1.2 .231

Note: Test statistic was ␹2 or Fisher exact test for categorical variables and Student t tests for continuous variables. P values in bold indicate statistical
significance or values approaching statistical significance. BUP ⫽ 12-week extended buprenorphine/naloxone therapy assignment; DETOX ⫽ 14-day
buprenorphine/naloxone detoxification assignment; N/A ⫽ not available.

JOURNAL OF THE AMERICAN ACADEMY OF CHILD & ADOLESCENT PSYCHIATRY

1124 www.jaacap.org VOLUME 50 NUMBER 11 NOVEMBER 2011
 PREDICTORS OF YOUTH OPIOID TREATMENT OUTCOMES

TABLE 2 Bivariate Comparisons of During-Treatment Factors With Week-12 Opioid-Positive Urine Tests
Intent-to-Treat Sample

Urine Negative Urine Positive at

at 12 wk 12 wk
Variable n (%)/Mean (SD) n (%)/Mean (SD) Test Statistic p

Early treatment predictors (1 and 2 wk)

Opioid-negative urine 6.854 .009
Yes 23 (57.5) 22 (31.9)
No 17 (42.5) 47 (68.1)
Any study therapy compliant 8.448 .004
Yes 36 (78.3) 53 (53.5)
No 10 (21.7) 46 (46.5)
Medication compliant N/A <.0001
Yes 42 (91.3) 63 (61.8)
No 4 (8.7) 39 (38.2)
Any moderate to severe AEs 2.156 .142
Yes 11 (23.9) 15 (14.2)
No 35 (76.1) 91 (85.8)
Mean SOWS scores 5.63 (4.39) 7.08 (6.49) 1.34 .184
12-week treatment predictors
Health status-related characteristics
Any serious AEs N/A .724
Yes 2 (4.3) 7 (6.6)
No 44 (95.7) 99 (93.4)
Liver enzymes (at 4 or 12 wk) 11.107 .001
Any enzyme elevated 24 (52.2) 26 (24.5)
None 22 (47.8) 80 (75.5)
No. of active medical or psychiatric problems (12 wk) 1.52 (1.4) 0.44 (1.0) 5.43 <.0001
Study treatment characteristics
Bup/Nal dosing groups 0.279 .870
Low 17 (37.0) 43 (41.3)
Med 20 (43.5) 43 (41.3)
High 9 (19.6) 18 (17.3)
Maximum Bup/Nal dose 14.13 (4.4) 12.85 (4.4) 1.65 .104
No. of any study therapy sessions 18.24 (16.6) 7.77 (9.3) 4.93 <.0001
Treatment dropout (⬍12 wk) 9.624 .002
Yes 17 (37.0) 68 (64.2)
No 29 (63.0) 38 (35.8)
Outside study treatment characteristics
Any nonstudy treatment services 17.525 <.0001
Yes 15 (32.6) 7 (6.6)
No 31 (67.4) 99 (93.4)
Receiving nonstudy medications 12.055 .0005
Yes 38 (82.6) 56 (59.6)
No 8 (17.4) 50 (47.2)

Note: Test statistic was ␹2 or Fisher exact test for categorical variables and Student t tests for continuous variables. P values in bold indicate statistical
significance or values approaching statistical significance. AE ⫽ adverse event; BUP ⫽ 12-week extended buprenorphine/naloxone therapy
assignment; Bup/Nal ⫽ buprenorphine/naloxone; N/A ⫽ not available; SOWS ⫽ Short Opioid Withdrawal Scale.

test in weeks 1 and 2 (OR 0.24, 95% CI 0.09 – 0.66),
those receiving nonstudy treatment services (OR
0.11, 95% CI 0.03– 0.43), and those receiving any
nonstudy medications (OR 0.11, 95% CI 0.03–
0.41) were less likely to have an OPU. Partici-
pants who dropped out of treatment before the
12th week (OR 4.71, 95% CI 1.16 –19.06) were
more likely to have an OPU.
Logistic regression analyses were repeated us-
ing the completer sample (i.e., those who pro-

