The Development and Initial Validation of the Trauma-Related Alcohol Use Coping Measure (TRAC)

Abstract

Objective:

Posttraumatic stress disorder (PTSD) symptoms and alcohol use commonly co-occur and present a prevalent clinical comorbidity. The self-medication/coping model has been applied most consistently to understand the PTSD-alcohol use association. However, there is a relative paucity of self-report measures designed to assess motivations for alcohol use specifically for coping with PTSD symptoms. The goals of the current study were to develop and validate a measure that assesses the use of alcohol to cope with specific facets of PTSD symptomatology across two independent samples.

Methods:

Two samples were evaluated: a university-based sample (N=617; 77.0% women; M_age=22.3; SD=5.20) comprised of racially diverse trauma-exposed students and a nationally representative sample (N=510; 52.5% women; M_age=39.5; SD=10.9) of trauma-exposed adults who endorsed PTSD symptoms and past-year hazardous drinking. Both samples completed identical online questionnaire batteries. A Trauma-Related Alcohol Use Coping (TRAC) measure was developed and validated across both samples.

Results:

Confirmatory factor analysis was used to support the latent, hierarchical structure of the TRAC measure (total score; coping with intrusion, avoidance, negative alterations in cognitions and mood, and arousal/reactivity symptoms) and supported an 18-item version of the TRAC measure (university-based sample [N=617]: RMSEA=0.047, 90% CI [.04, .05]; SRMR=0.043; CFI=0.95; TLI=0.95; nationally representative sample [N=510]: RMSEA=0.045, 90% CI [.04, .05]; SRMR=0.021; CFI=0.98; TLI=0.97). The TRAC measure demonstrated excellent internal consistency, convergent and discriminant validity with well-established measures of mental health, known-groups validity, and incremental validity relative to non-PTSD coping-motivated drinking.

Conclusions:

Overall, the TRAC measure can be used to assess the extent to which alcohol use is related to coping with PTSD symptoms.

Introduction

Alcohol use frequently co-occurs with posttraumatic stress disorder (PTSD) symptoms and presents a complex comorbidity marked by a potentially chronic clinical course (Chiu et al., 2011; Debell et al., 2014; McCauley et al., 2012; Walter et al., 2018). Adults with PTSD symptomatology are also more likely engage in alcohol use and hazardous drinking (i.e., a pattern of alcohol use that increases risk for adverse health consequences; Saunders et al., 1993) compared to those without PTSD symptomatology (Kachadourian et al., 2014). Notably, the comorbidity between PTSD and alcohol use is associated with greater disability (Greene et al., 2016; Rehm et al., 2009) and poorer health (Mills et al., 2011; Turner et al., 2020) as compared to either condition alone (Berenz & Coffey, 2012; McCauley et al., 2012; Simpson et al., 2019). Moreover, hazardous drinking can adversely impact PTSD treatment by increasing dropout and reducing the rate of recovery (Berenz & Coffey, 2012; Brady et al., 2004; Zandberg et al., 2016).

Research exploring the PTSD-alcohol use co-occurrence highlight the bidirectional and transactional associations between PTSD symptomatology and the maintenance and/or exacerbation of alcohol use over time through the use of alcohol use to cope with negative emotional states (Brady et al., 2004; Hawn, Cusack, et al., 2020). For instance, adults with PTSD symptoms may use alcohol to cope (i.e., self-medicate; Khantzian, 1985) with the symptoms of PTSD (Haller & Chassin, 2014; Luciano et al., 2022). Alcohol craving has also been found to increase during periods of time when PTSD symptoms are elevated (Possemato et al., 2015; Simons et al., 2018; Simpson et al., 2014). Additionally, individuals with PTSD who engage in hazardous drinking evince increases in alcohol-related craving responses when presented with personalized trauma cues (Coffey et al., 2002; Coffey et al., 2010; Coffey et al., 2006). However, the extant literature focused on drinking to cope with PTSD is limited given the lack of measurement of PTSD-related coping motives.

