Using the Research Domain Criteria Framework to Track Domains of Change in Comorbid PTSD and SUD

Abstract

Objectives

Comorbidity in diagnosis is quite common and raises critical challenges for psychological assessment and treatment. The Research Domain Criteria (RDoC) Project, put forth by the National Institutes of Mental Health, proposes domains of functioning as a way to conceptualize the overlap between comorbid conditions and inform treatment selection. However, further research is needed to understand common comorbidities (e.g. posttraumatic stress disorder (PTSD) and substance use disorder (SUD)) from an RDoC framework and how existing evidence based treatments would be expected to promote change in the RDoC domains of functioning. To address these gaps, the current study examined change in three RDoC domains of interest (Negative Valence Systems, Arousal/Regulatory Systems, and Cognitive Systems) during concurrent prolonged exposure (PE) and substance use treatment.

Method

Participants were 85 individuals with co-occurring PTSD and SUD who received PE in a residential substance use treatment facility. They completed an experimental task to assess physiological reactivity to trauma and alcohol cues at pre- and post-treatment.

Results

Results showed decreased severity in all three RDoC domains of interest across the study period. Pairwise comparisons between domains revealed that Arousal/Regulatory Systems had the lowest severity at post-treatment. Subsequent hierarchical linear regression analyses showed that post-treatment domain scores were associated with post-treatment cue reactivity for trauma and alcohol cues.

Conclusions

The findings provide preliminary evidence of how the RDoC domains of functioning may change with evidence-based treatments for DSM-5 disorders, and are discussed in terms of the assessment and treatment of mental health problems using the RDoC framework.

Posttraumatic stress disorder (PTSD) commonly co-occurs with substance use disorders (SUD); a comorbidity linked to poorer social adjustment, worse treatment outcomes, and greater risk for early treatment termination relative to PTSD or SUD alone (Ouimette, Brown, & Najavits, 1998). Limited evidence-based treatments exist that directly address this comorbidity; instead, treatment of comorbid PTSD-SUD involves administering individual treatments either in conjunction or sequential order. Consistent with research initiatives from the National Institutes of Mental Health (NIMH), one way to develop novel approaches involves conceptualizing comorbidities in terms of shared domains of functioning or transdiagnostic factors present in this comorbidity. In turn, conceptualizing these comorbidities will improve our understanding of treatment response and inform treatment selection and development. In line with this strategy, the current study sought to assess shared domains of functioning in PTSD-SUD and examine whether existing evidence-based treatments promote change in these shared factors.

Despite promising attempts to shift focus towards transdiagnostic treatment (Ellard, Fairholme, Boisseau, Farchione, & Barlow, 2010), the Diagnostic and Statistical Manual of Mental Disorders (DSM-5; American Psychiatric Association, 2013) uses a categorical approach, basing diagnoses on the identification of symptoms of a prototypical disorder as defined through expert clinical consensus. These categories emphasize reliability by ensuring the same criteria are used across diverse settings, but lack validity by failing to support the findings of physiological and genetic research (Insel et al., 2010). Furthermore, the a priori acceptance of DSM-5 diagnoses has limited etiological research to investigating mental health problems in a top down approach rather than allowing research to inform diagnostic categories (Cuthbert & Insel, 2013).

As a result of systematic failure to integrate research on clinical neuroscience, the NIMH launched the Research Domain Criteria (RDoC) Project. RDoC is a research framework aimed at overcoming the limitations of traditional categorical approaches by identifying transdiagnostic mechanisms of mental disorders and incorporating multiple units of analyses (e.g., behavioral and biological) with the goals of improving the current classification system and informing treatment development (Cuthbert & Insel, 2013). To do this, RDoC has identified five domains, each representing a grouping of behaviors with common functional purpose (NIMH, n.d.) that may be examined at biological, behavioral, and cognitive levels: the Negative Valence System represents emotional responses to aversive stimuli; the Positive Valence System focuses on motivated behaviors such as reward-seeking; the Cognitive System relates to behavioral measurement of attention, perception, memory, language, and cognitive control; the Arousal/Regulatory System examines general physiological responses to internal/external stimuli, which represent activation of the nervous system and homeostatic regulation; and the System for Social Processes focuses on responses in interpersonal settings including affiliation, attachment, and social communication.

RDoC aims to create and use a research-informed classification system to identify targets for treatment along with clinical subgroups that would be more likely to respond positively to targeted treatments (Keshavan & Ongur, 2014). RDoC may also be used to explore the sorely under-investigated effects of existing evidence-based treatments on common features across diagnoses. Indeed, treatments for specific disorders often lead to broader improvement than just in the target diagnosis (e.g., Hien, Cohen, Miele, Litt, & Capstick, 2004); however, it is not always clear how treatment effects are generalized, or the extent to which they are the result of addressing shared mechanisms. The RDoC domains present a useful framework for examining these shared mechanisms that could address this gap in the literature and inform the selection of treatments, particularly for comorbid conditions.

