Safety and Efficacy of Exposure-Based Risk Reduction Through Family Therapy for Co-occurring Substance Use Problems and Posttraumatic Stress Disorder Symptoms Among Adolescents

This randomized clinical trial assesses whether risk reduction through family therapy for adolescents with substance use problems and posttraumatic stress disorder symptoms results in improved outcomes relative to treatment as usual consisting primarily of trauma-focused cognitive behavioral therapy.

Key Points

Question

Is exposure-based risk reduction through family therapy a safe and efficacious treatment compared with treatment as usual for adolescents with co-occurring substance use problems and posttraumatic stress disorder symptoms?

Findings

In this randomized clinical trial that included 124 adolescents aged 13 to 18 years who were followed up 18 months after baseline, greater reductions in substance-using days, marijuana use, and polysubstance use, as well as posttraumatic stress disorder avoidance and hyperarousal symptoms, were observed among participants in the risk reduction through family therapy condition over time.

Meaning

For adolescents who have substance use problems and posttraumatic stress disorder symptoms, an exposure-based treatment for posttraumatic stress disorder symptoms that incorporates substance abuse interventions may yield the best long-term outcomes for these problems.

Abstract

Importance

No empirically supported treatments have been evaluated to address co-occurring substance use problems (SUP) and posttraumatic stress disorder (PTSD) symptoms among adolescents in an integrative fashion. This lack is partially owing to untested clinical lore suggesting that delivery of exposure-based PTSD treatments to youth with SUP might be iatrogenic.

Objective

To determine whether an exposure-based, integrative intervention for adolescents with SUP and PTSD symptoms—risk reduction through family therapy (RRFT)—resulted in improved outcomes relative to a treatment-as-usual (TAU) control condition consisting primarily of trauma-focused cognitive behavioral therapy.

Design, Setting, and Participants

This randomized clinical trial enrolled 124 participants who were recruited from November 1, 2012, through January 30, 2017. Adolescents (aged 13-18 years) who engaged in nontobacco substance use at least once during the past 90 days, experienced at least 1 interpersonal traumatic event, and reported 5 or more PTSD symptoms were enrolled. Blinded assessments were collected at baseline and at 3, 6, 12, and 18 months after baseline. Recruitment and treatment took place in community-based child advocacy centers in the Southeastern United States. Data were analyzed from August 2 through October 4, 2018, and were based on intention to treat.

Interventions

Participants were randomized to receive RRFT (n = 61) or TAU (n = 63).

Main Outcomes and Measures

Primary outcomes focused on number of nontobacco substance–using days measured with the timeline follow-back method and PTSD symptom severity using the UCLA (University of California, Los Angeles) PTSD Reaction Index for DSM-IV completed by adolescents and caregivers. Secondary outcomes focused on marijuana, alcohol, and polysubstance use and PTSD criterion standard (re-experiencing, avoidance, and hyperarousal) symptom severity.

Results

In all, 124 adolescents (mean [SD] age, 15.4 [1.3] years; 108 female [87.1%]) were randomized. For primary outcomes relative to TAU, RRFT yielded significantly greater reductions in substance-using days from baseline to month 12 (event rate [ER], 0.28; 95% CI, 0.12-0.65) and month 18 (ER, 0.10; 95% CI, 0.04-0.24). Significant reductions in PTSD symptoms were observed within groups for RRFT from baseline to months 3 (β = −9.25; 95% CI, −12.95 to −5.55), 6 (β = −16.63; 95% CI = −20.40 to −12.87), 12 (β = −17.51; 95% CI, −21.62 to −13.40), and 18 (β = −19.02; 95% CI, −23.07 to −14.96) and for TAU from baseline to months 3 (β = −9.62; 95% CI, −13.16 to −6.08), 6 (β = −13.73; 95% CI, −17.43 to −10.03), 12 (β = −15.53; 95% CI, −19.52 to −11.55), and 18 (β = −13.88; 95% CI, −17.69 to −10.09); however, between-group differences were not observed.

Conclusions and Relevance

In this study, RRFT and TAU demonstrated within-group improvements in SUP and PTSD symptoms, with greater improvement for substance use and PTSD avoidance and hyperarousal symptoms among adolescents randomized to RRFT compared with TAU. No evidence of the worsening of SUP was observed in either condition. These results suggest that this exposure-based treatment is safe, feasibly delivered by community-based clinicians, and offers an effective approach to inform clinical practice.

Trial Registration

ClinicalTrials.gov Identifier: NCT01751035

Introduction

Interpersonal violence (IPV) is prevalent throughout adolescence.^1,2 By 18 years of age, 1 in 2 youth in the United States experience IPV.^2,3 Interpersonal violence elevates risk for mental health disorders,⁴ particularly posttraumatic stress disorder (PTSD).⁵ Exposure to IPV and PTSD symptoms elevates the risk for serious substance use problems (SUPs) among adolescents.⁶ These risks include early initiation and misuse of various substances, such as marijuana, cocaine, and prescription drugs.⁷

A theoretical model linking SUP and PTSD symptoms is the self-medication hypothesis.⁸ This negative reinforcement model purports that when an individual endures IPV-related distress (PTSD symptoms), motivation shifts to stress reduction strategies. If the individual uses drugs or alcohol as a coping method—and distress temporarily remits—he or she is more likely to use substances again when faced with IPV-related cues. Evidence for this model in adolescents⁹ and adults,⁸ coupled with issues related to client burden (eg, navigating separate SUP and PTSD treatments) and clinician desire for tools that address varied problems representative of real-world clients,¹⁰ underscores the need for an integrative rather than compartmentalized treatment approach for adolescents with SUP–PTSD symptom comorbidity.

Despite progress made in treating adults with co-occurring SUP and PTSD symptoms,^{11,12,13,14,15} no empirically supported, integrative treatments targeting SUP–PTSD symptom comorbidity in adolescents are available at present. The field has several effective treatments for adolescent SUP (eg, contingency management¹⁶ and multisystemic therapy¹⁷) and separate effective treatments for adolescent PTSD (eg, trauma-focused cognitive behavioral therapy [TF-CBT]¹⁸ and developmentally adapted cognitive processing therapy¹⁹). However, no existing interventions address co-occurring SUP and PTSD symptoms among adolescents in an integrated fashion, perhaps owing to the highly compartmentalized nature of the mental health and addiction fields, the well-recognized challenges in the treatment of co-occurring SUP and PTSD symptoms across the life span,¹⁵ and the long-standing but unsupported clinical lore asserting that engaging youth with SUP and PTSD symptoms in exposure-based treatment (criterion standard treatment for PTSD symptoms) is unsafe owing to concerns that exposure-induced distress might exacerbate SUPs.²⁰

Risk reduction through family therapy (RRFT)²¹ was developed to address the need for an efficacious intervention for co-occurring SUP and PTSD symptoms among adolescents. An open trial²² and a pilot randomized clinical trial (RCT)²³ supported the feasibility and preliminary efficacy of RRFT. The present RCT evaluated the safety and effectiveness of RRFT when delivered to a larger sample of adolescents in a community-based treatment setting. We hypothesized that adolescents receiving RRFT would demonstrate significantly greater reductions in SUP (overall use and frequency) and PTSD symptoms relative to adolescents in a treatment-as-usual (TAU) control condition.

Methods

A copy of the trial protocol is found in Supplement 1. The trial was approved by the institutional review board at the Medical University of South Carolina, and an independent Data and Safety Monitoring Board (DSMB) monitored the study. All participants provided written informed consent. This study followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline for RCTs.