Various self-report measures assess drinking patterns and related consequences of drinking (e.g., Alcohol Use Disorder Identification Test [AUDIT]; Saunders et al., 1993), alcohol-related expectancies (e.g., Alcohol Outcome Expectancies; Leigh & Stacy, 1993), and motives for use (e.g., Drinking Motives Questionnaire [DMQ] and DMQ-Revised [DMQ-R]; Cooper, 1994). Although these measures are helpful in elucidating the severity of alcohol use and its impairment on overall health and wellbeing (Hawn, Cusack, et al., 2020), they are not typically tailored to specific disorders, such as PTSD, or specific symptom management motives, such as drinking alcohol to cope with PTSD symptoms. The DMQ and DMQ-R are among the most frequency used measures to assess PTSD/AUD self-medication or coping models (Hawn, Cusack, et al., 2020), and they focus on negative emotional states (e.g., worry, nervousness, depression) rather than PTSD-related symptoms. This measurement approach is limited because of the lack of specificity to target symptoms that define PTSD (i.e., intrusion, avoidance, negative alterations in cognitions and mood, and arousal/reactivity symptoms; Smith & Cottler, 2018; Walton et al., 2018). As such, there is an opportunity to enhance the precision of measurement of alcohol-related coping motives in the context of PTSD by developing a conceptually grounded measure that captures alcohol use in the context of PTSD coping motives.

There have been two measures developed that examine alcohol use within the context of trauma and PTSD symptomatology. First, Norman et al. (2008) developed and empirically validated a measure that assesses PTSD-related alcohol use expectancies (e.g., PTSD-Alcohol Expectancies Questionnaire [P-AEQ]), which asks respondents to consider their expectations for alcohol use (e.g., “After a few drinks, my past traumatic experiences would feel less real”) and how they may relate to their PTSD symptoms using a 5-point Likert scale (1=Strongly disagree to 5=Strongly agree). The P-AEQ assesses expectations for drinking, rather than motives for use, and does not assess the extent of alcohol use due to PTSD symptomatology. Second, Hawn, Bountress, et al. (2020) developed a four-item measure to assess trauma-related drinking to cope (TRD) by modifying the DMQ-R to inquire about current PTSD symptoms (e.g., “How often do you drink alcohol to cope with symptoms including repeated, disturbing, unwanted memories, dreams, or feelings about the stressful experience?”) across each of the four PTSD symptom clusters (i.e., one item per PTSD symptom cluster). Although the TRD assesses drinking to cope due to some PTSD symptoms, using the DMQ-R’s 5-point Likert scale (1=Almost never/Never to 5=Almost always/Always), the brief questionnaire was a modification of the DMQ-R and does not fully capture drinking to cope as they relate to coping with a variety of PTSD symptoms across the four DSM-5 PTSD symptom clusters.

Therefore, an empirically supported assessment measure to index coping-motivated drinking in relation to specific PTSD symptomatology, specifically, is needed. Due to the diversity of PTSD symptoms and their potential to interplay with drinking behavior in distinct ways, a novel measure that assesses PTSD-related coping motives has both clinical and research utility. For instance, distinguishing between drinking to cope with trauma-related sleep disturbances and trauma-related flashbacks may be relevant in clinical and research settings as it provides meaningful context for when trauma-specific coping-motivated alcohol use may occur (e.g., in the evenings, before bedtime). This may be particularly useful within a clinical context to identify and address alcohol-related avoidance behaviors (e.g., Possemato et al., 2015). Moreover, a measure that can examine the nuanced associations between coping-motivated drinking and specific PTSD symptoms (e.g., hyperarousal) could inform monitoring efforts to reduce dropout among patients in the early stages of PTSD treatment (Kehle-Forbes et al., 2016; Niles et al., 2018), and thus, go beyond expectancy approaches to further elucidate drinking behaviors within the context of PTSD.

The current study was designed to develop and validate a novel measure (i.e., Trauma-Related Alcohol Use Coping [TRAC]) that assesses the use of alcohol to cope with specific symptoms of PTSD across the four DSM-5 PTSD symptom clusters. The TRAC measure was developed across three phases and two independent samples. Phase I included the initial development of the measure, which incorporated expert review of established measures, expert consultation, and item creation and reduction. Phase II-Study 1 included measure validation using a university-based sample of trauma-exposed students (N=617) and a subsample of students with probable PTSD/AUD (n=129). Specifically, the best fitting model structure for the TRAC was evaluated and then the incremental and construct validity of the measure was examined, including its internal consistency and convergent and discriminant validity with well-established measures of mental health. Consistent with the DSM-5 conceptualization of PTSD symptomatology (American Psychiatric Association, 2013), it was hypothesized that a hierarchical four-factor model (i.e., a four-factor model that includes a single, higher-level factor and four lower-level factors) would provide the best fitting model. Phase II-Study 2 focused on examining the psychometric properties of the TRAC for replicability as well as further testing its construct validity using a nationally representative sample (N=510) of trauma-exposed adults who endorsed PTSD symptoms and a subsample who met criteria for co-occurring probable PTSD/AUD (n=356). Known-groups validity was also examined in a subsample of past-year hazardous drinking by comparing those who met probable criteria for PTSD (n=385) with those who did not (n=125).