Traumatic stress has been posited as a vulnerability factor for SUD (Hien, Cohen, & Campbell, 2005) as a considerable proportion of patients (15–41%) seeking treatment for SUD meet diagnostic criteria for PTSD (Brady, Back, & Coffey, 2004). Indeed, substance use is believed to initially serve as an avoidance coping strategy for the intrusive trauma-related cognitions and negative affect hallmark of PTSD (Khantzian, 1997). Subsequently, pairing distressing memories with substance use may lead to conditioned behavioral responses (Stasiewicz & Maisto, 1993). Evidence shows that individuals with PTSD-SUD have increased craving and salivation in response to trauma cues even when no alcohol is present (Coffey et al., 2010). Additionally, research suggests that extinction of this conditioned response may be possible in that exposure-based treatments for PTSD treatment, such as prolonged exposure therapy (PE; Foa, Hembree, & Rothbaum, 2007), lead to reduced distress to traumatic memories and decreased drug cravings as measured by self-report and salivation levels (Nosen, Littlefield, Schumacher, Stasiewicz, & Coffey, 2014). Taken together with evidence that PE produces transdiagnostic reductions in symptoms (e.g., depression, anxiety, and substance use; Coffey et al., 2016), these findings posit this comorbidity as a strong candidate for elucidating how treatment (e.g., PE) may impact transdiagnostic domains.

Research, however, has not directly investigated how transdiagnostic domains of functioning, such as RDoC domains, can help explain changes in shared symptoms in PTSD-SUD treatment. Some RDoC domains may be of particular relevance to the PTSD-SUD comorbidity, like those relevant to cognitive-behavioral theory (CBT; Beck, 2011) which uses the connection between thoughts, emotions, and behaviors to understand psychopathology. For instance, RDoC’s Cognitive Systems domain refers to constructs such as attention, perception, memory, and cognitive control. Beyond speaking to the difficulty concentrating and sustaining attention that affects both PTSD and SUD, this domain can be of particular relevance to CBT. Indeed, CBT models for PTSD and SUD identify maladaptive cognitions related to control, self-efficacy, and esteem/self-perception for both disorders (e.g., “I am in control of my drinking,” “My memories are uncontrollable,” “I am a bad person”). These thoughts can be intrusive in nature and lead to distress and avoidance attempts, which in turn can lead to further behavioral difficulties in PTSD-SUD. Also related to the CBT model is the Negative Valence Systems domain that refers to emotional responses to aversive stimuli (e.g., anxiety, fear, and loss). Negative affect is considered a shared vulnerability for psychopathology and comorbidities, including PTSD (Zoellner, Pruitt, Farach, & Jun, 2014). It has been implicated in the maintenance of SUD by acting as a conditioned stimulus through which conditioned drug responses are elicited (Stasiewicz & Maisto, 1993). In fact, trauma-related negative affect has been found to induce craving in SUD (Coffey et al., 2010) and it is believed to be partly responsible for the strong PTSD-SUD comorbidity. Another RDoC domain of functioning that is relevant to CBT, particularly its model for anxiety is the Arousal/Regulatory Systems domain that includes physiological responses to internal/external stimuli representing nervous system activation. This domain is related to PTSD through the arousal symptom cluster (e.g., heightened startle response, difficulty sleeping). Moreover, the physiological arousal elicited from substance withdrawal is believed to exacerbate PTSD symptoms, which in turn leads to relapse of substance use (Jacobsen, Southwick, & Kosten, 2001). While the Positive Valence and Social Processes Systems, focused on reward seeking and interpersonal affiliations, respectively, may relate to PTSD’s interpersonal relational deficits or SUD through reward seeking behavior, research is more limited regarding how these may explain its shared symptoms. Instead, drawing upon CBT theory and research, it appears Cognitive, Negative Valence, and Arousal Systems are more closely related to shared PTSD-SUD symptomatology thus offering insight into etiology and treatment response.

As such, the primary aim of this study was to better understand the PTSD-SUD comorbidity from an RDoC framework. This included an examination of how concurrent PTSD-SUD evidence-based treatment influenced change in the RDoC defined Cognitive, Negative Valence, and Arousal Systems. To our knowledge, there are currently no validated measures of the RDoC domains of functioning. Therefore, for the purposes of the current study, we generated an assessment tool that was intended to serve as a proxy of the RDoC domains of interest. Using data from a study examining the efficacy of PE among individuals receiving residential SUD treatment (Coffey et al., 2016), we categorized items from existing measures of psychological symptomatology into Cognitive, Negative Valence, and Arousal Systems and tested whether they changed over the course of concurrent PTSD-SUD treatment. With the intention of serving as a preliminary step to integrating the RDoC framework into existing treatment research, we hypothesized that these transdiagnostic self-report domain scores would decrease significantly from pre- to post- treatment. Furthermore, over and above decreases in symptoms, the present study permitted the examination of whether certain transdiagnostic domains of functioning are differentially or similarly impacted by concurrent PTSD-SUD treatment. Additionally, to incorporate biological measurement in mental health research as recommended by RDoC, our secondary aim was to investigate whether changes in these self-report domain scores over the course of treatment would be associated with decreased trauma-cued subjective distress and physiological craving.