Design and Participants

In this rater-blinded, parallel-assignment RCT, participants were adolescents aged 13 to 18 years who experienced IPV (child sexual abuse, physical abuse, physical assault, threat with a weapon, and/or witnessing violence), reported at least 1 nontobacco substance–using day in the past 90 days, and had at least 5 PTSD symptoms as reported on the UCLA (University of California, Los Angeles) PTSD Reaction Index (UCLA-PTSD-RI).^24,25,26 In the first 17 months of the study, sexual abuse was a required form of IPV. However, consistent with the literature demonstrating a well-established link between other forms of IPV and SUP across the life span, many youth at the study sites presented with a wide range of IPV events and substance use problems. Thus, the decision was made to modify the inclusion criteria such that sexual abuse was not a required form of IPV. This recommendation was made by the study’s DSMB after monitoring the screening data and was approved by the study sponsor and the institutional review board overseeing the study. Youth with pervasive developmental disability, intellectual disability, active suicidal or homicidal ideation, or active psychosis or those already engaged in treatment were excluded. An independent DSMB monitored the study. Among the potential serious adverse events reported over the course of the study across the 2 conditions that were reviewed by the DSMB, none were determined to be due to the treatments or study procedures (eMethods in Supplement 2).

Participants were enrolled from November 1, 2012, through January 30, 2017. Participants were identified through referrals to community-based child advocacy centers (CACs) in the Southeastern United States. Child advocacy centers provide services, including mental health treatment, to youth experiencing IPV and their families on referral from community agencies, such as law enforcement, child welfare, schools, and health care facilities. Participants gave written informed consent and verbal assent before enrollment by study staff. Immediately after the baseline assessment, adolescents and their caregivers were urn randomized to the RRFT or TAU conditions by a trained CAC staff member to keep assessors blinded. The urn balanced condition assignment on sex, pretreatment substance use frequency (≥4 substance-using days during the past 30 days), and pretreatment PTSD symptom severity (score on UCLA-PTSD-RI, ≥38 [scores range from 0 to 68, with higher scores indicating greater symptom severity]).

Procedures

Potential participants were screened by study staff using the timeline follow-back method²⁷ to assess nontobacco substance use in the past 90 days and a clinical interview, which included the UCLA-PTSD-RI,^24,25,26 to assess IPV exposure and PTSD symptoms. All cases were vetted by the principal investigator (C.K.D.) to confirm eligibility. Treatment in both conditions was administered by master’s level CAC clinicians who had previous training in trauma-focused treatment and no prior training in SUP treatment. The RRFT clinicians received weekly supervision from the RRFT developer. The TAU clinicians received weekly supervision from an experienced clinician with expertise in TF-CBT¹⁸ and other evidence-based, trauma-focused treatments. The RRFT and TAU clinicians were assigned to 1 treatment condition exclusively.

Interventions

Risk Reduction Through Family Therapy

As detailed elsewhere,²¹ RRFT is an adaptation and integration of evidence-based, cognitive-behavioral interventions that address PTSD symptoms (TF-CBT¹⁸), SUP (multisystemic therapy¹⁷), and sexual health.²⁸ Interventions are organized across 7 components: psychoeducation and engagement, family communication, SUP, coping, PTSD, healthy dating and sexual decision-making, and re-victimization risk reduction.

Risk reduction through family therapy targets individualized risk and protective factors for symptoms/problems across levels of each youth’s ecological features.²⁹ Consistent with negative reinforcement theory,³⁰ RRFT clinicians target self-regulation deficits³¹ to decrease trauma-related SUP. Based on its adaptation from TF-CBT and Mowrer’s 2-factor theory,³² RRFT also includes gradual exposure to address PTSD symptoms via development of detailed written and/or oral narratives of traumatic events to extinguish distress symptoms that have become paired with trauma cues.

The RRFT clinicians had been trained in TF-CBT and completed a 3-day intensive RRFT training. Risk reduction through family therapy was administered through weekly individual youth sessions (mean, 1 hour; range, 40-90 minutes), brief caregiver and family sessions when applicable, and occasional telephone or text check-ins between sessions. Duration of RRFT varies. Treatment progress was tracked with weekly timeline follow-back and PTSD symptom assessments,³³ random urine drug screens, and ongoing updates to the functional assessment of risk and protective factors for PTSD symptoms and SUP.

Treatment as Usual

The TAU clinicians were trained in evidence-based, trauma-focused treatments and were instructed to provide whichever treatment(s) they would typically implement with youth who present to the CAC with PTSD symptoms and SUP. Of note, TAU clinicians completed criterion standard training in TF-CBT (2-day in-person workshop followed by 6 months of consultation calls with national trainers) and received weekly TF-CBT supervision. Trauma-focused CBT¹⁷ is a leading evidence-based intervention for youth affected by trauma. This components-based model consists of weekly, individual child and parent sessions as well as joint child-parent sessions. Each component includes elements of graded exposure to a youth’s trauma cues because fear habituation is viewed as a critical feature of the model. Treatment as usual did not directly target SUP in the participating youth. The TAU clinicians could refer their clients to other community agencies for substance use treatment, if desired.

Treatment fidelity, characterization, and duration were measured using a comprehensive review of participants’ clinical records and observational coding of audio-recorded therapy sessions. All therapy sessions were recorded, and 20% of sessions across both conditions (560 sessions total) were coded by 3 raters (including M.R.M.) trained to greater than 80% interrater reliability using the Family Therapy Scale from the Therapy Procedures Checklist³⁴ and the RRFT Therapist Adherence Measure.³⁵

Outcomes

The primary outcome was number of substance-using days, measured at baseline and 3, 6, 12, and 18 months after baseline using the timeline follow-back method.²⁷ Participants reported on the specific daily amounts of nontobacco drugs and alcohol consumed during the previous 90 days. Timeline follow-back data were used to calculate variables representing the number of days of substance use as well as any marijuana use, number of days of marijuana use, any alcohol use, number of days of alcohol use, and polysubstance use (count of different substance types used) as secondary substance use outcomes.

Given the focus on comorbidity, reductions in overall DSM-IV PTSD symptom severity also was considered a primary outcome, and DSM-IV PTSD symptom criteria count was considered the secondary outcome. Symptoms of PTSD were assessed at all 5 time points using the UCLA-PTSD-RI, Adolescent and Caregiver versions.^24,25,26 The UCLA-PTSD-RI assesses all 17 DSM-IV PTSD symptoms, has been shown to be reliable and valid,^25,26 and has been used extensively in clinical evaluation, traumatic stress research, and posttrauma screening. The UCLA-PTSD-RI was administered as a clinical interview by highly trained study staff. A score of at least 38 on this measure indicates a DSM-IV PTSD diagnosis.

Statistical Analyses

Data were analyzed from August 2 through October 4, 2018. Statistical power was estimated for the difference between RRFT and TAU in change from baseline to each later occasion. With a final sample of 124 participants, the design effect formula was used to calculate the effective number of independent observations provided by each pair of measurements (ρ = 0.50),³⁶ for which power was estimated using G*Power.³⁷ With 165 effectively independent observations, power was 0.80 to detect a small to medium effect of f = 0.13 for the between-group differences in change.

The data collection and analyses followed intention-to-treat methods, and the analyses used all available data. The data had a 2-level structure, with 5 repeated measurements of PTSD symptoms and substance use (level 1) nested within youth (level 2), analyzed using mixed-effects formulations of general linear (continuous; maximum likelihood estimation) and generalized linear (Bernoulli and negative binomial; numerical integration and adaptive quadrature) regression models³⁸ in SuperMix software, version 2.1 (Scientific Software International).³⁹ The model included a contrast for each postbaseline occasion,⁴⁰ an indicator for intervention condition (0 indicates TAU; 1, RRFT), and the interaction between condition and each contrast. This procedure tested for differences between RRFT and TAU at baseline and in changes between baseline and months 3, 6, 12, and 18. This model also tested TAU’s change from baseline to each later occasion, and the same test was obtained for RRFT by using planned contrasts. The reported results include estimates, 95% CIs, and effect size estimates^41,42 for significant between-group differences. For nonlinear outcomes, the magnitude of effects is reflected by odds ratios (ORs) and event rates (ERs). Of note, ORs and ERs less than 1.00 reflect greater reductions for RRFT relative to TAU. Two-sided P < .05 indicated significance.