Phase I: Initial development

Phase I focused on the initial development of the TRAC, including item creation and reduction. We evaluated a pool of 95 items for inclusion, which were derived from established measures of avoidance coping (DMQ; Kuntsche & Kuntsche, 2009), general coping strategies (COPE Inventory; Carver et al., 1989), alcohol-related problems drinking (AUDIT; Saunders et al., 1993), and PTSD symptomology (PCL-5; Blevins et al., 2015). Initially, the pool of items was meant to capture the various types of trauma-related distress (i.e., maladaptive cognitions, physical symptoms, social impairments) that would be theoretically relevant to coping-motivated drinking, including general trauma-related distress (e.g., “To feel less tense or anxious about past traumatic events”), emotion regulation, feelings of powerlessness, and interpersonal functioning (15 TRAC items), as well as specific DSM-5 PTSD symptoms across each symptom cluster (Cluster B: intrusion [10 TRAC items], Cluster C: avoidance [22 TRAC items], Cluster D: negative alterations in cognitions and mood [30 TRAC items], and Cluster E: arousal/reactivity symptoms [18 TRAC items]) and general trauma-related distress; 15 TRAC items).

Further evaluation, which included a review of language clarity and item face validity, resulted in an 85-item version that was then sent to and reviewed by three doctoral-level experts in PTSD/AUD treatment and/or research. These PTSD/AUD experts were selected given their expertise in both clinically treating and researching PTSD/AUD comorbidity. After thorough review and correspondence with these three external PTSD/AUD experts, additional modifications were implemented and items were tailored to specific PTSD-related symptomatology (e.g., “When I have memories of the [event]”), as described in the DSM-5 PTSD criteria (American Psychiatric Association, 2013). Specifically, each expert identified items that most strongly aligned with each symptom across all four PTSD symptom clusters and contained clear language that would explicitly connect current alcohol use to current symptoms of PTSD (i.e., the construct of trauma-related alcohol use coping). In tandem with expert review, pilot testing of the 95-item version among a sample of undergraduate students (N=211) was conducted and revealed high inter-item correlations across items, which informed additional item reduction and merging to address these redundancies. For instance, high inter-item correlations were found among PTSD Cluster D TRAC items and resulted in substantial reductions (e.g., removal of the item and/or merging items that were closely related) from 30 items to 8 items. When coupled with expert feedback and further review of item face validity, a 19-item measure trauma-related alcohol use coping (i.e., TRAC) was finalized. It is important to note that while the TRAC measure includes items that represent DSM-5 PTSD symptoms and their respective clusters, the measure was designed to reflect a more nuanced measure of reasons for alcohol use with careful attention to item wording, clarity, and question prompts.

Phase II

The 19-item TRAC was examined in two independent studies across two separate samples: a university-based sample (Study 1; N=617) comprised of trauma-exposed university students and a nationally representative sample (Study 2; N=510) of trauma-exposed adults who endorsed PTSD symptoms and past-year hazardous drinking. Specifically, Phase II-Study 1 focused on measure validation using the university-based sample and a subsample of students with probable PTSD/AUD (n=129) by evaluating its internal consistency, incremental validity, and convergent and discriminant validity with well-established measures of mental health (see Table 2). Phase II-Study 2 focused on measure validation among the nationally representative sample and a subsample with probable PTSD/AUD (n=356) on the TRAC as well as evaluating its known-groups validity by comparing participants who met probable criteria for PTSD (n=385) with those who did not (n=125). The best fitting model structure for the TRAC identified in Phase II-Study 1 was evaluated for replicability and construct validity using the nationally representative sample (see Table 3 and 4).

Table 2.

Descriptive statistics and bivariate correlations of convergent and discriminant variables.

