Study design, interventions, and baseline characteristics for the Substance use and TRauma Intervention for VEterans (STRIVE) trial

Abstract

While comorbidity between posttraumatic stress disorder (PTSD) and substance use disorders (SUD) is common among veterans, there is debate regarding how to best treat individuals suffering from both conditions. Despite data supporting the effectiveness of integrated treatments that simultaneously address both disorders, due to concerns that an early focus on trauma may increase dropout and reduce the likelihood of achieving SUD-related goals, providers continue to prefer a sequential approach, where the addiction is treated first and PTSD treatment is instituted following sustained abstinence or reduced use. This project is designed to directly examine these provider concerns by evaluating the benefits and harms of an integrated versus a sequential approach to treating comorbid PTSD and SUD. This paper reviews the study's methodology, treatment approaches, and baseline participant characteristics. In this randomized clinical trial, one hundred eighty-three veterans with co-occurring PTSD and SUD have been randomized to one of two psychotherapies that include the same treatment components for SUD and PTSD (Motivational Enhancement Therapy and Prolonged Exposure respectively), but differ by whether the components are delivered sequentially or are integrated such that PTSD and SUD symptoms are addressed concurrently. We hypothesize that veterans assigned to integrated treatment will show greater improvement in PTSD and SUD symptoms than veterans assigned to sequential treatment. If this hypothesis is supported, the findings have the potential to change clinicians' beliefs and challenge long-standing practice patterns that require participation in SUD treatment prior to initiating trauma-focused therapies for PTSD.

1. Introduction

Posttraumatic stress disorder (PTSD) and substance use disorders (SUD) are two of the most common mental health problems for which veterans seek care within the Veterans Health Administration (VHA) [1,2]. While comorbidity between the two conditions is common (approximately one-quarter of veterans seeking VHA outpatient specialty PTSD care have a comorbid SUD diagnoses), there is an ongoing debate regarding how to best treat individuals suffering from PTSD concurrently with a SUD [3-6]. Historically, the most common recommendation for addressing this comorbidity has been to provide sequential treatment, where the addiction is treated first and PTSD treatment is instituted following sustained abstinence [7-9]. However, data supporting the functional relationship between symptoms of PTSD and substance use (e.g., self-medication) have led to concerns that an early focus on substance use at the exclusion of PTSD may make it difficult for individuals to reduce use or obtain abstinence as trauma-related symptoms go unaddressed [7-16].

As a result, several protocols for integrating psychotherapies for PTSD and SUD symptoms have been developed and evaluated [8,17-19]. These protocols are generally based on cognitive-behavioral principles of change, although they vary in the extent to which they incorporate exposure to trauma memories or cues (e.g., trauma-focus) [19]. Non-trauma focused integrated therapies (e.g., Seeking Safety, Integrated Cognitive Behavioral Therapy) have been shown to reduce PTSD and SUD symptomology in a number of studies; however, a recent meta-analysis concluded that there is insufficient evidence to support their use (e.g., pooled analyses did not find superior PTSD or SUD outcomes for these therapies) [19-21]. The evidence is stronger for trauma-focused integrated treatments, such as the Concurrent Treatment of PTSD and Substance Use Disorders using Prolonged Exposure (COPE) protocol and Sannibale and colleagues' Integrated Therapy [18, 22]. The aforementioned meta-analysis concluded that trauma-focused integrated treatments result in greater reductions in PTSD and SUD symptoms than SUD-only focused treatments [8,19].

Despite this growing body of evidence in support of integrated psychotherapies for the treatment of co-occurring PTSD and SUD, many continue to advocate for sequential treatment, arguing that the stress associated with trauma-focused treatment may lead to early treatment dropout, resulting in both PTSD and SUD going untreated [7-9,13-16]. These persistent beliefs reduce the likelihood that effective integrated treatments will be adopted into regular clinical practice, despite their being the preferred model of treatment by a majority of veterans with comorbid PTSD and SUD [10,23]. No published studies have directly compared the benefits and harms (e.g., dropout) of an integrated versus sequential approach; as such, there are no data available to allay these concerns. Thus, we are using randomized clinical trial (RCT) methodology to experimentally evaluate the clinical effectiveness of these two approaches to treating comorbid PTSD and SUD. This paper describes the design, methodology, and baseline data for Project STRIVE (Substance use and TRauma Intervention for VEterans), a RCT in which veterans are randomized to one of two trauma-focused psychotherapies that include the same treatment components for SUD and PTSD (Motivational Enhancement Therapy [MET] and Prolonged Exposure [PE] respectively), but differ by whether the components are delivered sequentially or are integrated such that PTSD and SUD symptoms are addressed concurrently [24,25].