Method

Detailed information on the study design and method can be found in Coffey et al. (2016).

Participants

Participants were 85 individuals who were recruited from a residential substance use treatment facility. Inclusion criteria for study were as follows: 18–64 years of age, current alcohol dependence, at least one heavy drinking day in the past 60 days (i.e., women: four standard drinks; men: five standard drinks), and DSM-IV diagnosis of non-combat related PTSD. Participants were excluded from the study for the following reasons: currently experiencing a manic or psychotic episode, acute risk for suicide, current neurological condition, and currently taking benzodiazepine or medication for cravings¹ or alcohol use reduction. Nineteen participants completed fewer than 8 sessions of PE (non-completers; Foa et al., 2005) and did not return for their post-treatment assessment. Therefore, only the 66 participants who completed 8 or more PE sessions and had complete pre and post treatment data were included in analyses. The local IRB approved the study.

The majority of participants were male (n = 48, 56.5%) and ranged in age from 18 to 57 years old (M = 34.54, SD = 10.42). Most participants identified their ethnic background as White/Caucasian (n = 66, 77.6%), with the rest identifying as Black/African-American (n = 17, 20.0%), American Indian/Alaskan Native (n = 1, 1.2%), and Biracial (n = 1, 1.2%). Thirty-eight participants were single (44.7%), 24 were divorced (28.2%), and 59 had completed high school or further education (69.4%). All participants met the DSM-IV diagnostic criteria for current PTSD and alcohol dependence. Participants reported a variety of Criterion A traumatic events including accidents (n = 49, 57.6%), natural disasters (n = 35, 41.2%), sexual abuse (n = 25, 29.4%), parental abuse (n = 36, 42.4%), assaulted by weapon (n = 53, 62.4%), physical assault (n = 46, 54.1%), serious injury (n = 11, 13%), fear of serious injury (n = 30, 35.3%), witnessed traumatic event (n = 51, 60%), tragic death of a family member (n = 33, 38.8%), and other stressors (n = 47, 55.3%). In addition to alcohol dependence, other common SUDs were crack/cocaine dependence (n = 43, 50.6%), sedative dependence (n = 33, 38.8%), and opiate dependence (n = 31, 36.5%). Most participants (n = 82, 96.5%) had comorbid diagnoses including: major depressive disorder (n = 71, 83.5%), generalized anxiety disorder (n = 52, 61.2%), and social anxiety disorder (n = 43, 50.6%).

Measures

Interview for PTSD

The National Women’s Study PTSD Module (NWSP; Resnick, 1996) is a psychometrically sound structured interview that assesses trauma history. In the current study, it was administered at pre-treatment to establish DSM-IV Criterion A for a diagnosis of PTSD. The Clinician Administered PTSD Scale (CAPS; Blake et al., 1995) is a well-validated structured interview that was administered at pre-treatment in order to diagnose current PTSD; it has shown strong internal consistency and inter-rater reliability (Weathers, Keane, & Davidson, 2001). The CAPS assesses both the frequency and severity of symptoms within the past month on a scale from zero to four, with a symptom considered present when the frequency was rated at least one and severity rating at least two (Weathers et al., 2001). The CAPS was administered by trained research team members who completed rigorous training protocol in CAPS administration and received supervision from experts in PTSD assessment and treatment. Inter-rater reliability for the CAPS was kappa = .94 (~ 20% of all interviews). In the event of classification disagreement, the measures were reviewed during key personnel meetings and discussed with the last two authors who are licensed psychologists and experts in the assessment and treatment of PTSD.

Interview for SUD

The Computerized Diagnostic Interview Schedule (C-DIS IV; Robins et al., 2000) is a computerized structured diagnostic interview for DSM-IV psychiatric disorders that was used to diagnose SUD in the current study.

Self-report questionnaires

The following questionnaires were administered at pre- and post-treatment and used for generating the transdiagnostic self-report domain scores.

The Beck Anxiety Inventory (BAI; Beck, Epstein, Brown, & Steer, 1988) is a 21-item measure that assesses the severity of anxiety symptoms. It has been validated for use in both clinical and non-clinical samples with strong internal consistency (current study was α = .94) and reliability as well as discriminative and concurrent validity (Beck et al., 1988). Items are rated on a 4-point Likert scale (0 = not at all and 3 = severely – it bothered me a lot).