Study 1 (N= 617)	1	2	3	4	5	6	7	8
1. Trauma-related alcohol use coping (TRAC-19)	--
2. PTSD symptom severity (PCL-5)	.53**	--
3. Alcohol use severity (AUDIT)	.59**	.30**	--
4. Drinking to cope (DMQ-R Coping)	.66**	.44**	.49**	--
5. Negative affect (PANAS)	.43**	.63**	.26**	.39**	--
6. Positive affect (PANAS)	−.11**	−.17**	−.10*	−.12**	−.10*	--
7. Self-efficacy in resisting drinking (DTCQ)	−.32**	−.18**	−.28**	−.28**	−.11**	.19**	--
8. Distress tolerance (DTS)	−.22**	−.38**	−.18**	−.26**	−.39**	.28**	.29**	--
Mean	13.26	27.70	7.05	11.42	23.46	28.51	55.13	45.85
Standard Deviation	15.88	19.05	5.30	5.40	8.19	8.59	21.42	13.23
Internal Consistency (α)	0.96	0.96	0.81	0.88	0.88	0.91	0.91	0.93
Study 2 (N= 510)	1	2	3	4	5
1. Trauma-related alcohol use coping (TRAC-19)	--
2. PTSD symptom severity (PCL-5)	.64**	--
3. Alcohol use severity (AUDIT)	.64**	.42**	--
4. Drinking to cope (DMQ-R Coping)	.61**	.39**	.43**	--
5. Life satisfaction (SWLS)	.75	.04	.14**	−.10*	--
Mean	38.73	46.90	19.02	6.72	18.45
Standard Deviation	22.85	19.88	9.68	1.84	8.57
Internal Consistency (α)	0.98	0.96	0.90	0.83	0.94

Note. N = 617. **p < .01; *p < .05. TRAC = Trauma-Related Alcohol use Coping; PCL-5 = PTSD Checklist for DSM-5 (Blevins et al., 2015); AUDIT = Alcohol Use Disorders Identification Test (Saunders et al., 1993); DMQ-R = Drinking Motives Questionnaire, Revised, Coping subscale (Cooper, 1994); PANAS = Positive and Negative Affect Schedule (Watson et al., 1988); DTCQ = Drug Taking Confidence Questionnaire (Sklar & Turner, 1999); DTS = Distress Tolerance Scale (Simons & Gaher, 2005); SWLS = Satisfaction with Life Scale (Diener et al., 1985). All measures demonstrated good to excellent internal consistency (Cronbach’s alpha).

Table 3.

Standardized factor loadings for the hierarchical four-factor TRAC.

Prompt: For each item, please state how much you have used alcohol to cope in the past month using this scale: Not at all (0); (1); Moderately (2); (3); Very much (4).	Study 1 (N = 617)		Study 2 (N=510)
Subscale/Item	19-item	18-item	19-item	18-item
Cluster B
1. When I remember or am reminded of the trauma/stressor.	0.871	0.871	0.894	0.893
2. When I feel like the trauma/stressor is happening all over again.	0.851	0.850	0.897	0.898
3. When I have physical reactions to a trauma/stressor reminder (e.g., heart racing or slowing, palms sweating, shaking or trembling).	0.830	0.831	0.889	0.888
4. When I have emotional reactions to a trauma/stressor reminder.	0.884	0.884	0.909	0.909
Cluster C
5. When I don’t want to think, feel, or remember the trauma/stressor.	0.855	0.856	0.870	0.868
6. When I am around people, activities, or places that remind me of the trauma/stressor.	0.757	0.756	0.854	0.857
Cluster D
7. When I feel stuck in upsetting emotions (e.g., shame, guilt, anger, fear) about the trauma/stressor.	0.832	0.838	0.881	0.880
8. When I blame myself for the trauma/stressor.	0.831	0.837	0.848	0.851
9. When I think I am a bad or damaged person.	0.820	0.830	0.863	0.865
10. When I think the world is dangerous or other people cannot be trusted.	0.763	0.760	0.845	0.845
11. When I think about how the trauma/stressor affected me.	0.878	--	0.887	--
12. When I think about what caused the trauma/stressor.	0.887	0.864	0.889	0.885
13. When I want to be intimate or close with someone, emotionally, physically, or sexually.	0.648	0.655	0.746	0.748
14. When I feel lonely or disconnected from others.	0.752	0.755	0.811	0.812
Cluster E
15. When I have trouble concentrating.	0.762	0.762	0.868	0.869
16. When I feel irritated or angry.	0.794	0.795	0.883	0.884
17. When I have trouble sleeping.	0.789	0.788	0.842	0.842
18. When I feel jumpy, on guard, or on edge.	0.805	0.804	0.895	0.894
19. When I want to feel a “rush” or a thrill.	0.522	0.522	0.730	0.731

Note. TRAC = Trauma-Related Alcohol use Coping. For more information on the structure and response scale for this measure, please see the Method section of Phase II-Study 1.

Table 4.

Goodness-of-fit statistics for TRAC confirmatory factor analyses in Study 1 and Study 2.