2. Method

2.1. Overview

Project STRIVE is a two-arm RCT with blinded assessment designed to directly evaluate the benefits and harms of two approaches (sequential versus integrated) for treating comorbid PTSD and SUD. Participants are male and female veterans recruited from February 2011 and June 2015 with co-occurring Diagnostic and Statistical Manual of Mental Disorders – Fourth Edition (DSM-IV) diagnosed PTSD and SUD [26]. Veterans in both conditions receive sixteen 90 minute psychotherapy sessions during which full courses of MET and PE are delivered. However, veterans assigned to the sequential treatment condition receive four weeks of MET followed by twelve weeks of PE, while those randomized to the integrated treatment arm receive components of both MET and PE during each of the 16 treatment sessions. Our primary hypotheses are that veterans assigned to the integrated condition will demonstrate greater improvements in PTSD (as measured by the PTSD Checklist [PCL]) and SUD (abstinence from heavy drinking and non-prescribed drug use as measured by the percent days with either heavy drinking or drug use reported on the Timeline Followback Interview [TLFB]) symptoms than veterans assigned to the sequential treatment during the sixteen week treatment period [27-29]. We secondarily hypothesize that veterans assigned to the integrated treatment will demonstrate (a) greater reductions in PTSD and SUD symptomology (as measured for the primary hypotheses) at six months post-treatment, (b) lower post-treatment clinician-assessed PTSD symptomology, and (c) larger gains at post-treatment and six-month follow-up on secondary outcomes frequently impacted by, and reported in, treatment trials for PTSD and/or SUD (depression, quality of life, alternative drinking measures, alternative anxiety measures) [18,30,31]. The demonstrated functional relationship between PTSD and SUD (e.g. PTSD symptoms often serve as triggers for cravings and use) leads us to hypothesize that a simultaneous focus on both conditions (e.g. integrated treatment) will allow symptoms to improve in concert, yielding optimal outcomes as compared to addressing the conditions one at a time [12,32]. Further, prior data demonstrating that veterans prefer integrated treatment for PTSD/SUD leads us to expect higher levels of veteran engagement and retention in the integrated condition, leading to superior outcomes [10].

2.2. Interventions

Regardless of the condition to which they are randomized, veterans receive full courses of MET and PE; those in the sequential condition receive four sessions of MET followed by twelve sessions of PE, while those in the integrated condition receive elements of both treatments in each of the 16 sessions. The integrated treatment for this study is similar to existing trauma-focused integrated treatments (most existing integrated treatments include non-exposure cognitive-behavioral strategies for managing SUD and PTSD, while our treatment does not) [18,22]. However, due to the difficulty of breaking existing integrated protocols into free-standing, evidence-based sequentially-delivered treatments that contain all of the integrated treatment elements, we developed our own protocol. We chose MET and PE due to their strong evidence base for treating SUD and PTSD respectively and their recent widespread implementation within VHA [33]. It is worth noting that in order to provide veterans in both conditions an equivalent number of sessions (the sequential condition necessitated 16 sessions), our integrated treatment is three to four sessions longer than existing trauma-focused integrated protocols (both intervention conditions consisted of sixteen 90-minute sessions).

PE is an evidence-based psychotherapy for PTSD that is based on the Emotional Processing Theory of PTSD; the four components of PE are: 1) exposure to safe situations, objects, or people that cause distress and are avoided because they are trauma reminders (in vivo exposure), 2) revisiting and processing of the trauma memory (imaginal exposure), 3) psychoeducation about trauma-related symptoms, and 4) breathing retraining [24]. MET is an evidence-based, structured adaptation of motivational interviewing that uses assessment and feedback to encourage change [25]. Motivational interviewing is defined as “a collaborative, goal-oriented style of communication with particular attention to the language of change…designed to strengthen personal motivation for and commitment to a specific goal by eliciting and exploring the person's own reasons for change within an atmosphere of acceptance and compassion.” [34] Abstinence is encouraged, but not mandated, as a treatment goal. In both the integrated and sequential conditions, PE and MET are provided by the same therapist. To account for sessions missed by the veteran or provider, veterans are allowed 20 weeks from randomization to complete the 16 sessions. Veterans who present to treatment sessions intoxicated are seen at the discretion of the therapist.

2.2.1. Sequential condition

In the sequential condition, veterans begin treatment with four 45-minute weekly MET sessions (Table 1). Session one focuses on assessing veterans' motivation for change, severity of use, and consequences of addiction. During sessions two through four, therapists review substance use and urges to use since the last session, provide feedback from the assessment, and use motivational interviewing strategies to promote the patients' motivation to reduce their substance use. Specifically, therapists work with patients to resolve ambivalence for change, facilitate change, and support any self-change efforts. In the sequential condition, health education content is also presented for 45 min in sessions one through four in order to ensure equal therapist contact in the two intervention conditions. The topics of the health behavior psychoeducation sessions include: diet/weight management, exercise, smoking and medical treatments, and healthy lifestyle changes. A 12-session course of PE begins at session five of the sequential condition and continues through the end of treatment. Beginning with session five, no MET is delivered and while substance use is assessed regularly, addiction is discussed only as it directly relates to trauma triggers or resistance to exposure exercises.

Table 1.