JOURNAL OF THE AMERICAN ACADEMY OF CHILD & ADOLESCENT PSYCHIATRY

VOLUME 50 NUMBER 11 NOVEMBER 2011 www.jaacap.org 1125
SUBRAMANIAM et al.
TABLE 3 Multivariate Models: Independent Predictors of Week-12 Opioid-Positive Urine Tests
Predictors
Baseline factors
Treatment group BUP vs. DETOX
Injection drug use in previous 30 days
No. of active medical or psychiatric systems
During-treatment factors
Treatment group BUP vs. DETOX
Opioid-negative urine in first 2 weeks
Any nonstudy medications during treatment
Dropping out of treatment
Receiving any nonstudy treatment
Note: BUP ⫽ 12-week extended buprenorphine/naloxone therapy assignment; DETOX ⫽ 14-day buprenorphine/naloxone detoxification assignment;
ITT ⫽ intent-to-treat sample.
vided a urine drug screen result at week 12).
Among baseline factors, previous 30-day IDU
remained significant (OR 0.28, 95% CI 0.08 – 0.93).
Among the during-treatment factors, only two
factors significantly decreased the risk of having
an OPU at week 12: providing opioid-negative
urine in weeks 1 and 2 (OR 0.22, 95% CI 0.07–
0.70) and receiving any nonstudy medications
during treatment (OR 0.10, 95% CI 0.03– 0.41).
DISCUSSION
This article presents new information on baseline
and during-treatment factors related to treatment
success in this outpatient, community-based, mul-
tisite trial of Bup/Nal-treated opioid-dependent
adolescents and young adults.
First, IDU at baseline was a significant and
independent predictor of lower rates of OPU at
week 12. This finding is consistent with prior
research that injection drug users had better
Bup/Nal outcomes or were more likely to believe
that it would be helpful,15,28,29 lending further
support for the role of Bup/Nal as a potentially
effective tool in decreasing HIV and hepatitis C
infections and risk.28 Second, having more active
medical and/or psychiatric problems at treat-
ment entry emerged as a significant independent
predictor of lower rates of OPU. Higher mean
baseline scores on self-reported internalizing
problems were associated with better opioid-use
outcomes (in bivariate analyses) but were not a
significant independent predictor of lower rates
of OPU. This finding is concordant with reports
of adult patients treated with buprenorphine
showing that higher levels of depression was
1126 www.jaacap.org
ITT Sample Completer Sample
Odds Ratio (95% Confidence Interval) Odds Ratio (95% Confidence Interval)
0.540 (0.258–1.129) 0.685 (0.287–1.635)
0.320 (0.127–0.802) 0.276 (0.082–0.934)
0.766 (0.599–0.980) —
1.325 (0.336–5.231) 0.960 (0.331–2.782)
0.241 (0.089–0.656) 0.222 (0.070–0.701)
0.112 (0.030–0.419) 0.103 (0.025–0.414)
4.705 (1.162–19.057) —
0.114 (0.031–0.426) —
associated with better opioid outcomes.12,13,15 For
youth, the association between IDU and high
distress levels and better treatment outcome may
be explained by their awareness of being in a
downward spiral and being tired of devoting so
much time and resources to obtaining and using
drugs to the exclusion of prosocial activities. This
could be a motivating factor for treatment.
The third significant finding was that Bup/
Nal treatment led to similar rates of opioid absti-
nence regardless of the type of opioid subjects
reported using (i.e., heroin, non-heroin prescrip-
tion opioid analgesic, or both), consistent with
findings from another adolescent opioid treat-
ment study.17 Also nonsignificant were baseline
predictors such as race, education/employment
status, and concomitant cocaine, tobacco or mar-
ijuana use, with a trend toward significance for
gender and elevated liver enzymes, in contrast to
results from adult treatment studies.12-16
The during-treatment factors associated with
treatment success spanned three areas: nonstudy
medications and other nonstudy treatment ser-
vices, early treatment phase opioid abstinence, and
study treatment completion. The significance of
receiving ancillary treatments and medications (to
ease withdrawal symptoms, insomnia, pain, and co-
occurring psychiatric symptoms) to augment Bup/
Nal interventions was not supported with adult
patients.15,30 It is likely that the improved outcomes
for patients who received nonstudy medications
and/or treatment reflect benefits of treating other
medical/psychiatric disorders while in the study (as
shown in studies with adults12) and the important
benefits of receiving additional treatment elsewhere
JOURNAL OF THE AMERICAN ACADEMY OF CHILD & ADOLESCENT PSYCHIATRY
VOLUME 50 NUMBER 11 NOVEMBER 2011