Model	χ 2	df	CFI	TLI	RMSEA [90% CI]	SRMR
Study 1: University Students (N=617)
One-factor (19-item)	523.538	135	.900	.887	.068 [.062, .075]	.050
Four-factor (19-item)	362.339	148	.948	.940	.048 [.042, .055]	.043
Four-factor (18-item)	308.236	131	.953	.946	.047 [.040, .054]	.043
Study 1: PTSD/AUD Subsample (n=129)
Four-factor (19-item)	246.024	148	.918	.905	.072 [.055, .087]	.060
Four-factor (18-item)	224.781	131	.914	.900	.074 [.058, .091]	.060
Study 2: National Sample (N=510)
Four-factor (19-item)	284.128	148	.978	.975	.042 [.035, .050]	.021
Four-factor (18-item)	263.928	131	.977	.974	.045 [.037, .052]	.021
Study 2: PTSD/AUD Subsample (n=356)
Four-factor (19-item)	275.149	148	.964	.959	.049 [.040, .058]	.030
Four-factor (18-item)	257.676	131	.962	.956	.052 [.043, .061]	.031

Notes. TRAC = Trauma-Related Alcohol use Coping. Study 1 was conducted among trauma-exposed university students (N = 617) and a subsample of those who met criteria for probable PTSD/AUD (n=129); Study 2 was conducted nationally and included trauma-exposed adults who endorsed PTSD symptoms and past-year hazardous drinking (N = 510), as well as a subsample who met criteria for probable PTSD/AUD (n=356). All four-factor models included a single, higher-level factor, as well as the four lower-level factors. All chi-square tests are statistically significant at p < .001. CFI = comparative fit index; TLI = Tucker-Lewis index; RMSEA = root mean square error of approximation; SRMR = standardized root mean square residual.

Phase II: Study 1

Methods

Participants

A summary of sample characteristics is presented in Table 1. Less than 5% of the data were missing across study variables and was handled via listwise deletion for descriptive and correlation analyses. Item-level missing data on the TRAC measure was handled using the maximum likelihood with robust standard errors (MLR) estimator in Mplus version 8.0 (Muthén & Muthén, 2015). Distributions for all study variables approximated normality (skewness < ∣2.25∣; George & Mallery, 2003). The sample was comprised of 617 trauma-exposed students attending a large southwestern university (77.0% women; M_age=22.3; SD=5.20). The sample was racially and ethnically diverse (see Table 1) and the average number of traumatic event types endorsed was 10.35 (SD=4.42). Approximately 39.1% of the sample met diagnostic criteria for probable PTSD per the recommended PCL-5 total score diagnostic cut-off (Bovin et al., 2016a); 40.8% met diagnostic criteria for probable AUD per the recommended AUDIT total score diagnostic cut-off (Babor et al., 1992); and 20.9% met diagnostic criteria for probable PTSD/AUD. The subsample was comprised of 129 trauma-exposed students (76.7% women; M_age=21.7; SD=3.51) who met criteria for co-occurring probable PTSD/AUD. The subsample reported an average of 11.41 (SD=4.13) traumatic event types.

Table 1.

Participant demographic characteristics.

Variable	Study 1 N (%)	Study 2 N (%)
Gender
Men	132 (21.4%)	241 (47.3%)
Women	475 (77.0%)	268 (52.2%)
Transgender	2 (0.3%)	0 (0%)
Non-binary	4 (0.6%)	0 (0%)
Other (e.g., agender, genderfluid, demi-, etc.)	4 (0.6%)	1 (0.2%)
Race
White	310 (50.5%)	401 (78.6%)
Black or African American	102 (16.6%)	74 (14.5%)
Native American/Alaska Native	19 (3.1%)	8 (1.6%)
Asian	137 (22.3%)	17 (3.3%)
Native Hawaiian or Other Pacific Islander	2 (0.3%)	0 (0%)
Mixed Race/Other	44 (7.2%)	10 (2.0%)
Ethnicity
Not Hispanic or Latino	380 (61.6%)	445 (87.3%)
Hispanic or Latino	237 (38.4%)	65 (12.7%)
Marital status
Single	--	153 (30.0%)
Married	--	216 (42.4%)
Divorced	--	47 (9.2%)
Living with partner	--	71 (13.9%)
Widowed	--	12 (2.4%)
Separated	--	10 (2.0%)
Educational Attainment
Less than 7 years of school	--	5 (1.0%)
Junior High School	--	3 (0.6%)
Partial completion of High School	--	9 (1.8%)
High School graduate	--	113 (22.2%)
Partial completion of college	--	127 (24.9%)
College graduate	--	168 (32.9%)
Graduate school	--	85 (16.7%)
Academic Year
Freshman (0-29 semester hours)	81 (13.1%)	--
Sophomore (30-59 semester hours)	101 (16.4%)	--
Junior (60-89 semester hours)	196 (31.8%)	--
Senior (90 or more semester hours)	228 (37.0%)	--
Post-Baccalaureate	9 (1.5%)	--
Graduate Student	2 (0.3%)	--
Probable Diagnosis
Probable PTSD (PCL-5 ≥ 33)	241 (39.1%)	385 (75.5%)
Probable AUD (AUDIT ≥ 8 for men, ≥ 7 for women)	252 (40.8%)	446 (87.5%)
Probable PTSD and AUD	129 (20.9%)	356 (69.8%)