Overview of study design: Interventions and assessments.

Visit	SC	BL/S1	S2	S3	S4	S5	S6	S7	S8	S9	S10	S11	S12	S13	S14	S15	S16	6-Month FU
Integrated condition		MET (45 min)		MET check-up (15 min)
		PE (45 min)		PE (75 min)
Sequential condition		MET (45 min)				PE (90 min)
		Health education (45 min)
Assessments	X	X			X				X				X				X	X

SC = screening; BL = baseline; S = session; FU = follow-up; MET = Motivational Enhancement Therapy, PE = Prolonged Exposure.

2.2.2. Integrated condition

In the integrated condition, all sessions include MET and PE components. Sessions one through four include the same MET content as sessions one through four of the sequential condition, although the MET portion of the sessions is shortened to approximately 15–30 min from session 3 forward. Starting with session five, the MET portion of the session focuses on current substance use and progress toward goals. During the final few sessions, therapists also discuss post-treatment plans and provided encouragement for sustained reduced substance use or abstinence (as applicable). Regarding PE, following the session one MET components, therapists provide an overview of PE and how PE reduces PTSD symptoms and teach breathing retraining. During session two, the PE components include a discussion of common reactions to trauma, presentation of the rationale for in vivo exposure, and construction of the in vivo exposure hierarchy. During session three, the rationale for imaginal exposure is presented and imaginal exposure and processing are conducted. Imaginal exposure and processing of the imaginal exposure experience continue for the remainder of the sessions, with exposure to hot spots (e.g. the most intense part of the trauma memory) typically beginning around session six. In the integrated condition, PTSD and SUD are conceptually linked throughout the treatment process. For example, during MET veterans are asked to reflect on how their use relates to their PTSD, SUD treatment is woven into the PE rational, substance use is emphasized as a common reaction to trauma, attention is given to make sure that no in vivo hierarchy items are avoided because they help reduce use or maintain abstinence, urges to use are assessed prior to and following imaginal exposure, and processing includes a discussion of the relationship between veterans' trauma and substance use.

2.2.3. Therapists, training, and supervision

Thirteen licensed doctoral-level psychologists and masters-level clinical social workers deliver the treatment. One part-time therapist is paid by the study at each site; all other therapists are existing VA clinicians who have been approved to donate time to the study and treat participants as part of their regular caseload. In order to reduce the risk of imbalance in therapist skill across the two treatments, all therapists deliver both the integrated and sequential treatments. Treatment fidelity assessments will be conducted in order to verify adherence to the randomized treatment condition (see below). At the start of the study, therapists completed a one and one-half day in person MET training with Dr. Van Horn, a member of the Motivational Interviewing Network of Trainers. Therapists that join during the course of the study (e.g. after the first randomization) complete one-on-one MET training that includes videotaped practice. All study therapists complete four days of PE training and supervision of two cases as offered by VHA's Mental Health Dissemination Initiative [35]. Therapists who completed the PE training prior to the start of the study (e.g. through their clinical VHA positions) also completed a two-day in person PE refresher led by Dr. Foa before treating their first study participant. This training also included strategies for integrating MET and PE (e.g., integrated rational for treatment; integrating checks of substance use and craving into PE homework review). All therapists are provided with treatment manuals for both the integrated and sequential treatments. While treating participants, therapists participate in weekly case consultation led by experts in PE (Dr. Yusko) and MET (Drs. Van Horn and Drapkin). Supervision includes regular review of session video, encourages skillful adherence to the treatment condition including strategies for integration, and provides a forum for problem-solving challenges, Study therapists from both sites attend the same weekly supervision phone call to minimize site differences in treatment delivery.

2.2.4. Fidelity assessment

Treatment sessions are videotaped and 100 randomly selected videotapes will be coded to ensure fidelity to the treatment conditions (25 sessions will be selected from each of the following groups: sessions 1–4, session 5, sessions 6–9, and sessions 10 and above). Fidelity assessments will be conducted by experienced PE fidelity assessors at the Center for the Treatment and Study of Anxiety and Dr. Van Horn using established methods for assessing PE fidelity and the Yale Adherence and Competence Scale for assessing MET fidelity [36]. The first three sessions will be rated by all assessors and discussed until consensus is reached in order to establish interrater reliability. Sessions will be rated for both the presence and absence (e.g., no PE components during the first four sessions of the sequential condition) of PE and MET elements.

2.3. Measures

Following psychotherapy trial best practices, all post-randomization measures are completed by an assessor blinded to intervention condition. See Table 2 for the full assessment schedule. Completion of assessment measures is estimated to take approximately 2.5 h at the screening visit; 30–40 min at the baseline, week 4, week 8, week 12, and six-month post-treatment visits; and 60 min at the post-treatment visit. Participants are compensated up to $470 for completing study assessments.

Table 2.

Assessment schedule.