The Beck Depression Inventory-II (BDI-II; Beck, Steer, & Brown, 1988) is a widely used and well validated 21-item measure of depressive symptom severity. Items are rated on a 4-point Likert scale, with higher scores indicating more severe symptoms. Internal consistency was α = .91.

The Impact of Event Scale-Revised (IES-R; Weiss & Marmar, 1997) is a 22-item measure of arousal, hypervigilance, and intrusive symptoms related to trauma. Items are rated on a 5-point Likert scale (0 = not at all and 4 = extreme) to indicate the distress associated with each symptom. The IES-R has demonstrated excellent internal consistency in individuals with co-occurring PTSD and SUD (Rash, Coffey, Baschnagel, Drobes, & Saladin, 2008); for the present study α = .91.

The Alcohol Craving Questionnaire-Now (ACQ-Now; Singleton, Tiffany, & Henningfield, 1994) is a 47-item measure of current alcohol craving symptoms. Items are rated on a 7-point Likert scale (1 = strongly disagree and 7 = strongly agree). The ACQ-Now has demonstrated sound psychometric properties (Connolly, Coffey, Baschnagel, Drobes, & Saladin, 2009) with excellent internal consistency (α = .97) in the current study.

Cue reactivity measures

The following measures were used during the cue reactivity task (see Procedure) and served as additional units of analysis to observe the impact of treatment on self-reported RDoC domain scores.

Subjective distress

The Subjective Units of Distress Scale (SUDS) was used to assess subjective distress during the cue reactivity task, and it consisted of asking participants to rate their level of distress on a 0–100 scale (0 = no distress and 100 = extreme distress).

Salivation level

Salivation level was used as a proxy of craving as it has been found to correlate with self-reported alcohol craving (r = .44, Monti et al., 1987) with good test-retest reliability (r =.68, see Navazesh & Christensen, 1982). To measure salivary flow, pre-weighted dental cotton rolls were inserted under the tongue and between the inner cheek and lower gum on each side of the mouth, immediately prior to the cue reactivity task. The dental cotton rolls were weighed again after each imagery-in vivo cue. The amount of salivation is calculated by subtracting the pre- from the post-trial weight (for detailed description, see Monti et al., 1987).

Treatments

Participants were randomized to a modified PE or “healthy lifestyles” control condition; given that the current paper focuses on symptom domain changes associated with PE and that a recent study (Coffey et al., 2016) shows that modified PE had a greater effect relative to control condition, only the modified PE condition was included in the analyses.

Modified PE

PE is a widely used evidenced-based treatment for PTSD involving both imaginal and in vivo exposure techniques. Participants are first provided with psychoeducation about PTSD symptoms, a rationale for PE treatment, and breathing retraining to help them manage arousal symptoms. Participants were initially offered nine sessions, which occurred once or twice a week (Foa et al., 2005). If participants did not experience a 70% reduction in their PTSD symptoms after completing nine sessions, they were offered an additional three sessions. The modified PE sessions lasted 60 minutes instead of the traditional 90-minute sessions to enhance adoption of PE within community substance use treatment facilities. Preliminary work suggests that outcomes do not differ between 60- and 90-minute PE sessions (Nacasch et al., 2015). Although modified PE did not explicitly focus on SUD, participants were provided with psychoeducation about the relationship between PTSD and SUD symptoms (e.g., substances may be used to cope with PTSD symptoms).

Approximately half of participants assigned to the modified PE condition were provided with a 90-minute pre-treatment session of trauma-focused Motivational Enhancement Therapy (MET-PTSD) while the other half was provided one pre-treatment session of relaxation. Given no significant outcome or drop-out rate differences with or without the motivational session (Coffey et al., 2016), these two conditions were collapsed into a single modified PE condition.

Substance abuse treatment as usual

As part of their stay at the residential substance use facility, all participants received standard 6-week treatment as usual for SUD in addition to modified PE. This treatment consisted of daily 3-hour group therapy, recreation therapy, 12-step meetings, individual drug counseling sessions, and drug counseling homework, which was provided by staff from the residential treatment facility unaffiliated with the current study.

Procedure

Brief prescreening and randomization

Prospective participants completed the PTSD Checklist (PCL; Weathers, Litz, Herman, Huska, & Keane, 1993) and the Alcohol Use Identification Test (AUDIT; Saunders, Aasland, Babor, de la Fuente, & Grant, 1993) to determine the likelihood of meeting diagnostic criteria for PTSD and SUD. Individuals were eligible for participation if they scored greater than 43 on the PCL and greater than seven on the AUDIT. Eligible participants were then randomized to modified PE, modified PE + MET, or “healthy lifestyles.”