Note. Study 1 (N = 617); Study 2 (N = 510). All variables assessed via demographics questionnaires unless otherwise stated. PCL-5 = PTSD Checklist for DSM-5 (Blevins et al., 2015). AUDIT = Alcohol Use Disorders Identification Test (Saunders et al., 1993).

Procedure

Participants were recruited from a large public university in the southwestern United States. Students were informed of the opportunity to participate in the study through flyers posted on campus and online. Virtual flyers were shared with course instructors who had the option to share their flyers with students. Students were also able to find the study through browsing an online research management system operated by university staff. The online research management system is only accessible to those currently enrolled as students, 18 years of age or older, and who report English proficiency. Eligibility criteria for the present study required participants to endorse lifetime trauma exposure and alcohol use. The study consisted of a survey administered via Qualtrics that was estimated to take between 1.5 and 2 hours to complete. Upon completion of the study, students were automatically awarded research credit through the online research management system. The study was conducted in accordance with the Declaration of Helsinki, approved by all relevant institutional review boards, and was not preregistered. Data and analysis code for the present study is available from the corresponding author upon request.

Measures

Demographics questionnaire

Participant sociodemographic information was collected via a brief, self-report questionnaire. A summary of sample characteristics is provided in Table 1.

Life Events Checklist for DSM-5 (LEC-5; Weathers et al., 2013)

The LEC-5 is a 17-item self-report questionnaire designed to screen for the lifetime experience of potentially traumatic events. The measure includes a list of 16 potentially traumatic events (e.g., natural disaster, physical/sexual assault, combat, etc.), including an option for “other” potentially traumatic events not listed, and asks participants to indicate whether the event “happened to me”, “witnessed it”, “learned about it”, “part of my job”, or “not sure”. Positive exposure to a certain type of traumatic event was coded if the participant selected that the event “happened to me”, “witnessed it”, or “part of my job.” Events were summed to create a ‘trauma load’ variable that indicated the number of traumatic event types experienced across the lifespan. Participant eligibility for this study was confirmed by a nonzero LEC-5 total score.

Alcohol Use Disorders Identification Test (AUDIT; Saunders et al., 1993)

The AUDIT is a 10-item self-report measure used to identify alcohol use problems. It has been repeatedly validated across populations and demonstrates reliability in detecting hazardous alcohol consumption, alcohol dependence, and alcohol-related problems (Bohn et al., 1995; Selin, 2003). A pre-screen question is included to identify individuals who endorse lifetime alcohol consumption. Scores range from 0 to 40, and the cut-off score to identify probably alcohol use disorder is 7 among women and 8 among men (Saunders et al., 1993). In the present study, the AUDIT was used to confirm eligibility regarding lifetime alcohol use, as well as demonstrate convergent validity with TRAC.

Trauma-Related Alcohol Use Coping (TRAC)

The TRAC is a novel, 19-item self-report measure created to assess trauma-specific coping-motivated alcohol use. Participants are asked to indicate on a 5-item Likert scale (0=Not at all to 4=Very much) how often in the past month they have used alcohol to cope with each PTSD-related symptom. Please see Table 3 for a list of measure items.

Posttraumatic Stress Disorders Checklist for DSM-5 (PCL-5; Blevins et al., 2015)

The PCL-5 is a 20-item self-report questionnaire used to evaluate past-month PTSD symptom severity, per DSM-5 PTSD criteria (American Psychiatric Association, 2013). Participants are asked to respond to the PCL-5 regarding the “worst” traumatic event endorsed on the LEC-5, by rating how much they have been bothered by each symptom in the past month, using a 5-point Likert scale (0=Not at all to 4=Extremely). Total scores range from 0 to 80, with a score of 33 or higher considered indicative of a probable PTSD diagnosis (Bovin et al., 2016b). In the current study, the PCL-5 was used to examine convergent validity with TRAC.