Measure	Screening visit	Baseline visit/session 1	Sessions 4, 8, & 12	Session 16 (post-treatment)	6-Month follow-up
Clinician-administered
Structured Clinical Interview for DSM-IV (SCID)	X
Mini International Neuropsychiatric Interview (MINI)	X
PTSD Symptom Scale Interview (PSS-I)	X			X
Timeline Follow Back Interview (TLFB)	X	X	X	X	X
Self-report
PTSD Checklist (PCL)	X	X	X	X	X
State-Trait Anxiety Inventory (STAI)	X			X	X
Short Inventory of Problems (SIP-R)		X	X	X	X
Brief Addictions Monitor (BAM)		X	X	X	X
Patient Health Questionnaire – 9 (PHQ-9)		X	X	X	X
Medical Outcomes Study Short Form (SF-12)		X	X	X	X
Non-VA mental health services		X	X	X	X
Other
Urine drug screen	X	X	X	X	X

PTSD = posttraumatic stress disorder; SUD = substance use disorder.

2.3.1. Clinician-administered scales

The SUD and PTSD modules of the Structured Clinical Interview for DSM-IV (SCID-IV) are used to establish eligibility [37]. All modules of the Mini International Neuropsychiatric Interview except SUD, PTSD, Bulimia, and Anorexia are administered to assess current comorbid mental health diagnoses in order to determine eligibility and describe the sample [38]. A clinician-assessed measure of current PTSD symptomology is obtained using the PTSD Symptom Scale-Interview Version (PSS-I; secondary outcome) [39]. The TLFB is used to determine eligibility and will be used as a primary outcome measure; the TLFB yields a complete record of drinking/drug use and time spent in controlled environments (e.g. inpatient settings) from sixty days prior to the study through the six-month follow-up [29].

2.3.2. Self-report measures

The PCL is a 17 item self-report measure designed to evaluate the severity of PTSD symptoms in the prior month; it is used to determine eligibility and will serve as a primary outcome [27,28]. The State-Trait Anxiety Inventory, a 40-item inventory that evaluates state anxiety at the time the questionnaire is completed and the enduring trait to experience anxiety, is being administered as a secondary anxiety outcome [40]. Self-report measures related to drinking and drug use include: the Short Inventory of Problems (SIP-R), a 15-item questionnaire designed to measure adverse interpersonal, physical, social, impulsive, and intrapersonal consequences of alcohol and drug abuse (secondary outcome); and the Brief Addiction Monitor, a 17-item assessment used as a national performance measure for addiction treatment throughout VA (secondary outcome) [41,42]. Depressive symptoms are assessed using the Patient Health Questionnaire – 9, a checklist of DSM-IV depression symptoms that is a reliable measure of the severity of depressive symptoms (secondary outcome) [43]. The Medical Outcomes Study Short Form (SF-12) is administered as a measure of quality of life (secondary outcome); it is divided into physical and mental component subscales [44,45]. Participants also provide demographic information and report on non-VA mental health service use and participation in self-help groups (e.g. Alcoholics or Narcotics Anonymous).

2.3.3. Other measures

Urine drug screens are administered at each research visit and are used to detect the presence of opiates, cocaine, benzodiazepines, and marijuana. VA mental health and substance use service utilization data in the twelve months prior to study entry, the duration of the study, and twelve months after the study period and psychiatric medications prescribed during the course of the study will be extracted from the electronic medical record.

2.4. Procedures

This study is approved by the Institutional Review Boards of the Minneapolis and Philadelphia VHA Medical Centers. Participants are recruited at these two sites through a variety of channels, including provider referrals, advertisements in the VHA medical centers, and direct recruitment of potentially eligible veterans identified via the electronic medical record. Potential participants are prescreened in person or by telephone; the prescreening consists of general questions regarding age, alcohol and drug use, and PTSD symptomology. Those who meet the prescreening eligibility and indicate interest in participating are scheduled for an in-person screening evaluation.

At the start of the in-person screening evaluation, voluntary informed consent is obtained following an explanation of all assessments, lab procedures, treatment conditions, confidentiality, study payments, and under what circumstances they may be prematurely discontinued from treatment. Participants are also asked for permission to video record their treatment sessions and to authorize research staff to contact one or two significant others to help locate subjects if they fail to return for study visits. Finally, participants are notified that a blood alcohol content reading will be required at each research visit and that the reading must be 0.00 to complete the research visit; if participants have a blood alcohol content reading greater than 0.00 the research visit is rescheduled. Once informed consent is obtained, a diagnostician and a study staff member administer the baseline assessments as outlined in Table 2. At this time, outpatient medical detoxification is initiated as clinically indicated. Participants who meet the inclusion/exclusion criteria are scheduled for a randomization visit within 30 days of the screening visit, during which they meet with study staff to complete the final pre-randomization assessments (see Table 2). Participants are then randomized to either the sequential or integrated condition and meet with their therapist for the first session. High levels of treatment dropout reported in prior trials of participants with co-occurring PTSD and SUD (e.g. only 18% of participants randomized to receive COPE completed the full course of treatment) lead us to anticipate that many veterans will not complete a full course of treatment [18,19]. However, given that we hypothesize that better adherence may lead to superior outcomes for the integrated condition, in order to increase the external validity of our findings, strategies beyond those typically used by VHA clinicians to re-engage veterans in treatment (e.g. three attempts at contact following a missed treatment session) are not being employed.