Results

Participant Characteristics

Figure 1 shows the CONSORT flow diagram. A total of 124 adolescents were enrolled in the RCT and randomized to RRFT (n = 61) or TAU (n = 63) groups (mean [SD] age, 15.4 [1.3] years; 108 female (87.1%] and 16 male [12.9%]). Table 1 provides baseline sample characteristics. At baseline, 92 participants (74.2%) endorsed alcohol use, 83 (66.9%) endorsed marijuana use, and 16 (12.9%) endorsed other nonnicotine drug use (eg, cocaine, pills, stimulants, club drugs, or hallucinogens). Youth in both conditions reported high rates of IPV and other forms of trauma, with a mean (SD) of 3.6 (2.0) different traumatic event types.

Figure 1. Trial Profile.

RRFT indicates risk reduction through family therapy; PTSD, posttraumatic stress disorder; TAU, treatment as usual.

^aFifty-eight (95.1%) RRFT youth and 58 (92.1%) TAU youth received at least partial treatment of the allocated intervention. Of these, 42 (72.4%) RRFT youth and 38 (65.5%) TAU youth completed treatment (ie, attending ≥8 sessions and designated as graduated by the clinician and supervisor).

Table 1. Demographics and Baseline Substance Use and Posttraumatic Stress Disorder Symptoms.

Characteristic	All (N = 124)a	RRFT Group (n = 61)a	TAU Group (n = 63)a	t or χ2 Value	P Value
Youth Characteristics
Age, mean (SD) [range], y	15.4 (1.3) [13-18]	15.5 (1.3)	15.3 (1.2)	t = −0.69	.49
Sex
Female	108 (87.1)	52 (85.2)	56 (88.9)	χ2 = 0.37	.55
Male	16 (12.9)	9 (14.8)	7 (11.1)
Race
White	78 (62.9)	40 (65.6)	38 (60.3)	χ2 = 0.05	.83
Black	36 (29.0)	17 (27.9)	19 (30.2)
Latino	7 (5.6)	3 (4.9)	4 (6.3)
Biracial	3 (2.4)	1 (1.6)	2 (3.2)
Hispanic ethnicity	6 (4.8)	2 (3.3)	4 (6.3)	χ2 = 0.64	.43
Trauma exposure
No. of potentially traumatic events, mean (SD)	3.6 (2.0)	3.4 (1.8)	3.7 (2.2)	t = 0.81	.42
Sexual abuse	85 (68.5)	41 (67.2)	44 (69.8)	χ2 = 0.09	.75
Learned about violent death/serious injury of loved one	64 (51.6)	30 (49.2)	34 (54.0)	χ2 = 0.28	.59
Community violence	49 (39.5)	23 (37.7)	26 (41.3)	χ2 = 0.17	.69
Domestic violence	45 (36.3)	21 (34.4)	24 (38.1)	χ2 = 0.18	.67
Physical abuse	42 (33.9)	16 (26.2)	26 (41.3)	χ2 = 3.10	.08
Physical assault	35 (28.2)	16 (26.2)	19 (30.2)	χ2 = 0.24	.63
Natural disaster	26 (21.0)	11 (18.0)	15 (23.8)	χ2 = 0.62	.43
Serious accident	18 (14.5)	7 (11.5)	11 (17.5)	χ2 = 0.90	.34
Medical trauma	17 (13.7)	6 (9.8)	11 (17.5)	χ2 = 1.50	.22
Saw a dead body	14 (11.3)	6 (9.8)	8 (12.7)	χ2 = 0.25	.62
Exposure to war/political violence	3 (2.4)	3 (4.9)	0	χ2 = 3.10	.08
Other	39 (31.5)	24 (39.3)	15 (23.8)	χ2 = 2.90	.09
Psychiatric comorbidities,b No. (%)
Generalized anxiety disorder	12 (11.4)	6 (11.8)	6 (11.1)	χ2 = 0.01	.92
Panic disorder	20 (20.2)	8 (17.4)	12 (22.6)	χ2 = 0.42	.52
Major depressive episode	17 (16.2)	8 (15.7)	9 (16.7)	χ2 = 0.02	.89
Dysthymic disorder	13 (14.3)	5 (11.9)	8 (16.3)	χ2 = 0.36	.55
Manic episode	1 (1.0)	1 (2.0)	0	χ2 = 1.07	.30
Hypomanic episode	5 (4.8)	2 (3.9)	3 (5.7)	χ2 = 0.17	.68
Oppositional defiant disorder	27 (25.7)	14 (27.5)	13 (24.1)	χ2 = 0.16	.69
Conduct disorder	27 (26.5)	14 (27.5)	13 (25.5)	NA	NA
Using psychotropic medications	44 (35.5)	19 (31.1)	25 (39.7)	χ2 = 0.99	.32
Caregiver Characteristics
Sex
Female	107 (89.9)	57 (95.0)	50 (84.7)	χ2 = 3.50	.06
Male	12 (10.1)	3 (5.0)	9 (15.3)
Race
White	79 (64.8)	40 (67.8)	39 (61.9)	χ2 = 2.50	.48
Black	41 (33.6)	18 (30.5)	23 (36.5)
Biracial	1 (0.8)	1 (1.7)	0
Other (not specified)	1 (0.8)	0	1 (1.6)
Hispanic ethnicity	4 (3.3)	2 (3.4)	4 (6.3)	χ2 = 3.90	.05
Relationship to youth
Mother	73 (58.9)	39 (63.9)	34 (54.0)	χ2 = 9.70	.14
Father	13 (10.5)	3 (4.9)	10 (15.9)
Grandmother	10 (8.1)	5 (8.2)	5 (7.9)
Aunt	2 (1.6)	2 (3.3)	0
Foster mother	9 (7.3)	6 (9.8)	3 (4.8)
Stepmother	1 (0.8)	0	1 (1.6)
Case worker	15 (12.1)	5 (8.2)	10 (15.9)
Has secondary caregiver	63 (52.5)	31 (53.4)	32 (51.6)	χ2 = 0.14	.84
Marital statusc
Married/cohabiting with other parent	31 (28.9)	19 (34.5)	12 (23.1)	χ2 = 3.39	.64
Married/cohabiting with nonparent	28 (26.1)	11 (20.0)	17 (32.7)
Never married	13 (12.1)	6 (10.9)	7 (13.5)
Separated	16 (15.0)	8 (14.5)	8 (15.4)
Divorced	17 (15.9)	10 (18.2)	7 (13.5)
Widowed	2 (1.9)	1 (1.8)	1 (1.9)
Employmentc,d
Full-time	54 (51.4)	25 (46.3)	29 (56.9)	χ2 = 4.91	.56
Part-time	15 (14.3)	6 (11.1)	9 (17.6)
Student	3 (2.9)	2 (3.7)	1 (2.0)
Homemaker	12 (11.4)	9 (16.7)	3 (5.9)
Unemployed	15 (14.3)	8 (14.8)	7 (13.7)
Retired	6 (5.7)	4 (7.4)	2 (3.9)
Educational attainmentc
Less than high school	19 (18.4)	14 (26.4)	5 (10.0)	χ2 = 4.91	.56
High school diploma/GED	14 (13.6)	6 (11.3)	8 (16.0)
Some college	38 (36.9)	18 (34.0)	20 (40.0)
College degree	22 (21.4)	10 (18.9)	12 (24.0)
Graduate degree	10 (9.7)	5 (9.4)	5 (10.0)
Substance use
Positive urine drug screen	40 (34.2)	23 (45.9)	17 (54.0)	χ2 = −1.23	.22
TLFB, mean (SD)
No. of SU days of 90	14.5 (18.3)	15.5 (17.0)	13.6 (19.6)	t = −0.78	.57
No. of alcohol use days	3.7 (7.5)	4.8 (9.3)	2.6 (4.9)	t = −1.67	.10
No. of marijuana use days	10.4 (17.8)	11.2 (16.5)	9.7 (19.0)	t = −0.45	.66
Polysubstance usee	62 (50.0)	28 (45.9)	34 (54.0)	χ2 = 0.81	.37
PTSD Symptomsf
Youth report
Met PTSD diagnostic threshold, No. (%)	77 (65.8)	35 (61.4)	42 (70.0)	χ2 = 0.96	.33
Severity score, mean (SD)
Total PTSD	36.6 (14.0)	36.5 (14.6)	36.8 (13.6)	t = 0.09	.93
Criterion B (re-experiencing)	13.4 (6.6)	13.0 (6.5)	13.9 (6.7)	t = 0.73	.47
Criterion C (avoidance)	14.0 (5.8)	14.3 (5.8)	13.7 (5.8)	t = −0.57	.57
Criterion D (hyperarousal)	11.9 (4.8)	12.0 (4.9)	11.7 (4.6)	t = −0.37	.71
Caregiver report
Met PTSD diagnostic threshold, No. (%)	79 (76.7)	38 (77.6)	41 (75.9)	χ2 = 0.04	.85
Severity score, mean (SD)
Total PTSD	45.1 (17.2)	48.4 (16.3)	42.2 (17.6)	t = −1.87	.45
Criterion B (re-experiencing)	20.2 (7.9)	21.3 (7.6)	19.1 (8.0)	t = −1.49	.14
Criterion C (avoidance)	16.7 (8.2)	18.9 (7.7)	14.8 (8.3)	t = −2.83	.006
Criterion D (hyperarousal)	12.1 (4.9)	12.7 (4.8)	11.5 (5.0)	t = −1.43	.15