for co-occurring problems even after dropping out of
the study.
Achieving opioid abstinence during the early
treatment phase may serve as an important marker
of treatment success, as in a previous study with
adults,31 and may have been driven by a motiva-
tion to get well among those severely addicted,
resulting in adherence to medication and counseling.
Similarly, it was no surprise that treatment comple-
tion was associated with better outcomes because
longer periods of treatment participation have consis-
tently been linked to better buprenorphine and
other substance-abuse treatment outcomes.32-34
This study was not adequately powered to de-
tect a clinically significant interaction with all the
baseline and during-treatment factors examined.
Although the lack of statistical significance in the
results may be a result of sample size, the present
analyses were meant to generate hypotheses for
future study. Future adequately powered studies
are needed to replicate these findings, reevaluate
the nonsignificant findings from this study, evalu-
ate specific co-occurring psychiatric disorders, con-
sider study endpoints/outcomes other than end-of-
study urine results, and include other predictors,
e.g., biological and genetic markers that were not
examined in this study. Although youth younger
than 18 years (n ⫽ 26, 18%) did not fare differently
from those who were older, this finding may be an
artifact of the small sample. The low rates of study
treatment completion/retention may have ad-
versely affected the outcomes. Consistent with the
primary study, positive results for missing urine
results were conservatively imputed, which may
have incorrectly estimated those who achieved
recovery.
The primary study demonstrated the efficacy
of 12 weeks of Bup/Nal in decreasing opioid use
and improving treatment retention among opioid-
dependent youth 15 to 21 years old. The present
study has contributed new and important clinical
information on baseline and during-treatment fac-
tors that were linked to lower rates of OPU. Pa-
tients presenting with an advanced opioid-use pat-
tern (i.e., IDU and other health issues) were more
likely to have lower rates of OPU at week 12,
suggesting that they can respond well to outpatient
Bup/Nal treatment. Those who were able to
achieve early opioid abstinence, receive supple-
mental treatment services and medications outside
the study, and remain in treatment for the entire 12
weeks were more likely to have better opioid
JOURNAL OF THE AMERICAN ACADEMY OF CHILD & ADOLESCENT PSYCHIATRY
VOLUME 50 NUMBER 11 NOVEMBER 2011
PREDICTORS OF YOUTH OPIOID TREATMENT OUTCOMES
outcomes. These new findings have the potential to
facilitate tailored treatments for opioid-dependent
youth and inform the design of future treatment
trials for this largely understudied population. &
Accepted July 15, 2011.
Dr. Subramaniam is with the National Institute on Drug Abuse. Drs.
Subramaniam, Fishman, and Stitzer are with Johns Hopkins University
School of Medicine. Drs. Warden, Minhajuddin, Adinoff, and Trivedi
are with the University of Texas Southwestern Medical Center. Dr.
Weiss is with Harvard Medical School and McLean Hospital. Dr.
Potter is with University of Texas Health Science Center. Ms. Poole and
Dr. Woody are with the University of Pennsylvania and Treatment
Research Institute. Dr. Adinoff is also with the Veterans Affains North
Texas Health Care System. Dr. Fishman is also with Mountain Manor
Treatment Center.
This study was supported by the following grants from the National
Institutes on Drug Abuse: U10-DA13043 and KO5 DA-17009
(G.E.W.) to the University of Pennsylvania, U10-013034 (M.L.S.) and
K12-DA000357 (G.A.S., Paula Riggs, M.D.) to Johns Hopkins
University, U10-DA020024 (B.A.) to University of Texas Southwestern
Medical Center at Dallas, and U10-DA015831 (R.W.) and K24-
DA022288 (R.W.) to McLean Hospital.
Disclosure: Dr. Warden holds stock in Pfizer and Bristol-Myers Squibb.
He has received funding from the National Alliance for Research on
Schizophrenia and Depression. Dr. Fishman is the medical director of
Mountain Manor Treatment Center (MMTC), one of multiple research
sites in this study. He is a beneficiary of the trust that owns MMTC and
serves on the governing board of the trust and the board of directors of
MMTC. The terms of Dr. Fishman’s potential conflict of interest in