2.4.1. Randomization procedures

Subjects are randomized within each site to the integrated or sequential treatment. Randomization is computer generated and stratified by site, the presence of illicit drug use (versus alcohol use only), the severity of PTSD symptoms as measured by the PSS-I (high versus low with cut-point of 25), and service era (pre-Gulf War I versus Gulf War I and later). In order to maintain blinded research assessments, prior to the first session therapists are provided with a sealed envelope containing a postcard indicating the model of care. The therapist then adds the participant's study identification number to the postcard and mails it to an unblinded staff member at the Philadelphia VA Medical Center. This results in all assessors (study staff and diagnostician) remaining blinded to the treatment condition.

2.4.2. Coordination across study sites

In addition to adhering to the same study protocol, standardization of procedures across the two study sites is ensured via weekly phone conferences between the project coordinators and twice-monthly phone conferences with the principal investigator (PI), all co-investigators (including the Minneapolis VAMC site PI), and study staff at both sites. The PI and Minneapolis VA Medical Center site PI also have annual in person meetings to review study progress and procedures.

2.4.3. Data monitoring committee

The project has a three-person data monitoring committee assigned by the funding agency. The committee meets quarterly to review participant enrollment and withdrawal, changes to the study protocol, and adverse events. The committee also reviews serious and unexpected adverse events on an ongoing basis as they occur.

2.5. Participants

Participants include 183 male and female veterans with comorbid PTSD and SUD at two VHA Medical Centers (Minneapolis and Philadelphia). See Fig. 1 for the study recruitment flowchart. Eligibility requirements include: (1) age 18 or older, (2) enrollment in VHA care, (3) current DSM-IV diagnosis of chronic PTSD, (4) a PCL score of at least 50, (5) current abuse or dependence on alcohol, stimulants such as cocaine, opioids (including prescription opioids), or benzodiazepines, (6) alcohol or drug use at least 10 out of 30 days prior to signing consent, (7) able speak and read English and provide informed consent. Of note, participants can be abusing or dependent on nicotine or marijuana but those are not considered sufficient for inclusion. Veterans are excluded from the study if they meet any of the following: (1) current suicidal or homicidal ideation with intent and/or plan that, in the judgment of the investigator, should be the focus of treatment, (2) current DSM-IV-based diagnosis of bipolar affective disorder, schizophrenia or any psychotic disorder, (3) unstable or serious medical illness including history of stroke, seizure disorder, or unstable cardiac disease, (4) history of moderate or severe traumatic brain injury with cognitive impairments that may interfere with treatment, (5) participation in PE in the last 6 months as we believe beginning a new course of PE immediately following a prior trial would not be clinically indicated in most cases, (6) initiation of a new psychotherapy program in the last 2 months, (7) active participation in a formal addiction treatment program (e.g. any visit in the program in the prior month and pending future appointments for the treatment of addictions), (8) change in psychotropic medication in the one month prior to randomization except for the use of oxazepam for alcohol detoxification or a taper of a previously used benzodiazepine, and (9) therapeutic use of a benzodiazepine greater than the equivalent of 40 mg of diazepam at the time of randomization. While veterans are excluded if they recently initiated new treatment (psychotherapy or medication), given that MET action plans often include engagement in services, initiation of new services following randomization is allowed. Further, veterans are allowed to participate in ongoing maintenance treatments (e.g. case management, medication management) prior to randomization and during treatment.

Fig. 1.

Study recruitment and randomization. *Participants may have met more than one exclusion criteria; the primary reason for exclusion provided by the study coordinator is reported.

2.5.1. Baseline characteristics

Table 3 presents select demographic and baseline characteristics for the 183 randomized veterans. Consistent with the larger VHA population, our sample is predominately male. The most commonly reported race (approximately 50%) is Black or African American. The most commonly reported service era is Gulf War (40.4%; which includes Gulf War I/Operation Enduring Freedom [OEF]/Operation Iraqi Freedom [OIF]/Operation New Dawn [OND]); 15.3% of the sample is Vietnam era veterans. The majority (85.2%) of participants meet DSM-IV criteria for alcohol dependence, 8.7% meet criteria for alcohol abuse but not dependence, and 6.0% do not meet criteria for an alcohol use diagnosis. No participants meet criteria for drug abuse, while 18% meet criteria for drug dependence. Participants' baseline PTSD scores are similar to those reported in other clinical trials of veterans with PTSD [30]. On average, participants used non-prescription drugs (including marijuana) 23.6% of the 30 days prior to randomization.

Table 3.

Selected baseline characteristics of the intent to treat sample.