Drinking Motives Questionnaire, Revised (Cooper, 1994)

The DMQ-R is a 20-item self-report measure designed to assess reasons for alcohol use and includes a coping subscale (i.e., drinking to cope with negative emotions) The subscale is comprised of 5 items, and summed subscale scores range from 5 to 25. For the present study, we examined the coping subscale to assess convergent validity with TRAC.

Positive and Negative Affect Schedule (PANAS; Watson et al., 1988)

The PANAS is a 20-item self-report questionnaire used to assess a participant’s mood in the past week. Participants are asked to indicate the extent to which they have felt each emotion within the past week using a 5-point Likert scale (1=Very slightly or not at all to 4=extremely). The PANAS has two subscales to assess positive emotion and negative emotion, respectively. In the current study, the PANAS subscales were used to assess convergent and discriminant validity with TRAC.

Distress Tolerance Scale (DTS; Simons & Gaher, 2005)

The DTS is a 15-item self-report measure designed to assess participants’ perceived capacity to withstand negative affect states (i.e., distress tolerance; Leyro et al., 2010). Total scores range from 15 to 75, with higher scores indicative of higher distress tolerance (i.e., a greater perceived ability to tolerate distress). For the current study, the DTS was used to examine discriminant validity with TRAC.

Drug Taking Confidence Questionnaire (DTCQ-8; Sklar & Turner, 1999)

The DTCQ is an 8-item self-report measure derived from the 50-item DTCQ, each designed to assess an individual’s self-efficacy in resisting the urge to drink in certain situations or events (e.g., “If I had trouble sleeping,” “If I were angry at the way things had turned out,” “If I wanted to celebrate with a friend,” etc.). Participants are asked to indicate how confident they are (using a percentage) in their ability to “resist the urge to drink heavily” in each situation. In the present study, the DTCQ was used to explore discriminant validity with TRAC.

Data Analytic Plan

Data were first examined for missing data, multivariate outliers, and normality. Descriptive statistics were used to evaluate the sample of university students who endorsed trauma exposure and consuming alcohol (N=617) as well as the subsample of students (n=129) who met criteria for co-occurring probable PTSD (a PCL-5 total score cut-off score of 33 or greater; Bovin et al., 2016a) and probable AUD (an AUDIT total cut-off score of ≥ 8 for men and ≥ 7 for women; Babor et al., 1992). Bivariate correlations were conducted across all study variables for both the overall sample and subsample. Examination of inter-item correlations for the initial TRAC were conducted using IBM SPSS version 29.0 (IBM Corporation). Items that were highly correlated (>0.80; Campbell et al., 2013) were reviewed and then removed in order to reduce redundancy in the TRAC. Cronbach’s alpha was computed to assess the internal reliability of the TRAC in the overall sample and subsample and interpreted using established guidelines (George & Mallery, 2003).

Confirmatory factor analysis (CFA) was used to examine the hypothesized latent factor structure of the proposed 19-item TRAC in the overall sample and the subsample. CFA was conducted using Mplus and utilized full information MLR estimation across all models considering the continuous nature of the TRAC and to account for outlying cases in the data. This approach was preferred to an exploratory factor analysis approach given that the TRAC was designed to map onto the existing factor structure of PTSD (i.e., symptom clusters B through E; American Psychiatric Association, 2013). Several structures were evaluated to identify the best fitting model, including a unidimensional model and a hierarchical four-factor model that incorporated a single, higher-level factor and four lower-level symptom cluster factors. The hierarchical four-factor model was favored over other CFA models to assess multidimensionality as the general, higher-order factor accounts for the shared variance among all the observed items, while the specific, lower-order factors account for the unique variance of each subset of items (Morin et al., 2013). Model fit was assessed using several fit indices, including (1) chi-square goodness-of-fit statistics, (2) the root mean square error of approximation (RMSEA), (3) the standardized root mean square residual (SRMR), (4) the comparative fit index (CFI), and (5) the Tucker-Lewis Index (TLI). Fit statistics were assessed using Hu and Bentler (1999) criteria for good model fit across all models. Specifically, RMSEA and SRMR values that are <.08 indicate an acceptable fit and values that are <.06 indicate an excellent fit. Furthermore, CFI and TLI values that are ≥.90 indicate acceptable fit and values that are ≥.95 indicate excellent fit.