Assessments

Participants completed in-person assessments before treatment and in-person or telephone assessments (n = 6) at post-treatment. Participants were compensated for assessments. At the end of the pre-treatment assessment, participants were asked to provide a detailed description of their worst traumatic event including thoughts, emotions, sensations, and visual details. Responses were recorded on a research form that has been used in several previous studies (e.g., Coffey, Saladin, Drobes, Brady, Dansky, & Kilpatrick, 2002; Coffey, Stasiewicz, Hughes, & Brimo, 2006). The form is available from the last author upon the request.

Laboratory sessions

Participants completed two laboratory sessions: during the first week of starting residential SUD treatment (prior to initiating PTSD treatment) and approximately six weeks later (after completing PTSD treatment). Laboratory sessions were conducted between 1:00–3:00 p.m. to control for diurnal variations that may influence cue reactivity. At the beginning of each session, participants were given a urine drug screen and a breathalyzer to assess for recent drug use and alcohol intoxication. Participants were rescheduled if they tested positive for substances. Participants were fitted for their dental cotton rolls, and then completed the cue reactivity task (see Coffey et al., 2010) in a sound-attenuated experimental room. During this task, participants were presented with four counterbalanced imagery-in vivo cue combinations (trauma imagery cue + alcohol cue, trauma imagery cue + neutral cue, neutral imagery cue + alcohol cue, and neutral imagery cue + neutral cue). The trauma imagery cue was a 60-s description of participant’s worst traumatic event derived from the details provided during the pre-treatment assessment. The alcohol cue was each participant’s preferred alcoholic beverage; bottled water served as a neutral cue. The neutral imagery cue was a 60-s description of changing a light bulb. Participants were asked to close their eyes and listen to the trauma imagery cue. Following the imagery cue, participants were asked to open their eyes and were presented with the beverage cue. Participants were then instructed to continue to imagine the scene from the imagery cue for two minutes. The dental cotton rolls were then removed from the participants’ mouths and they were asked to complete their SUDS rating.

Statistical Analyses

To generate the three self-report RDoC domain scores, the first through sixth authors independently categorized each item from the ACQ-Now, IES-R, BDI, and BAI into one of four categories: Cognitive Systems, Negative Valence Systems, Arousal/Regulatory Systems, or no category (i.e., did not fit into any of the domains of interest). These four self-report measures had data available at all time points and collectively represent potential transdiagnostic factors related to but not exclusive of PTSD. It was determined that at minimum over half of the six authors (i.e., four authors) categorizing should agree on a given domain for a particular item in order for that item to be categorized into the domain. Thus, items with < 67% agreement rate were excluded from all domains (see Supplemental Materials for items in each domain).

To create domain-level composite scales, Likert scales were transformed so that all measures conformed to a 4-point scale with response options ranging from 0 to 3. Rescaled items were then computed into total sum and average scores for each RDoC domain of interest. Following the computation of domain scores, coefficient alphas were computed for each domain at pre- and post-treatment to assess internal consistency. Domains had different numbers of items (i.e., Negative Valence Systems had 30 items, Cognitive Systems had 29 items, Arousal/Regulatory Systems had 26 items), therefore, average scores were used in analyses in order to make the domain scores comparable (average scores for all domains ranged from 0–3) at each time point. Dependent samples t tests were employed to examine change within each domain between pre- and post-assessment as well as to examine change across domains at each time point.

A series of hierarchical linear regression analyses was conducted to assess the relationship of each self-report domain score to post-treatment cue reactivity. Dependent variables included post-treatment SUDS and salivation levels during the trauma and alcohol cue condition. In the first step, the corresponding pre-treatment domain (i.e., Cognitive Systems, Negative Valence Systems, and Arousal/Regulatory Systems) and pre-treatment variable of interest (i.e., SUDS or salivation level) were entered to control for pre-treatment values. In the second step, the corresponding post-treatment self-report domain score was entered to assess the unique predictive power of post-treatment domain score while controlling for pre-treatment values.

Results

Categorization into Domains

A total of 85 items were categorized into one of the three domains, leaving 26 items uncategorized (see Appendix A for domains and items). As seen in Table 1, reliability estimates were high for all three self-report domain scores across the study period. Regarding the validity of the self-report domain scores, domains at pre-treatment assessment were significantly correlated with CAPS total score at pre-treatment (Cog. r = .29, p = .008; Neg. Val. r = .50, p < .001; Arousal r = .47, p < .001), and domains at post-treatment were correlated to post-treatment CAPS (Cog. r = .51, p < .001; Neg. Val. r = .65, p < .001; Arousal r = .64, p < .001). Furthermore, both pre-treatment Negative Valence and Arousal/Regulatory domain scores correlated with CAPS total score at post-treatment (Cog. r = .13, p = ns; Neg. Val. r = .32, p = .011; Arousal r = .32, p = .010), providing evidence for the predictive validity of the employed RDoC measurement method.