	Total sample (N = 183)	Integrated condition (N = 95)	Sequential condition (N = 88)	Test statistic	P-value
Age, years: M(SD)	44.1(13.0)	44.4 (13.1)	43.8 (13.0)	t(181) = 0.35	0.73
Sex, N (%) male	169 (92.3)	90 (94.7)	79 (89.8)	χ2 = 1.59	0.207
Race, N (%) White	75 (41.0)	37(38.9)	38(43.2)	χ2 = 0.34	0.561
Ethnicity, N (%) Hispanic or Latino	9 (4.9)	7(7.4)	2(2.3)	χ2 = 2.54	0.111
Marital status, N (%) married	67 (36.6)	44(46.3)	23(26.1)	χ2 = 8.02	0.005
PTSD Checklist (PCL)	63.0 (10.9)	62.5(11.3)	63.6(10.3)	t(181) = −0.66	0.513
% days heavy drinking	53.8 (29.6)	51.7(29.5)	56.0(29.6)	t(181) = −0.99	0.325
PTSD Symptom Scale Interview (PSS-I)	36.4 (6.6)	35.8(6.9)	36.9(6.4)	t(181) = −1.12	0.263
Patient Health Questionnaire (PHQ-9)	15.4 (5.7)	14.7(5.8)	16.1(5.5)	t(181) = −1.75	0.083
Medical Outcomes Study Short Form (SF-12) mental component	29.2 (9.7)	30.9(9.4)	27.3(9.8)	t(181) = 2.52	0.013
Medical Outcomes Study Short Form (SF-12) physical component	45.3 (7.8)	45.3(7.6)	45.2(8.0)	t(181) = 0.05	0.96
Short Inventory of Problems (SIP-R)a	17.9 (10.7)	16.4(10.6)	19.5(10.6)	t(180) = −2	0.047

PTSD = posttraumatic stress disorder.

^an = 182 for this variable.

Randomization was checked by comparing the integrated and sequential groups on relevant background variables (e.g., sociodemographic characteristics, baseline symptomatology). We used independent sample t-tests for continuous variables and Pearson chi-square tests for categorical variables. Randomization successfully balanced most demographic variables and baseline symptomology across the two conditions, although participants in the integrated condition are more likely to be married, have high higher SF-12 mental component scores, and report lower SIP-R scores than those in the sequential condition.

2.6. Sample size and power considerations

Our study is powered to detect between-group differences in the outcomes associated with our primary hypotheses. We based our power estimates on the methods of Hedeker et al. [46]. We assumed that attrition from research assessments will result in 20% loss to follow-up through the 16 weeks and assumed an exchangeable covariance matrix with correlation coefficients of 0.5 between any two time points (correlations were purposefully inflated to adjust for the planned use of an unstructured covariance matrix for data analysis). To account for having two primary hypotheses, we used a two-sided alpha level of 0.025. With these assumptions, our sample size of approximately 91 per group provides over 80% power to detect a standardized main effect of d = 0.40 (small-sized effect) for either outcome [47]. The sample size yields over 80% power for a standardized linear group by time interaction effect of d = 0.45 (small to medium-sized effect) at the post-treatment time point [47].

2.7. Data analysis plan for the completed sample

Prior to performing analyses, we will apply standard data cleaning procedures [48]. Specifically, we will (1) screen the data for data-entry errors, (2) check for outliers, (3) assess the extent and pattern of missing data, and (4) check that appropriate assumptions of normality for data distributions are met whenever necessary.

2.7.1. Primary hypotheses

To test our primary hypotheses (veterans assigned to the integrated condition will demonstrate greater improvements in PTSD [as measured by the PCL] and SUD [percent days with heavy drinking or drug use as measured by the TLFB] symptoms than veterans assigned to the sequential treatment during the study period), we will perform mixed-effects regression analysis of the PCL data and a mixed effects Poisson regression (with offset of number of available days per time period) for the TLFB data. The two measures will be analyzed using an information-criterion-selected covariance matrix for the four repeated measures within subject. This analytic procedure will make use of all available data for all participants and accounts for participant-level clustering. We will also test for group by time interactions. To assess the cross-correlations between the two primary responses (association between changes in PTSD and SUD symptoms), we will also fit a bivariate mixed-effects model using the four repeated responses for each of the two outcomes to create an eight-dimensional repeated measures response, with group, time, domain (PTSD or SUD) and their interactions as fixed effects.

We will analyze the data under an intent-to-treat principle, in which all subjects randomized will be included in these analyses. However, outcomes for our primary hypothesis may be affected by non-adherence to the assigned intervention condition (e.g. missed sessions and treatment dropout). Thus, we will conduct analyses to examine the role of participation in care on outcomes. We will compare the groups on number of sessions missed using a zero-inflated negative binomial regression model and will use an instrumental variable approach to assess the influence of non-adherence on the analyses addressing the main hypotheses [49]. This method is preferable to a completer analysis as it tries to allow for the fact that differences in adherence to treatment may be due to differences between patients that may be correlated with treatment outcomes.