Abbreviations: GED, General Education Development; NA, not applicable; PTSD, posttraumatic stress disorder; RRFT, risk reduction through family therapy; SU, substance use; TAU, treatment as usual; TLFB, timeline follow-back.

^aUnless otherwise specified, data are expressed as number (percentage) of patients. Percentages represent percentage of respondents to each item; sample size for each question varies somewhat.

^bBased on responses to the Clinical Data Interchange Standards Consortium.

^cIncludes non–case worker caregivers only.

^dIn reference to primary caregiver only.

^eDefined as endorsement of more than 1 nonnicotine substance at baseline per TLFB method.

^fScores derived from the UCLA (University of California, Los Angeles) PTSD Reaction Index for DSM-IV.^23,24,25

Retention and Treatment Delivery

Research retention was high, with 50 (82.0%) of the RRFT group and 56 (88.9%) of the TAU group completing at least 1 postbaseline research assessment. Mean (SD) treatment length was 18.53 (12.08) sessions for RRFT and 12.51 (8.49) for TAU; 58 (95.1%) of the RRFT group and 58 (92.1%) of the TAU group received at least part of the allocated intervention. Of these, 42 (72.4%) of the RRFT group and 38 (65.5%) of the TAU group completed treatment (ie, attended ≥8 sessions and designated as graduated by the clinician and supervisor). Referrals for outside services occurred in 16 cases (25.4%) in the TAU group, including SUP treatment (9 cases [14.3%]), hospitalization (13 [20.6%]) and other mental health services (14 [22.2%]). Therapist Procedures Checklist results indicated that, of the 19 RRFT families and 24 TAU families included in the tape coding protocol, the percentage receiving family therapy techniques as part of their study treatment sessions was higher in the RRFT group (10 families [52.6%]) relative to TAU (9 families [37.5%]). The RRFT Therapist Adherence Measure was used to measure treatment adherence, with Rasch measurement models indicating acceptable psychometric performance (97% interrater agreement, session-level reliability of 0.59-0.62, and family-level reliability of 0.85). Adherence scores revealed consistency in the delivery of TF-CBT elements across both conditions. Of the 19 RRFT families and 24 TAU families included in the tape coding protocol, the percentage receiving features of multisystemic therapy for SUP (eg, development of alternative skills for managing SUP triggers) was higher among individuals receiving RRFT (17 families [89.5%]) relative to TAU (12 families [50.0%]). These data support the intended intervention condition similarities and differences.

SUP–PTSD Symptom Outcomes

Substance Use

Substance use outcome results are reported in Table 2 and Figure 2A. For the number of days with substance use, RRFT and TAU did not differ significantly at baseline. From baseline to months 3 (β = −1.09; 95% CI, −1.58 to −0.60), 6 (β = −1.03; 95% CI, −1.55 to −0.50), and 12 (β = −1.02; 95% CI, −1.61 to −0.42), TAU had significant reductions in substance-using days. For RRFT, the reductions were significant from baseline to months 3 (β = −1.38; 95% CI, −1.88 to −0.88), 6 (β = −1.54; 95% CI, −2.11 to −0.97), 12 (β = −2.28; 95% CI, −2.87 to −1.69), and 18 (β = −2.33; 95% CI, −2.96 to −1.70). Relative to TAU, RRFT had significantly greater decreases from baseline to months 12 (−10.5 vs −4.8 days; ER, 0.28; 95% CI, 0.12-0.65) and 18 (−10.6 vs −0.4 days; ER, 0.10; 95% CI, 0.04-0.24). The ERs indicate that the decrease in the number of using days from baseline to month 12 for RRFT was 72% greater (ER, 0.28) than the decrease for TAU; from baseline to month 18, the decrease for RRFT was 90% greater (ER, 0.10).

Table 2. Mixed-Effects Regression Model Results for Youth Reports of Substance Use^a.