research are managed by Johns Hopkins University in accordance with
its conflict of interest policies. Dr. Weiss has received grants from the
National Institute of Drug Abuse (NIDA). He has served as a consultant
to Titan Pharmaceuticals. Dr. Adinoff has received grant support from
the National Institute of Alcohol Abuse and Alcoholism, NIDA, and the
Department of Veterans Affairs. He has served as a consultant to
Shook, Hardy, and Bacon LLP and Paul J. Passante, P.C. He has
received honoraria from the University of New Mexico, the Medical
University of South Carolina, the American Institute of Biological
Sciences, the American Academy of Addiction Psychiatry, the Meth-
odist Medical Center, Vanderbilt University, the University of North
Texas Health Care System, John Peter Smith Hospital, and Texas Tech
University. Dr. Trivedi has served as a consultant for Abbott,
Alkermes, AstraZeneca, Axon Advisors, Bristol-Myers Squibb,
Cephalon, CME Institute of Physicians, Eli Lilly and Co., Evotek,
Forest, GlaxoSmithKline, Johnson and Johnson, Lundbeck, MedAvante,
Neuronetics, Ostuka, Pamlab, Pfizer, PgxHealth, Rexahn, Shire, Takeda,
Tal Medical/Puretech Venture, and Transcept. He has served on the
speakers’ bureau for Axon Advisors, Bristol-Myers Squibb, CME Institute
of Physicians, Eli Lilly and Co., Forest, GlaxoSmithKline, Lundbeck,
MedAvante, Otsuka, Pamlab, Pfizer, PgxHealth, Rexahn, and Takeda.
He has received research support from the Agency for Healthcare
Research and Quality, the National Institute of Mental Health, NIDA,
Naurex, Targacept, and Valient. Dr. Woody has served on the advisory
board of the Researched Abuse, Diversion, and Addiction-Related Surveil-
lance (RADARS) System. Drs. Subramaniam, Minhajuddin, Stitzer, and
Potter, and Ms. Poole report no biomedical financial interests or potential
conflicts of interest.
Correspondence to Geetha A. Subramaniam, M.D., Division of
Clinical Neuroscience and Behavioral Research, National Institute
on Drug Abuse, 6001 Executive Blvd., Room 3173, MSC 9593,
Bethesda, MD 20892; e-mail: geetha.subramaniam@nih.gov
0890-8567/$36.00/©2011 American Academy of Child and
Adolescent Psychiatry
DOI: 10.1016/j.jaac.2011.07.010
www.jaacap.org 1127
SUBRAMANIAM et al.
REFERENCES
1. Johnston LD, O’Malley PM, Bachman JG, Schulenberg JE. Moni-
toring the Future National Results on Adolescent Drug Use: Overview
of Key Findings, 2009. Publication No. 10-7583. Bethesda, MD: 18.
National Institute on Drug Abuse; 2010.
2. Subramaniam GA, Stitzer MA, Woody GE, Fishman MJ, Kolod-
ner K. Clinical characteristics of treatment-seeking adolescents
with opioid versus cannabis/alcohol use disorders. Drug Alcohol 19.
Depend. 2009;99:141-149.
3. Subramaniam GA, Ives ML, Stitzer ML, Dennis ML. The added
risk of opioid problem use among treatment-seeking youth with
marijuana and/or alcohol problem use. Addiction. 2010;105:686-
698. 20.
4. DAWN. Detailed tables: national estimates, drug-related emer-
gency department visits for 2004 –2009. 2009. https://dawninfo.
samhsa.gov/data//. Accessed February 1, 2011.
5. Mattick RP, Kimber J, Breen C, Davoli M. Buprenorphine main- 21.
tenance versus placebo or methadone maintenance for opioid
dependence. Cochrane Database Syst Rev. 2008;2:CD002207.
6. Mattick RP, Breen C, Kimber J, Davoli M. Methadone mainte-
nance therapy versus no opioid replacement therapy for opioid 22.
dependence. Cochrane Database Syst Rev. 2009;3:CD002209.
7. Parrino M. State Methadone Treatment Guidelines. TIPS Series 1.
DHHS Publication No. (SMA) 93-1991. Rockville, MD: Depart- 23.
ment of Health and Human Services; 1993.
8. Marsch LA, Bickel WK, Badger GJ et al. Comparison of pharma- 24.
cological treatments for opioid-dependent adolescents: a random-
ized controlled trial. Arch Gen Psychiatry. 2005;62:1157-1164. 25.
9. Woody GE, Poole SA, Subramaniam G et al. Extended vs short-
term buprenorphine-naloxone for treatment of opioid-addicted
youth: a randomized trial. JAMA. 2008;300:2003-2011. 26.
10. Kraemer HC, Frank E, Kupfer DJ. Moderators of treatment
outcomes: clinical, research, and policy importance. JAMA. 2006; 27.
296:1286-1289. 28.
11. Kravitz RL, Duan N, Braslow J. Evidence-based medicine, heter-
ogeneity of treatment effects, and the trouble with averages.
Milbank Q. 2004;82:661-687. 29.