Finally, we will likely have missing data resulting from participants not attending research visits. The mixed effects models described above will allow us to use all data, complete or incomplete, provided by all subjects. The results of these analyses will be valid so long as missing data is ignorable. To address the possibility that the missing data is not ignorable, we will also combine models for the missing data (“selection models”) with those for the complete data, and will re-run the primary analyses described above in this more general framework [50]. When there are a relatively small number of time points such as in this study, this amounts to using repeated measures logistic regression models to predict missed versus made on a visit by visit basis, from covariates and responses collected prior to a visit. We will then use the predicted probabilities of completion of visits as inverse probability weights in reruns of the analyses for the main hypotheses.

2.7.2. Secondary hypotheses

To test secondary hypothesis “a” (veterans assigned to the integrated condition will demonstrate greater reductions in PTSD and SUD symptomology [as measured for the primary hypotheses] at six months post-treatment) the analyses described above will extend the treatment phase mixed effects models for PTSD and SUD to include the six-month follow-up visit. To examine secondary hypothesis “b” (veterans assigned to the integrated condition will demonstrate greater reductions in clinician-assessed PTSD symptomology), we will compare the groups on post-treatment clinician-assessed PTSD using a regression model, with post-treatment PSS-I score as response, baseline PSS-I score as a covariate, and intervention group as a binary factor. Finally, to test secondary hypotheses “c” (veterans assigned to the integrated condition will demonstrate greater reductions in secondary outcomes), we will use similar models as to those in the primary analyses to compare depression, quality of life, alternative drinking measures, and alternative anxiety measures at post-treatment and extend those models to include the six-month follow-up as described above, using generalized linear mixed effects models chosen according to the response distributions. Because our power calculations are based on our primary hypotheses, standardized effect sizes will also be calculated for all outcomes related to the secondary hypotheses to explore the possibility that clinically significant differences may be present in the absence of statistical significance.

3. Discussion

This paper describes the method and baseline data for a study that is using RCT methodology to directly examine two competing approaches to treating comorbid PTSD and SUD: an integrated approach in which evidence-based treatments for PTSD and SUD are delivered alongside each other with treatment elements targeted toward both disorders delivered during each session and a sequential approach during which the evidence-based treatments for SUD and PTSD are delivered one-after-another with minimal attention paid to the co-occurring disorder. Despite emerging data demonstrating the effectiveness of trauma-focused integrated treatments, persistent beliefs regarding the potential harms of integrated treatments as compared to a sequential approach may impede their uptake [23]. Clinicians and researchers who advocate for sequential treatment argue that the stress associated with PTSD treatment (especially trauma-focused therapy) may lead to difficulty achieving abstinence (or meeting other SUD-related treatment goals) or early treatment dropouts resulting in both SUD and PTSD ultimately going untreated for many patients [7-9,13-16]. This is the first study to directly examine this clinically important and widely debated issue regarding the relative merits and harms of these two approaches to treating comorbid PTSD and SUD. It is also responsive to the National Academy of Medicine's call for additional research on the treatment of military veterans with PTSD [51].

Our approach to examining the harms and benefits associated with an integrated versus a sequential approach to treatment has several strengths. We are employing methodological best practices for psychotherapy research including blinded assessments, monitoring of condition fidelity, assessment of longer-term outcomes, and continuing research assessments for veterans who discontinue treatment. An additional strength is the use of two psychotherapies with strong evidence for their efficacy that have already been widely disseminated throughout VHA [33]. If, as hypothesized, the integrated treatment yields better outcomes than the sequential condition, having clinicians already trained in PE and MET may facilitate a more rapid adoption of an integrated approach into practice than could occur if therapists had to be trained in an entirely new protocol. Further, the study is able to uniquely leverage VHA's national implementation of PE. By utilizing mostly existing VHA clinicians trained in PE through the Mental Health Dissemination Initiative as study therapists, we are able to conduct a larger trial than would have been practicable for a similar cost outside of the VHA system. A final strength is the incorporation of elements of practical clinical trials to increase the likelihood that our findings will generalize beyond the trial [52,53]. Examples of practical elements include having therapists deliver both treatment conditions, recruiting through a wide variety of channels, assessing a broad range of outcomes, using measures frequently employed in regular clinical care for the primary outcomes (e.g., PCL), enrolling participants who have previously engaged in SUD or trauma-focused therapies, and allowing concurrent mental health treatments (both pharmacological and psychotherapeutic).