Table 1.

Means and Standard Deviations of Domain Total and Average Scores

		Pre-Treatment		Post-Treatment		Comparison
Domain		M(SD)	α	M(SD)	α	t(df)	p
Cognitive	Total	38.47(12.29)	.89	16.39(8.99)	.83	15.49(61)	<.001
	Average	1.38(0.47)a		0.57(0.31)a
Negative Valence	Total	47.54(14.01)	.89	18.03(12.73)	.91	15.65(60)	<.001
	Average	1.60(0.48)b		0.57(0.31)a
Arousal/Regulatory	Total	25.18(13.34)	.92	8.77(9.93)	.95	11.31(61)	<.001
	Average	1.29(0.64)a, c		0.44(.50)b

Note. Means and Standard deviations of both Total and Average domain scores. For Total domain scores, internal consistency estimates (Cronbach’s alpha) and comparisons of domains scores pre and post treatment (t-test) are also presented. For Average domain scores, within each time point (column) the domain means that do not share a common letter subscript are significantly different from each other at that time point, p < .05. For instance, at Pre-Treatment Cognitive is significantly different than Negative Valence but not Arousal/Regulatory.

Change Over Time and Between Domains

All three self-report domain scores significantly decreased from pre- to post-treatment (Table 1). Subsequent pairwise comparisons at each time point revealed that at pre-treatment, Arousal/Regulatory Systems and Cognitive Systems were both significantly lower than Negative Valence Systems. At post-treatment, Arousal/Regulatory Systems was significantly lower than both Cognitive and Negative Valence Systems, which were not different from each other.

Relationship to Post-Treatment Cue Reactivity

All three post-treatment self-report domain scores were significantly associated with post-treatment SUDS ratings while controlling for pre-treatment values (Table 2). Similarly, Cognitive and Negative Valence System domains were significantly positively associated with post-treatment salivation levels while controlling for pre-treatment values (Table 3).

Table 2.

Hierarchical Linear Regression Analyses Related to Post-Treatment Subjective Distress Ratings

	B	SE B	β	ΔR2
Model 1 (Cognitive Systems)
Step 1				.15
Pre-Treatment Cognitive	14.97+	7.96	0.25
Pre-Treatment SUDS	0.32+	0.17	0.26
Step 2				.21
Post-Treatment Cognitive	49.66**	12.30	.58
Model 2 (Negative Valence Systems)
Step 1				.12
Pre-Treatment Negative	11.13	7.25	0.21
Pre-Treatment SUDS	0.32+	0.17	0.26
Step 2				.30
Post-Treatment Negative	39.69**	8.01	0.63
Model 3 (Arousal/Reg. Systems)
Step 1				.15
Pre-Treatment Arousal	9.51+	5.29	0.24
Pre-Treatment SUDS	0.31+	0.17	0.25
Step 2				.11
Post-Treatment Arousal	22.25*	8.20	0.41

Note. Domains are expressed as average scores ranging from 0–3. SUDS = Subjective Units of Distress Scale. Arousal/Reg. = Arousal/Regulatory.

⁺p < .10,

^*p < .01,

^**p < .001.

Table 3.

Hierarchical Linear Regression Analyses Related to Post-Treatment Salivation Levels

	B	SE B	β	ΔR2
Model 1 (Cognitive Systems)
Step 1				.16
Pre-Treatment Cognitive	0.35+	0.20	0.23
Pre-Treatment Salivation	0.16*	0.07	0.29
Step 2				.12
Post-Treatment Cognitive	0.98**	0.34	0.45
Model 2 (Negative Valence Systems)
Step 1				.12
Pre-Treatment Negative	0.15	0.19	0.10
Pre-Treatment Salivation	0.16*	0.07	0.30
Step 2				.13
Post-Treatment Negative	0.67**	0.23	0.41
Model 3 (Arousal/Reg. Systems)
Step 1				.32
Pre-Treatment Arousal	−0.04	0.09	−0.06
Pre-Treatment Salivation	0.28***	0.06	0.56
Step 2				.00
Post-Treatment Arousal	0.06	0.16	0.05

Note. Domains are expressed as average scores ranging from 0–3. Arousal/Reg. = Arousal/Regulatory.

⁺p < .10,

^*p < .05,

^**p < .01,

^***p < .001.

Discussion

RDoC represents a research framework that integrates research data and uses this knowledge to inform a novel classification system and targeted treatment development (Cuthbert & Insel, 2013). This study aimed to better understand the PTSD-SUD comorbidity from an RDoC framework and whether the impact of existing evidence-based treatments can be understood through RDoC domains of functioning. The study is unique in its examination of the association between concurrent PTSD-SUD treatment and changes in domains of functioning consistent with RDoC (i.e., Cognitive Systems, Negative Valence Systems, and Arousal/Regulatory Systems). In doing so, this study goes beyond the existing literature by showing a generalized symptom reduction after treatment in comorbid PTSD-SUD samples (Coffey et al., 2016) and by exploring whether these self-report domain scores can explain trauma-elicited distress and craving.