	Any Useb		No. of Days With Usec
	β (SE) [95% CI]	P Value	β (SE) [95% CI]	P Value
Any Substanced
Intercept	NA	NA	2.03 (0.23) [1.57 to 2.48]	<.001
RRFT	NA	NA	0.44 (0.33) [−0.21 to 1.08]	.19
Month 3e	0.74 (0.46) [−0.16 to 1.64]	.11	−1.09 (0.25) [−1.58 to −0.60]	<.001
RRFT × month 3	0.10 (0.66 [−1.19 to 1.40]	.88	−0.29 (0.36) [−0.99 to 0.41]	.41
Month 6	0.12 (0.55) [−0.95 to 1.19]	.83	−1.03 (0.27) [−1.55 to −0.50]	<.001
RRFT × month 6	−0.99 (0.78) [−2.52 to 0.54]	.20	−0.51 (0.39) [−1.28 to 0.26]	.20
Month 12	−0.73 (0.59) [−1.88 to 0.42]	.21	−1.02 (0.30) [−1.61 to −0.42]	.001
RRFT × month 12	−0.66 (0.83) [−2.28 to 0.96]	.43	−1.27 (0.43) [−2.10 to −0.43]	.003
Month 18	0.05 (0.57) [−1.07 to 1.16]	.94	−0.05 (0.28) [−0.61 to 0.50]	.85
RRFT × month 18	−1.54 (0.84) [−3.18 to 0.10]	.07	−2.28 (0.43) [−3.11 to −1.44]	<.001
RRFT estimatef
Month 3	NA	NA	−1.38 (0.26) [−1.88 to −0.88]	<.001
Month 6	−0.87 (0.56) [−1.97 to 0.22]	.12	−1.54 (0.29) [−2.11 to −0.97]	<.001
Month 12	−1.39 (0.60) [−2.56 to −0.21]	.02	−2.28 (0.30) [−2.87 to −1.69]	<.001
Month 18	−1.50 (0.61) [−2.70 to −0.29]	.02	−2.33 (0.32) [−2.96 to −1.70]	<.001
Marijuana Useg
Intercept	0.59 (0.41) [−0.22 to 1.39]	.16	0.82 (0.34) [0.16 to 1.49]	.02
RRFT	1.19 (0.62) [−0.02 to 2.40]	.05	0.76 (0.46) [−0.14 to 1.66]	.10
Month 3	−0.67 (0.49) [−1.64 to 0.29]	.17	−0.78 (0.29) [−1.36 to −0.20]	.01
RRFT × month 3	−0.85 (0.75) [−2.31 to 0.61]	.25	−0.65 (0.42) [−1.48 to 0.17]	.12
Month 6	−0.85 (0.52) [−1.87 to 0.18]	.11	−0.67 (0.31) [−1.28 to −0.06]	.03
RRFT × month 6	−1.71 (0.80) [−3.27 to −0.15]	.03	−0.87 (0.47) [−1.80 to 0.06]	.07
Month 12	−1.24 (0.58) [−2.37 to −0.11]	.03	−0.35 (0.36) [−1.05 to 0.35]	.32
RRFT × month 12	−2.35 (0.89) [−4.08 to −0.61]	.01	−2.11 (0.52) [−3.12 to −1.10]	<.001
Month 18	−0.05 (0.54) [−1.11 to 1.00]	.92	0.65 (0.33) [0 to 1.29]	.05
RRFT × month 18	−3.19 (0.87) [−4.90 to −1.48]	<.001	−3.10 (0.51) [−4.09 to −2.11]	<.001
RRFT estimatef
Month 3	−1.52 (0.56) [−2.63 to −0.42]	.01	−1.44 (0.30) [−2.02 to −0.85]	<.001
Month 6	−2.56 (0.60) [−3.74 to −1.37]	<.001	−1.54 (0.36) [−2.24 to −0.85]	<.001
Month 12	−3.59 (0.69) [−4.94 to −2.24]	<.001	−2.46 (0.37) [−3.19 to −1.73]	<.001
Month 18	−3.25 (0.68) [−4.58 to −1.91]	<.001	−2.46 (0.38) [−3.21 to −1.71]	<.001
Alcohol Useg
Intercept	1.94 (0.44) [1.07 to 2.81]	<.001	0.41 (0.24) [−0.05 to 0.88]	.08
RRFT	−0.69 (0.58) [−1.83 to 0.45]	.24	0.24 (0.35) [−0.44 to 0.91]	.50
Month 3	−2.69 (0.56) [−3.77 to −1.60]	<.001	−1.16 (0.31) [−1.77 to −0.56]	<.001
RRFT × month 3	0.87 (0.73) [−0.56 to 2.31]	.23	0.18 (0.43) [−0.66 to 1.02]	.68
Month 6	−2.41 (0.57) [−3.52 to −1.30]	<.001	−1.23 (0.33) [−1.87 to −0.59]	<.001
RRFT × month 6	−0.12 (0.77) [−1.62 to 1.38]	.87	0.21 (0.47) [−0.70 to 1.12]	.65
Month 12	−2.65 (0.61) [−3.85 to −1.45]	<.001	−1.00 (0.35) [−1.68 to −0.32]	.004
RRFT × month 12	0.41 (0.80) [−1.16 to 1.99]	.61	−0.69 (0.50) [−1.66 to 0.28]	.16
Month 18	−3.15 (0.63) [−4.39 to −1.91]	<.001	−0.90 (0.36) [−1.61 to −0.20]	.01
RRFT × month 18	0.43 (0.83) [−1.20 to 2.05]	.61	−0.58 (0.51) [−1.59 to 0.42]	.25
RRFT, estimatef
Month 3	−1.81 (0.53) [−2.84 to −0.78]	.001	−0.99 (0.28) [−1.53 to −0.44]	.001
Month 6	−2.53 (0.56) [−3.63 to −1.43]	<.001	−1.11 (0.34) [−1.76 to −0.45]	.003
Month 12	−2.24 (0.57) [−3.36 to −1.12]	<.001	−1.69 (0.35) [−2.38 to −1.00]	<.001
Month 18	−2.73 (0.61) [−3.92 to −1.53]	<.001	−1.49 (0.37) [−2.21 to −0.77]	<.001
Polysubstance Useg,h
Threshold 1	−2.73 (0.36) [−3.44 to −2.02]	<.001	NA	NA
Threshold 2	1.11 (0.32) [0.49 to 1.74]	<.001	NA	NA
RRFT	0.42 (0.44) [−0.45 to 1.28]	.93	NA	NA
Month 3	−1.92 (0.42) [−2.75 to −1.09]	<.001	NA	NA
RRFT × month 3	0.24 (0.59) [−0.92 to 1.39]	.69	NA	NA
Month 6	−1.92 (0.44) [−2.78 to −1.05]	<.001	NA	NA
RRFT × month 6	−0.62 (0.61) [−1.82 to 0.58]	.31	NA	NA
Month 12	−2.49 (0.49) [−3.45 to −1.53]	<.001	NA	NA
RRFT × month 12	−0.51 (0.67) [−1.81 to 0.79]	.44	NA	NA
Month 18	−1.69 (0.46) [−2.59 to −0.79]	<.001	NA	NA
RRFT × month 18	−1.37 (0.67) [−2.68 to −0.07]	.04	NA	NA
RRFT, estimateg
Month 3	−1.68 (0.44) [−2.54 to −0.82]	<.001	NA	NA
Month 6	−2.54 (0.46) [−3.44 to −1.63]	<.001	NA	NA
Month 12	−3.00 (0.51) [−3.99 to −2.01]	<.001	NA	NA
Month 18	−3.06 (0.52) [−4.08 to −2.04]	<.001	NA	NA

Abbreviations: NA, not applicable; RRFT, risk reduction through family therapy.

^aThe models included a dummy-coded indicator for each postbaseline measurement occasion: months 3, 6, 12, and 18. These terms reflect the change for treatment as usual between baseline and each later occasion. The contrasts reflect the same information for the RRFT group, and the RRFT × month terms tested for between-group differences in change. Test statistics are not reported but were computed as β coefficient (SE) unless reported as an estimate (SE), and the odds ratios (any use) and event rates (number of days with use) reported in the text were computed as exp (β).

^bIndicates dichotomous outcome (0 indicates no use; 1, use) with binomial sampling distribution (logit link).

^cIndicates count-distributed outcome with negative binomial sampling distribution (log link).

^dIndicates primary outcome.

^eBaseline substance use was required for participation, and as a result, month 3 status became the model intercept.

^fIndicates planned contrast for RRFT’s change between baseline and each later occasion.

^gIndicates secondary outcome.

^hIndicates ordered categorical outcome with ordinal sampling distribution (logit link).

Figure 2. Estimated Outcomes.

A, Estimated days with any substance use for risk reduction through family therapy (RRFT) vs treatment as usual (TAU). B and C, Estimated posttraumatic stress disorder (PTSD) total symptom severity from adolescent and caregiver reports for RRFT vs TAU. The statistical tests focused on change between baseline and each later occasion, both within and between groups.

For the any marijuana use outcome, RRFT and TAU did not differ significantly at baseline. For TAU, the change from baseline to month 12 was significant (β = −1.24; 95% CI, −2.37 to −0.11); for RRFT, the reductions from baseline were significant to months 3 (β = −1.52; 95% CI, −2.63 to −0.42), 6 (β = −2.56; 95% CI, −3.74 to −1.37), 12 (β = −3.59; 95% CI, −4.94 to −2.24), and 18 (β = −3.25; 95% CI, −4.58 to −1.91). Relative to TAU, RRFT had significantly greater reductions in any marijuana use from baseline to month 6 (−54% vs −21%; OR, 0.18; 95% CI, 0.04-0.86), month 12 (−71% vs −34%; OR, 0.10; 95% CI, 0.02-0.54), and month 18 (−67% vs −1%; OR, 0.04; 95% CI, 0.01-0.23). The ORs indicate that the decrease in odds of any marijuana use from baseline to month 6 for RRFT was 82% greater (OR, 0.18) than the decrease for TAU; from baseline to months 12 and 18, the decrease for RRFT was 90% (OR, 0.10) and 96% (OR, 0.04) greater, respectively, than for TAU.