12. Marsch LA, Stephens MA, Mudric T, Strain EC, Bigelow GE,
Johnson RE. Predictors of outcome in LAAM, buprenorphine, and
methadone treatment for opioid dependence. Exp Clin Psychop-
harmacol. 2005;13:293-302.
13. Gerra G, Borella F, Zaimovic A et al. Buprenorphine versus 30.
methadone for opioid dependence: predictor variables for treat-
ment outcome. Drug Alcohol Depend. 2004;75:37-45.
14. Fiellin DA, Pantalon MV, Chawarski MC et al. Counseling plus 31.
buprenorphine-naloxone maintenance therapy for opioid depen-
dence. N Engl J Med. 2006;355:365-374. 32.
15. Ziedonis DM, Amass L, Steinberg M et al. Predictors of outcome
for short-term medically supervised opioid withdrawal during a
randomized, multicenter trial of buprenorphine-naloxone and
clonidine in the NIDA clinical trials network drug and alcohol 33.
dependence. Drug Alcohol Depend. 2009;99:28-36.
16. Torrington M, Domier CP, Hillhouse M, Ling W. Buprenorphine
101: treating opioid dependence with buprenorphine in an office-
based setting. J Addict Dis. 2007;26:93-99. 34.
17. Motamed M, Marsch LA, Solhkhah R, Bickel WK, Badger GJ.
Differences in treatment outcomes between prescription opioid-
JOURNAL
1128 www.jaacap.org
dependent and heroin-dependent adolescents. J Addict Med.
2008;2:158-164. doi:110.1097/ADM.1090b1013e31816b31812f31884.
Mercer DE, Woody GE. An Individual Drug Counseling Approach to
Treat Cocaine Addiction: The Collaborative Cocaine Treatment Study
Model. NIH Publication No. 99-4380. Rockville, MD: National
Institute of Drug Abuse; 1999.
Chakrabarti A, Woody GE, Griffin ML, Subramaniam G, Weiss
RD. Predictors of buprenorphine-naloxone dosing in a 12-week
treatment trial for opioid-dependent youth: secondary analyses
from a NIDA Clinical Trials Network study. Drug Alcohol
Depend. 2010;107:253-256.
Meade CS, Weiss RD, Fitzmaurice GM et al. HIV risk behavior in
treatment-seeking opioid-dependent youth: results from a NIDA
clinical trials network multisite study. J Acquir Immune Defic
Syndr. 2010;55:65-72.
Miele GM, Carpenter KM, Smith Cockerham M, Dietz Trautman
K, Blaine J, Hasin DS. Concurrent and predictive validity of the
Substance Dependence Severity Scale (SDSS). Drug Alcohol De-
pend. 2000;59:77-88.
Needle R, Fisher DG, Weatherby N et al. Reliability of self-
reported HIV risk behaviors of drug users. Psychol Addict Behav.
1995; 9:242-250.
NIDA. Risk Behavior Survey. 3rd ed. Rockville, MD: National
Institute on Drug Abuse, Community Research Branch; 1993.
Achenbach TM. Manual for the YSR and 1991 Profile. Burlington,
VT: Department of Psychiatry, University of Vermont; 1991.
Achenbach T. Manual for the Young Adult Self-Report and 1997
Profiles. Burlington, VT: Department of Psychiatry, University of
Vermont; 1997.
Gossop M. The development of a Short Opiate Withdrawal Scale
(SOWS). Addict Behav. 1990;15:487-490.
SAS Institute. SAS, 9.2 Edition. Cary, NC: SAS Institute; 2010.
Sullivan LE, Fiellin DA. Buprenorphine: its role in preventing
HIV transmission and improving the care of HIV-infected pa-
tients with opioid dependence. Clin Infect Dis. 2005;41:891-896.
Bachireddy C, Bazazi AR, Kavasery R, Govindasamy S, Kamarul-
zaman A, Altice FL. Attitudes toward opioid substitution therapy
and pre-incarceration HIV transmission behaviors among HIV-
infected prisoners in Malaysia: Implications for secondary pre-
vention. Drug Alcohol Depend. 2011;116:151-157.
Hillhouse M, Domier CP, Chim D, Ling W. Provision of ancillary
medications during buprenorphine detoxification does not im-
prove treatment outcomes. J Addict Dis. 2010;29:23-29.
Stein MD, Cioe P, Friedmann PD. Buprenorphine retention in
primary care. J Gen Intern Med. 2005;20:1038-1041.
Kakko J, Svanborg KD, Kreek MJ, Heilig M. 1-Year retention and
social function after buprenorphine-assisted relapse prevention
treatment for heroin dependence in Sweden: a randomised,
placebo-controlled trial. Lancet. 2003;361:662-668.
Katz EC, Schwartz RP, King S et al. Brief vs. extended buprenor-
phine detoxification in a community treatment program: engage-
ment and short-term outcomes. Am J Drug Alcohol Abuse.
2009;35:63-67.
Villafranca SW, McKellar JD, Trafton JA, Humphreys K. Predic-
tors of retention in methadone programs: a signal detection
analysis. Drug Alcohol Depend. 2006;83:218-224.
OF THE AMERICAN ACADEMY OF CHILD & ADOLESCENT PSYCHIATRY
VOLUME 50 NUMBER 11 NOVEMBER 2011