Items were first categorized from existing and validated self-report measures into three domains of functioning (Cognitive, Negative Valence, and Arousal/Regulatory Systems), serving as a proxy of RDoC domains. Results demonstrated strong internal consistency for all domains across the study period along with concurrent and predictive validity of the CAPS, suggesting that this preliminary first step towards validated assessment tools for the RDoC domains resulted in domains that represent shared constructs. Although not without its limitations, this approach demonstrated how existing measures may be used to apply the RDoC framework to treatment research.

Consistent with hypotheses, the domains of interest decreased in severity over the course of treatment. These findings go beyond those of the parent study (Coffey et al., 2016), which demonstrated generalized symptom reduction within specific “diagnostic categories” (e.g., substance use, depression, anxiety) by investigating treatment-related changes in domains of functioning. This approach moves away from traditional diagnostic categories and allows for the assessment of shared mechanisms that may explain the PTSD-SUD comorbidity. Indeed, RDoC aims to elucidate underlying mechanisms of psychopathology to overcome the limitations of diagnostic categories and inform treatment selection and development. Furthermore, when controlling for pre-treatment values, all three self-report domain scores were positively associated with post-treatment subjective distress, and all but one domain were positively associated with post-treatment salivation levels, such that smaller domain scores were linked to less subjective distress and alcohol cravings. Although in this study subjective distress and salivation are considered different units of analysis, these could also be categorized in the context of domains of functioning. For instance, subjective distress is measured as a response to potentially aversive trauma-relevant stimuli and thus closely fit the definition of Negative Valence Systems. Salivation is stimulated by both the sympathetic and parasympathetic nervous system, and thus could be typically considered an autonomic response fitting of the Arousal/Regulatory Systems. Nonetheless, in the context of cue reactivity it may be considered a conditioned proxy for craving and thus more representative of negative affect rather than a physiological domain.

Taken together, these findings speak to the potential efficacy of concurrent PTSD-SUD treatment in abating symptoms beyond those exclusively related to trauma. More importantly, the results suggest potential advantages in developing and using measures of transdiagnostic factors consistent with RDoC, particularly when using different units of analysis, to inform our understanding of the effects of existing evidence-based treatments. The cue-reactivity task has been shown to be a worthwhile means of investigating the relationship between PTSD and SUD (e.g., Coffey et al., 2010); therefore, identifying how these domains relate to changes observed in reactivity to trauma cue elicitation can be a promising avenue for capitalizing on a different unit of analysis and elucidating the mechanism of change in concurrent PTSD-SUD treatment. To understand why and how PE may improve functioning in general, it is important to consider the theorized mechanisms underlying PE and the etiology of these comorbidities; RDoC provides a promising framework for these tasks.

Severity of the Cognitive Systems domain score significantly decreased after treatment and was associated with a decrease in post-treatment subjective distress and salivation. These findings support the notions that perceptions of uncontrollability and unpredictability are core elements of anxiety and are implicated in the comorbidity between anxiety and PTSD (Zoellner et al., 2014). Consistent with RDoC’s Cognitive Systems domain, prior research has also suggested that PE alters thought patterns by disconfirming dysfunctional trauma-related cognitions through exposure exercises (Foa, Huppert, & Cahill, 2006). In the present study, similar to our cue reactivity findings, changes in the Cognitive System were associated with a beneficial reduction in emotional distress as subjectively reported by participants when they were faced with trauma cues. These findings provide evidence for important and complex connections between cognitive and affective systems, both of which appear to be activated in concurrent PTSD-SUD treatment. Although PE may not seem to address cognitive restructuring to the same degree as other cognitive therapies, research has suggested that PE alone is linked to changes in cognitions about self, world, and self-directed blame (Foa & Rauch, 2004). In addition, dysfunctional cognitions have been strongly linked to disorders that commonly co-occur with PTSD (e.g., as depression, anxiety, and substance use) providing evidence for the examination of cognitive systems as a transdiagnostic mechanism (Lockwood & Forbes, 2014). The present results suggest that cognitive domains of functioning are an important mechanism of change in concurrent PTSD-SUD treatment, evincing an impact in both self-report and physiological measures.