We also acknowledge several limitations to our design. First, the sequential condition is not an exact proxy for how a similar sequential approach might be delivered in routine clinical practice. For example, the SUD and PTSD treatments may be delivered by different therapists, the SUD treatment may last longer than four weeks, and interventions other than MET may be provided (either in conjunction or in place of MET). Further, in considering sequential treatment, we are aware that some clinical programs require a patient to establish a set period of sobriety before engaging in trauma-focused therapies. We considered replicating this requirement in the design but rejected the idea because the amount of time required to establish abstinence varies considerably by patient and allowing patients to select whether or not abstinence is their treatment goal is more consistent with the spirit of MET [25]. A second limitation is the lack of a SUD and/or PTSD only treatment condition. While a recent systematic review provided evidence that integrated treatments outperform SUD-specific treatments, it is possible that trauma-focused therapies without a SUD component may perform as well (or better) than integrated treatments [8,19]. The only known study that has examined this issue found that while PE delivered with a pharmacological placebo resulted in significant decreases in PTSD and SUD symptoms, PE delivered concurrently with naltrexone (an evidence-based medication for SUD) resulted in a greater reduction in percent days drinking at post-treatment and six-month follow-up [17]. A further limitation is that we are not powered to statistically detect small differences between the two treatment arms; while we believe effects smaller than what we are powered to detect are unlikely to be clinically significant, in order to address this concern we will also examine standardized effect sizes to determine if a small between-group effect may be present in the absence of statistical significance. Finally, despite our efforts to make the two treatment conditions as similar as possible (other than the presence or lack of integration), depending on how much time therapists spend delivering MET in sessions three to sixteen of the integrated treatment, those assigned to that condition may spend more time on MET components and less time on PE components than those in the sequential condition. While we believe it is unlikely given recent data suggesting that outcomes obtained with sixty-minute PE sessions do not differ from those resulting from ninety-minute PE sessions, it is possible that differences that are observed between the two conditions may be due to differing amounts of time engaged in PE [54].

The baseline data presented above demonstrate that our sample is similar to the population of veterans seeking outpatient VHA specialty PTSD treatment (with and without SUD) in several important domains, including age and baseline PTSD symptom severity [55]. The racial and ethnic identity of our participants is less representative; we have a higher proportion of African American veterans and a lower proportion of Caucasian and Hispanic veterans than outpatient PTSD specialty treatment clinics [55]. Due to the underrepresentation of Hispanic veterans in our sample, caution will need to be exercised in generalizing findings to that population of veterans. However, our near equal number of Caucasian and African American veterans leaves us well positioned to explore racial differences in the presentation and treatment of comorbid PTSD and SUD. We also have a higher proportion of veterans from the Gulf I/OEF/OIF/OND service eras than is observed in outpatient PTSD clinics [55]. This is likely due to our exclusion of veterans from other service eras during the first two years of recruitment. The project is funded through a Research Funding Announcement specific to the treatment of OEF/OIF/OND veterans with PTSD and SUD. However, despite expanding our recruitment strategies beyond clinic-based referrals as originally proposed, we were unable to meet our recruitment targets with this exclusive focus. Several recently completed clinical trials that recruited veterans with PTSD have enrolled larger numbers of Vietnam era veterans than veterans from the recent conflicts, suggesting that OEF/OIF/OND veterans may be harder to engage in research than those from prior eras [21,56,57]. Given emerging evidence that Vietnam and OEF/OIF/OND era veterans respond similarly to trauma-focused therapies, we were granted a project modification to broaden our inclusion criteria to all eras [58]. Following the modification, we began stratifying randomization by service era which will enable us to explore differences in response by service era. Finally, while randomization successfully balanced the majority of baseline and demographic characteristics across conditions, the two groups do differ on some measures; variables for which the groups show significant differences will be considered for inclusion as covariates in sensitivity analyses related to the primary and secondary hypotheses.

4. Conclusions

Project STRIVE will provide the first direct evaluation of a long-debated topic in the treatment of a high priority group of veterans – whether an integrated or sequential approach is best suited to reducing associated symptomology among veterans with PTSD and SUD. Several evidence-based, trauma-focused integrated treatments are currently available to therapists, including COPE, which was recently suggested as an integrated treatment by the Substance Use and Mental Health Services Administration's Treatment Improvement Protocols for Trauma-Informed Care in Behavioral Health Services [59]. However, despite the demonstrated effectiveness of such treatments, persistent therapist beliefs regarding the harms of such approaches as compared to a sequential approach likely limit the implementation of these effective therapies [23]. Findings from this RCT will provide data that are able to explicitly address these concerns; if as hypothesized the integrated condition proves to be more effective than the sequential condition, the findings have the potential to change these therapist beliefs. Given VHA's ongoing efforts to reduce organizational barriers to integrated care (e.g. hiring of SUD/PTSD specialists at all VHA Medical Centers) and veterans' demonstrated preference for integrated approaches, changing providers' perceptions of the acceptability and safety of these treatments has the potential to result in a substantial shift in practice patterns, including the long-standing practice of requiring SUD treatment prior to initiating evidence-based, trauma-focused therapies for PTSD [10,23].

Abbreviations:

PE

Prolonged Exposure

MET

Motivational Enhancement Therapy

PCL

PTSD Checklist

TLFB

Timeline Followback

OEF

Operation Enduring Freedom

OIF

Operation Iraqi Freedom

OND

Operation New Dawn