For the number of days with marijuana use, RRFT and TAU did not differ significantly at baseline. From baseline to months 3 (β = −0.78; 95% CI, −1.36 to −0.20) and 6 (β = −0.67; 95% CI, −1.28 to −0.06), TAU had significant reductions in marijuana-using days; for RRFT, the reductions were significant from baseline to months 3 (β = −1.44; 95% CI, −2.02 to −0.85), 6 (β = −1.54; 95% CI, −2.24 to −0.85), 12 (β = −2.46; 95% CI, −3.19 to −1.73), and 18 (β = −2.46; 95% CI, −3.21 to −1.71). Relative to TAU, the reductions for RRFT were significantly greater between baseline and month 12 (−4.4 vs −0.7 days; ER, 0.12; 95% CI, 0.04-0.33) and month 18 (−4.4 vs 2.1 days; ER, 0.05; 95% CI, 0.02-0.12).

For the any alcohol use outcome, RRFT and TAU did not differ significantly at baseline. Significant reductions in the log-odds of alcohol use were observed from baseline for TAU at months 3 (β = −2.69; 95% CI, −3.77 to −1.60), 6 (β = −2.41; 95% CI, −3.52 to −1.30), 12 (β = −2.65; 95% CI, −3.85 to −1.45), and 18 (β = −3.15; 95% CI, −4.39 to −1.91) and for RRFT at months 3 (β = −1.81; 95% CI, −2.84 to −0.78), 6 (β = −2.53; 95% CI, −3.63 to −1.43), 12 (β = −2.24; 95% CI, −3.36 to −1.12), and 18 (β = −2.73; 95% CI, −3.92 to −1.53), but the reductions did not differ significantly for the 2 groups.

For the number of days with alcohol use, RRFT and TAU did not differ significantly at baseline. Significant reductions in alcohol-using days were observed from baseline for TAU at 3 (β = −1.16; 95% CI, −1.77 to −0.56), 6 (β = −1.23; 95% CI, −1.87 to −0.59), 12 (β = −1.00; 95% CI, −1.68 to −0.32), and 18 (β = −0.90; 95% CI, −1.61 to −0.20) and for RRFT at months 3 (β = −0.99; 95% CI, −1.53 to −0.44), 6 (β = −1.11; 95% CI, −1.76 to −0.45), 12 (β = −1.69; 95% CI, −2.38 to −1.00), and 18 (β = −1.49; 95% CI, −2.21 to −0.77), but the reductions did not differ significantly for the 2 groups.

For the ordinal polysubstance use outcome, RRFT and TAU did not differ at baseline. Significant reductions were observed from baseline for TAU at months 3 (β = −1.92; 95% CI, −2.75 to −1.09), 6 (β = −1.92; 95% CI, −2.78 to −1.05), 12 (β = −2.49; 95% CI, −3.45 to −1.53), and 18 (β = −1.69; 95% CI, −2.59 to −0.79) and for RRFT at months 3 (β = −1.68; 95% CI, −2.54 to −0.82), 6 (β = −2.54; 95% CI, −3.44 to −1.63), 12 (β = −3.00; 95% CI, −3.99 to −2.01), and 18 (β = −3.06; 95% CI, −4.08 to −2.04). Relative to TAU, RRFT had significantly greater reductions from baseline to month 18 (≥1 vs 0 drugs: −51% vs −20%; OR, 0.25; 95% CI, 0.07-0.94).

PTSD Symptoms

Outcome results for PTSD symptoms are reported in Table 3 and Figure 2B and C. For total PTSD symptom severity, RRFT had significantly higher scores than TAU at baseline for caregiver reports (49.10 vs 42.28; β = 6.82; 95% CI, 0.61-13.03; d = 0.39) but not youth reports. From baseline to later occasions, the same pattern held for TAU and RRFT. Caregivers reported significant reductions at months 6 (β = −10.05 [95% CI, −15.30 to −4.81] for TAU; β = −15.69 [95% CI, −21.20 to −10.17] for RRFT), 12 (β = −9.13 [95% CI, −14.76 to −3.49] for TAU; β = −14.26 [95% CI, −20.15 to −8.37] for RRFT), and 18 (β = −12.02 [95% CI, −17.76 to −6.28] for TAU; β = −15.53 [95% CI, −21.88 to −9.18] for RRFT), as did youth at months 3 (β = −9.62 [95% CI, −13.16 to −6.08] for TAU; β = −9.25 [95% CI, −12.95 to −5.55] for RRFT), 6 (β = −13.73 [95% CI, −17.43 to −10.03] for TAU; β = −16.63 [95% CI, −20.40 to −12.87] for RRFT), 12 (β = −15.53 [95% CI, −19.52 to −11.55] for TAU; β = −17.51 [95% CI, −21.62 to −13.40] for RRFT), and 18 (β = −13.89 [95% CI, −17.69 to −10.09] for TAU; β = −19.02 [95% CI, −23.07 to −14.96] for RRFT). The 2 groups did not differ significantly on these changes.

Table 3. Mixed-Effects Regression Model Results for PTSD Symptom Outcomes^a.