Symptoms in the Negative Valence Systems domain scale, with the lowest pre-treatment severity, also decreased over treatment and were associated with reductions in subjective distress and alcohol craving at post-treatment. Negative affect, thought to reflect feelings of distress and dissatisfaction, has been implicated as a shared vulnerability factor in PTSD comorbidity due to its strong genetic influences (Zoellner et al., 2014). Thus, it seems logical that this domain was associated with lower post-treatment distress scores. Furthermore, trauma-related negative affect has been found to induce craving in SUD with exposure to substance cues increasing negative affect (Coffey et al., 2010). This may explain why reduction in Negative Valence Systems was associated with decreased salivation and distress reactivity to trauma and alcohol cues. Within the context of PE, negative affect is believed to be reduced by the exposure-driven extinction to fear response, which in turn should lead to a reduction in cravings. Together this makes a strong argument for the important role of negative affect in explaining concurrent PTSD-SUD treatment outcomes.

Finally, decreases in the Arousal/Regulatory self-report domain scores, which was the lowest in severity at post-treatment, point to the important role that physiological symptoms may play in individuals with comorbid PTSD-SUD. Previous research has found decreases in physiological arousal (e.g., heart rate) after exposure-based treatments which may be driven by extinction learning and the gradual reduction of fear/distress over repeated exposure trials (Foa et al., 2006). This is consistent with the association between distress and the Arousal/Regulatory domain found in the present study. Interestingly and highlighting the importance of considering diverse units of analysis, while the Arousal/Regulatory domain decreased over treatment and was associated with trauma-elicited subjective distress, it was the only domain that failed to predict salivation during the post-treatment cue reactivity task. This is unexpected given that the autonomic nervous system regulates salivation and thus may be better represented by this domain. Nonetheless, if salivation is considered a proxy for craving in the cue reactivity task perhaps level of negative emotions or even negative cognitions may have a greater influence on craving than the other processes regulated by the autonomic nervous system. Indeed, Negative Valence and Cognitive Systems were associated with post-treatment salivation.

This study had several limitations that warrant consideration. First, we categorized items from existing measures to create the RDoC domains of interest, deviating from these measures’ original intended use. To our knowledge, there are currently no validated measures of the RDoC domains; rather, RDoC directs researchers to already-validated measures that may tap single constructs within each domain. By using different existing measures, our categorization system may have introduced measurement error; however, the constructed scales used in this study demonstrated acceptable psychometric properties. Furthermore, we were limited by the available data from the parent study (Coffey et al., 2016), which did not include some relevant measures for assessing PTSD and SUD (e.g., the PCL was administered prior to consent and data were not available to analyze; data were not available for the CAPS and C-DIS IV beyond the pre-treatment assessment). As such, these measures could not be included in the domains or used for further validation of the created measures. Second, although we selected these domains to be consistent with CBT and current evidence-based treatments, as well as representative of the shared symptomatology between PTSD and SUD, only three of the five RDoC domains of functioning were examined in the current study. Thus, it is unclear how the remaining two domains (i.e., Positive Valence, Social Processes) apply to PTSD-SUD and its treatment. Additionally, the study examined the RDoC domains of interest only in the context of PTSD-SUD treatment. This represents a unique comorbid population, yet the degree to which the changes observed in the current study are the result of PE, SUD treatment, or a combination of both is unknown. Nonetheless, a recent study using the same dataset found that PE had a superior effect relative to control group and a positive impact on psychological symptoms beyond PTSD (e.g., depression, anxiety, substance use), so it is likely that PE was at least partially responsible for the transdiagnostic changes observed in the current study. Despite these limitations, this study contributed to RDoC’s short-term goals of developing measurement tools to assess its identified domains. This represents a first attempt at assessing these in a clinically comorbid sample to inform future assessment approaches within the RDoC framework.

With these strengths and limitations in mind, there are a number of future directions for this research. It will be important to develop valid and reliable measures to assess the RDoC domains. As outlined in the RDoC matrix, these measurement tools should be comprehensive and include multiple units of analysis. Thus future attempts to develop measures may consider including well established measures of PTSD symptoms (e.g., PCL and CAPS). Furthermore, although the current study was restricted by sample size to run a confirmatory factor analysis on the created measures, such analysis can greatly improve the confidence in the developed measures. It is important to expand on this study by investigating change in all five RDoC domains, allowing investigators to test if certain RDoC domains of functioning are associated with greater change in treatment. This will, in turn, inform existing theories of the mechanisms of change in therapy and the development of specialized and tailored treatments targeting specific domains. Future research should include individuals with comorbid PTSD-SUD, but also should extend to other populations and treatments to clarify the generalizability of the findings to other mental health problems and interventions.

The present study uniquely investigated the PTSD-SUD comorbidity from an RDoC framework in relation to evidence-based treatments. In this way, beyond showing symptom reduction in treatment, the study served as a preliminary step to guide future research to merge the RDoC approach into the treatment literature. Our findings support the RDoC framework as a useful tool for understanding the transdiagnostic effects of existing treatments. Continued research has the potential to identify domains of functioning that serve as mechanisms of change in treatment.