Result	Youth Report		Caregiver Report
	β (SE) [95% CI]	P Value	β (SE) [95% CI]	P Value
PTSD Symptom Totalb
Intercept	36.26 (1.75) [32.82 to 39.69]	<.001	42.28 (2.23) [37.91 to 46.64]	<.001
RRFT	0.44 (2.50) [−4.46 to 5.34]	.86	6.82 (3.17) [0.61 to 13.03]	.03
Month 3	−9.62 (1.81) [−13.16 to −6.08]	<.001	−2.29 (2.64) [−7.46 to 2.88]	.39
RRFT × month 3	0.38 (2.61) [−4.75 to 5.50]	.89	−2.62 (3.89) [−10.23 to 4.99]	.50
Month 6	−13.73 (1.89) [−17.43 to −10.03]	<.001	−10.05 (2.68) [−15.30 to −4.81]	<.001
RRFT × month 6	−2.91 (2.70) [−8.19 to 2.38]	.28	−5.63 (3.88) [−13.24 to 1.98]	.15
Month 12	−15.53 (2.03) [−19.52 to −11.55]	<.001	−9.13 (2.88) [−14.76 to −3.49]	.002
RRFT × month 12	−1.98 (2.92) [−7.70 to 3.75]	.50	−5.13 (4.16) [−13.28 to 3.02]	.22
Month 18	−13.89 (1.94) [−17.69 to −10.09]	<.001	−12.02 (2.93) [−17.76 to −6.28]	<.001
RRFT × month 18	−5.12 (2.84) [−10.68 to 0.44]	.07	−3.51 (4.36) [−12.07 to 5.05]	.42
RRFT estimatec
Month 3	−9.25 (1.89) [−12.95 to −5.55]	<.001	−4.91 (2.85) [−10.50 to 0.68]	.09
Month 6	−16.63 (1.92) [−20.40 to −12.87]	<.001	−15.69 (2.82) [−21.20 to −10.17]	<.001
Month 12	−17.51 (2.10) [−21.62 to −13.40]	<.001	−14.26 (3.01) [−20.15 to −8.37]	<.001
Month 18	−19.02 (2.07) [−23.07 to −14.96]	<.001	−15.53 (3.24) [−21.88 to −9.18]	<.001
Criterion B (Re-experiencing)d
Intercept	13.73 (0.79) [12.18 to 15.28]	<.001	18.98 (1.09) [16.85 to 21.11]	<.001
RRFT	−0.83 (1.13) [−3.05 to 1.38]	.46	2.20 (1.55) [−0.83 to 5.23]	.16
Month 3	−4.95 (0.85) [−6.60 to −3.29]	<.001	−0.91 (1.36) [−3.58 to 1.76]	.50
RRFT × month 3	1.36 (1.22) [−1.04 to 3.75]	.27	0.72 (2.01) [−3.22 to 4.66]	.72
Month 6	−6.85 (0.88) [−8.57 to −5.13]	<.001	−5.55 (1.38) [−8.26 to −2.84]	<.001
RRFT × month 6	0.75 (1.26) [−1.72 to 3.21]	.55	−1.47 (2.00) [−5.39 to 2.45]	.46
Month 12	−7.23 (0.95) [−9.09 to −5.37]	<.001	−7.25 (1.48) [−10.15 to −4.35]	<.001
RRFT × month 12	0.59 (1.37) [−2.09 to 3.27]	.67	2.53 (2.14) [−1.65 to 6.72]	.24
Month 18	−7.28 (0.90) [−9.04 to −5.52]	<.001	−6.10 (1.52) [−9.07 to −3.12]	<.001
RRFT × month 18	0.40 (1.32) [−2.19 to 2.99]	.76	0.16 (2.26) [−4.27 to 4.59]	.95
RRFT estimatec
Month 3	−3.59 (0.88) [−5.32 to −1.86]	<.001	−0.20 (1.48) [−3.10 to 2.71]	.90
Month 6	−6.11 (0.90) [−7.87 to −4.34]	<.001	−7.02 (1.45) [−9.86 to −4.19]	<.001
Month 12	−6.64 (0.98) [−8.56 to −4.72]	<.001	−4.72 (1.54) [−7.74 to −1.70]	.002
Month 18	−6.88 (0.97) [−8.78 to −4.98]	<.001	−5.94 (1.67) [−9.22 to −2.66]	<.001
Criterion C (Avoidance)d
Intercept	13.69 (0.72) [12.28 to 15.11]	<.001	14.81 (1.02) [12.82 to 16.80]	<.001
RRFT	0.79 (1.03) [−1.23 to 2.82]	.44	4.14 (1.47) [1.27 to 7.01]	.01
Month 3	−4.02 (0.81) [−5.61 to −2.43]	<.001	0.05 (1.18) [−2.26 to 2.35]	.97
RRFT × month 3	0.32 (1.17) [−1.98 to 2.62]	.78	−3.27 (1.75) [−6.71 to 0.16]	.06
Month 6	−5.19 (0.84) [−6.84 to −3.54]	<.001	−2.80 (1.20) [−5.16 to −0.45]	.02
RRFT × month 6	−1.91 (1.20) [−4.27 to 0.45]	.11	−3.65 (1.76) [−7.10 to −0.20]	.04
Month 12	−6.38 (0.91) [−8.16 to −4.59]	<.001	−1.88 (1.29) [−4.41 to 0.65]	.15
RRFT × month 12	−0.88 (1.31) [−3.45 to 1.68]	.50	−4.94 (1.88) [−8.63 to −1.26]	.01
Month 18	−5.42 (0.86) [−7.11 to −3.73]	<.001	−3.82 (1.33) [−6.42 to −1.22]	.004
RRFT × month 18	−2.73 (1.27) [−5.21 to −0.25]	.03	−3.59 (1.98) [−7.47 to 0.29]	.07
RRFT estimatec
Month 3	−3.69 (0.85) [−5.35 to −2.04]	<.001	−3.22 (1.30) [−5.77 to −0.68]	.01
Month 6	−7.10 (0.86) [−8.79 to −5.41]	<.001	−6.45 (1.28) [−8.97 to −3.94]	<.001
Month 12	−7.26 (0.94) [−9.10 to −5.42]	<.001	−6.83 (1.37) [−9.50 to −4.15]	<.001
Month 18	−8.15 (0.93) [−9.97 to −6.33]	<.001	−7.41 (1.47) [−10.29 to −4.53]	<.001
Criterion D (Hyperarousal)d
Intercept	11.69 (0.59) [10.54 to 12.84]	<.001	11.42 (0.61) [10.24 to 12.61]	<.001
RRFT	0.41 (0.84) [−1.23 to 2.05]	.63	1.30 (0.87) [−0.41 to 3.00]	.14
Month 3	−1.76 (0.62) [−2.98 to −0.54]	.01	−1.03 (0.73) [−2.45 to 0.40]	.16
RRFT × month 3	−1.80 (0.90) [−2.94 to 0.58]	.19	−0.64 (1.07) [−2.75 to 1.46]	.55
Month 6	−3.23 (0.64) [−4.50 to −1.97]	<.001	−2.20 (0.74) [−3.64 to −0.75]	.003
RRFT × month 6	−1.71 (0.92) [−3.51 to 0.1]	.06	−1.23 (1.07) [−3.34 to 0.87]	.25
Month 12	−3.64 (0.70) [−5.00 to −2.27]	<.001	−1.30 (0.80) [−2.86 to 0.27]	.10
RRFT × month 12	−1.57 (1.00) [−3.54 to 0.39]	.12	−2.58 (1.16) [−4.85 to −0.31]	.03
Month 18	−3.02 (0.66) [−4.31 to −1.73]	<.001	−3.07 (0.81) [−4.66 to −1.47]	<.001
RRFT × month 18	−2.36 (0.97) [−4.26 to −0.46]	.02	−0.78 (1.21) [−3.15 to 1.59]	.52
RRFT estimatec
Month 3	−2.94 (0.65) [−4.21 to −1.67]	<.001	−1.67 (0.79) [−3.22 to −0.12]	.04
Month 6	−4.94 (0.66) [−6.23 to −3.64]	<.001	−3.43 (0.78) [−4.96 to −1.90]	<.001
Month 12	−5.21 (0.72) [−6.62 to −3.80]	<.001	−3.87 (0.84) [−5.52 to −2.22]	<.001
Month 18	−5.38 (0.71) [−6.77 to −3.99]	<.001	−3.84 (0.89) [−5.60 to −2.09]	<.001

Abbreviations: PTSD, posttraumatic stress disorder; RRFT, risk reduction through family therapy.

^aThe models included a dummy-coded indicator for each postbaseline measurement occasion: months 3, 6, 12, and 18. These terms reflect the change for treatment as usual between baseline and each later occasion. The contrasts reflect the same information for the RRFT group, and the RRFT × month terms tested for between-group differences in change. Test statistics are not reported but were computed as β coefficient (SE) unless otherwise expressed as estimate (SE).

^bIndicates primary outcome.

^cMeasured planned contrast for RRFT’s change between baseline and each later occasion.

^dIndicates secondary outcome.

For criterion B (re-experiencing) symptom severity, RRFT and TAU did not differ significantly at baseline. From baseline, both groups had significant reductions for caregiver reports at months 6 (β = −5.55 [95% CI, −8.26 to −2.84] for TAU; β = −7.02 [95% CI, −9.86 to −4.19] for RRFT), 12 (β = −7.25 [95% CI, −10.15 to −4.35] for TAU; β = −4.72 [95% CI, −7.74 to −1.70] for RRFT), and 18 (β = −6.10 [95% CI, −9.07 to −3.12] for TAU; β = −5.94 [95% CI, −9.22 to −2.66] for RRFT) and for youth reports at months 3 (β = −4.95 [95% CI, −6.60 to −3.29] for TAU; β = −3.59 [95% CI, −5.32 to −1.86] for RRFT), 6 (β = −6.85 [95% CI, −8.57 to −5.13] for TAU; β = −6.11 [95% CI, −7.87 to −4.34] for RRFT), 12 (β = −7.23 [95% CI, −9.09 to −5.37] for TAU; β = −6.64 [95% CI, −8.56 to −4.72] for RRFT), and 18 (β = −7.28 [95% CI, −9.04 to −5.52] for TAU; β = −6.88 [95% CI, −8.78 to −4.98] for RRFT). The 2 groups did not differ significantly on these changes.

For criterion C (avoidance) symptom severity, RRFT had significantly higher scores at baseline for caregiver reports (18.95 vs 14.81; β = 4.14; 95% CI, 1.27-7.01; d = 0.50) but not for youth reports. TAU had significant reductions for caregiver reports from baseline to months 6 (β = −2.80; 95% CI, −5.16 to −0.45) and 18 (β = −3.82; 95% CI, −6.42 to −1.22) and for adolescent reports from baseline to months 3 (β = −4.02; 95% CI, −5.61 to −2.43), 6 (β = −5.19; 95% CI, −6.84 to −3.54), 12 (β = −6.38; 95% CI, −8.16 to −4.59), and 18 (β = −5.42; 95% CI, −7.11 to −3.73). For RRFT, the reductions in avoidance were statistically significant for caregiver reports from baseline to months 3 (β = −3.22; 95% CI, −5.77 to −0.68), 6 (β = −6.45; 95% CI, −8.97 to −3.94), 12 (β = −6.83; 95% CI, −9.50 to −4.15), and 18 (β = −7.41; 95% CI, −10.29 to −4.53) and for youth reports from baseline to months 3 (β = −3.69; 95% CI, −5.35 to −2.04), 6 (β = −7.10; 95% CI, −8.79 to −5.41), 12 (β = −7.26; 95% CI, −9.10 to −5.42), and 18 (β = −8.15; 95% CI, −9.97 to −6.33). The changes for RRFT and TAU were significantly different, specifically from baseline to month 6 (−6.5 vs −2.8, respectively; β = −3.65; 95% CI, −7.10 to −0.20; d = −0.44) and month 12 (−6.8 vs −1.9, respectively; β = −4.94; 95% CI, −8.63 to −1.26; d = −0.60) based on caregiver reports and from baseline to month 18 (−8.1 vs −5.4, respectively; β = −2.73; 95% CI, −5.21 to 0.25; d = −0.47) based on adolescent reports.

For criterion D (hyperarousal) symptom severity, RRFT and TAU did not differ significantly at baseline. The TAU group had significant reductions for caregiver reports from baseline to months 6 (β = −2.20; 95% CI, −3.64 to −0.75) and 18 (β = −3.07; 95% CI, −4.66 to −1.47) and for youth reports from baseline to months 3 (β = −1.76; 95% CI, −2.98 to −0.54), 6 (β = −3.23; 95% CI, −4.50 to −1.97), 12 (β = −3.64; 95% CI, −5.00 to −2.27), and 18 (β = −3.02; 95% CI, −4.31 to −1.73). For RRFT, significant reductions were observed for caregiver reports from baseline to months 3 (β = −1.67; 95% CI, −3.22 to −0.12), 6 (β = −3.43; 95% CI, −4.96 to −1.90), 12 (β = −3.87; 95% CI, −5.52 to −2.22), and 18 (β = −3.84; 95% CI, −5.60 to −2.09) and for youth reports from baseline to months 3 (β = −2.94; 95% CI, −4.21 to −1.67), 6 (β = −4.94; 95% CI, −6.23 to −3.64), 12 (β = −5.21; 95% CI, −6.62 to −3.80), and 18 (β = −5.38; 95% CI, −6.77 to −3.99). For caregiver reports, RRFT had significantly greater reductions from baseline to month 12 (−3.9 vs −3.0; β = −2.58; 95% CI, −4.85 to −0.31; d = −0.52). For adolescent reports, RRFT had significantly greater reductions from baseline to month 18 (−5.4 vs −3.0; β = −2.36; 95% CI, −4.26 to −0.46; d = −0.52).

Discussion

This study establishes the efficacy of an integrated, exposure-based treatment approach for adolescents with co-occurring SUP and PTSD symptoms. Risk reduction through family therapy was safely administered with significant improvement in SUP and PTSD criterion symptoms and with no evidence of iatrogenic effects (eg, SUP exacerbation). Risk reduction through family therapy yielded significant within-group reductions across all SUP outcomes from baseline to all postbaseline assessments, and TAU yielded significant within-group reductions across multiple SUP outcomes. Between-group comparisons demonstrated greater reductions for RRFT than TAU on some SUP outcomes (eg, decrease in odds of any marijuana use was 82% greater for RRFT than for TAU from baseline to month 6, 90% greater at 12 months, and 96% greater at 18 months) but not all (eg, alcohol). Youth PTSD symptoms improved in both conditions because RRFT and TAU demonstrated within-group reductions, with some differences by informant. Given that a criterion standard treatment for PTSD symptoms (TF-CBT) was well represented in TAU, it is not surprising that significant between-group differences were not found. It is encouraging that clinically significant PTSD symptom treatment gains were maintained in both conditions through 18 months after baseline because this is the first RCT, to our knowledge, to evaluate these treatments in an adolescent sample with co-occurring SUP and PTSD symptoms. Observed findings in which RRFT yielded greater reductions in avoidance symptom severity (baseline to 6 and 12 months, per caregiver; baseline to 18 months, per adolescent) and hyperarousal symptom severity (baseline to 18 months, per adolescent) align with negative reinforcement theory,³⁰ in which SUP reductions are expected to parallel reductions in PTSD avoidance and/or hyperarousal.

High relapse rates are a major challenge in adolescent and adult SUP treatment.^43,44,45,46 The significant between-group differences in SUP outcomes at longer-term postbaseline assessments observed herein suggest RRFT effects are sustained in the long term. These results fit with RRFT’s focus on the long-term generalizability of skills for youth (eg, realistic refusal skills, addressing the urge to avoid trauma-related cues in more effective ways) and/or caregivers (eg, parental monitoring, random urine drug testing) to promote the durability of SUP-related treatment gains.

Without data to guide treatment decisions for adolescent SUP–PTSD symptom comorbidity, clinical lore that exposure-based treatment may exacerbate substance use has influenced decades of clinical practice. Prior treatment research targeting adolescents with SUP–PTSD symptoms has been limited to 2 controlled pilot studies.^23,47 Building on this modest empirical base, the present study was, to our knowledge, the first full-scale RCT to demonstrate the safety and efficacy of implementing the criterion standard (exposure-based) treatment for PTSD symptoms—integrated with substance use treatment—in a large sample of adolescents with SUP–PTSD symptoms. The significant within-group reductions in substance use outcomes in the TAU condition, in which an exposure-based treatment for PTSD symptoms was well represented, are also noteworthy. These results complement recent treatment research supporting the safety and utility of exposure-based treatments for adults with co-occurring SUP and PTSD.^{11,12,13,14,15,20} Together findings from this and prior research demonstrate that it is safe to provide exposure-based treatment to this population when combined with substance use treatment, and this approach also may lead to the best long-term SUP and PTSD symptom outcomes.

Strengths and Limitations

Study strengths include a focus on treatment outcomes in an understudied population in one of the largest samples of adolescents who experience IPV to date. We used rigorous methods for randomization, fidelity measurement, treatment characterization, and longitudinal assessment during 18 months. Study retention rates also were high; 109 of 124 participants (87.9%) completed at least 1 postbaseline assessment. Study implementation in a community-based setting (CAC) with real-world clinicians increases the likelihood that results will generalize to youth in other community settings and promotes greater likelihood of uptake of RRFT by frontline clinicians.

One notable limitation was that the sample was primarily female adolescents (87.1%). Male adolescents also experience IPV with PTSD symptoms and are at greater risk for illicit drug use than female adolescents, and replication is needed in a sample with greater representation of male youth. Second, length of treatment was not equivalent for RRFT and TAU. Although we controlled for dosage differences between conditions in the analyses, and there is a long-standing RCT literature wherein TAU and other comparison conditions with unequal dosage (wait-list controls) have been implemented, future studies should delineate the specific mechanisms underlying RRFT treatment gains to determine the effect of dosage on outcomes. Another limitation was the use of the UCLA-PTSD-RI as a clinical interview rather than a structured clinician-administered interview for the measurement of PTSD symptoms. Although the focus of the present study was on PTSD symptoms rather than PTSD diagnosis, a structured interview could have provided a richer clinical picture.

Conclusions

Risk reduction through family therapy is currently the only exposure-based, integrative treatment for adolescents with co-occurring SUP and PTSD symptoms with full-scale RCT support. This study offers empirical guidance for clinical practice where no standard of care exists.

Supplement 1.

Trial Protocol

Supplement 2.

eMethods. Monitoring of Serious Adverse Events

Supplement 3.

Data Sharing Statement