A series of zero-order correlations between study variables and the TRAC were conducted to demonstrate convergent and discriminant validity across the overall sample and the subsample. Variables to evaluate convergent validity included measures of alcohol use severity (AUDIT total score), PTSD symptom severity (PCL-5 total score), coping motives for alcohol use (DMQ-R coping subscale total score), and negative affectivity (PANAS-Negative subscale total score). Variables to evaluate discriminant validity included positive affectivity (PANAS-Positive subscale total score), distress tolerance (DTS total score), and alcohol-related self-efficacy (DTCQ total score). It was hypothesized that the TRAC total score would moderately (.3 ≤ r < .5) to strongly (.5 ≤ r < 1.0) correlate with alcohol use severity, PTSD symptom severity, coping motives for alcohol use, and negative affectivity, all of which are constructs that theoretically underlie trauma-specific coping-motivated alcohol use and should thus be related, yet distinct from TRAC. Moreover, it was hypothesized that the TRAC total score would weakly (i.e., r < .3) correlate with positive affectivity, distress tolerance, and alcohol-related self-efficacy. And while these constructs may be theoretically related to trauma-specific coping-motivated alcohol use (e.g., low distress tolerance or low alcohol-related self-efficacy may increase the urge to drink to cope), the degree to which they are related to TRAC should be weak.

Hierarchical regression analyses were also conducted across the overall sample and the subsample to examine the incremental validity of the TRAC (total score) on PTSD symptom severity (PCL-5 total score) and alcohol use severity (AUDIT total score) above and beyond the variance accounted for by the DMQ-R coping motives subscale. Specifically, the DMQ-R coping motives subscale was entered at Step 1 and then the TRAC was entered in Step 2. This approach therefore helps to test that potential observed effects at Step 2 of the hierarchical regression models are distinct from the variance accounted for at Step 1 and adheres to the following effect size ranges in order to facilitate interpretation: small (ΔR² < 0.02), medium (0.02 ≤ ΔR² < 0.13), and large (ΔR² ≥ 0.13) effect sizes (Cohen & Cohen, 1983)

Results

Item Evaluation

Examination of inter-item correlations for the 19-item TRAC revealed moderate to strong associations between items (r’s range: 0.33-0.85), with items 11 and 12 yielding a high correlation (i.e., >0.80). After an independent review between two doctoral-level experts and considering factors such as the extant literature on measure development (Campbell et al., 2013), item clarity, measure aims, and adequate representation of each PTSD symptom cluster, item 11 was removed in order to reduce redundancy in the TRAC. Accordingly, the modified 18-item TRAC was included for CFA and construct validity analyses; however, the initial 19-item measure was also examined to compare model fit.

Confirmatory Factor Analysis

See Table 3 for a summary of standardized factor loadings and Table 4 for a summary of model fit indices. The unidimensional model demonstrated poor fit. Although both RMSEA and SRMR values were below .08, indicating acceptable fit, TLI values were less than .90, which indicates poor fit. However, the hierarchical four-factor model (i.e., a four-factor model that included a single, higher-level factor, as well as four lower-level factors) demonstrated excellent fit. Both RMSEA and SRMR values were less than .06 and CFI and TLI values were greater than .90, all of which demonstrates excellent fit. Latent TRAC symptom cluster covariances were all significant and all items properly loaded onto their respective latent symptom clusters. Thus, the hierarchical four-factor model was favored over the unidimensional model. Additionally, CFAs were conducted with both the initial 19-item TRAC as well as the 18-item version to evaluate the best fitting model. Results indicated that the 18-item TRAC demonstrated improved model fit compared to the 19-item version (see Table 4). Finally, among the subsample of students with probable PTSD/AUD, both the initial 19-item and 18-item versions of TRAC demonstrated similar model fit when using the hierarchical four-factor model.

Construct Validity

A summary of zero-order, bivariate correlations for the overall sample and the internal consistency of each measure are presented in Table 2. The internal consistency of the TRAC in the overall sample was excellent (α=.96). Cronbach’s alpha scores for each TRAC symptom cluster ranged from fair to excellent: Cluster B’s α=.92, Cluster C’s α=.78, Cluster D’s α=.92, Cluster E’s α=.83. Regarding convergent validity, the TRAC total score was positively and statistically significantly correlated with alcohol use severity (r=.59, p<.001; 34.2% shared variance), PTSD symptom severity (r=.53, p<.001; 28% shared variance), coping motives for alcohol use (r=.66, p<.001; 43.6% shared variance), and negative affectivity (r=.43, p<.001; 18.3% shared variance). Regarding discriminant validity, the TRAC total score was negatively correlated with positive affectivity (r=−.11, p=.007; 1.2% shared variance), distress tolerance (r=−.22, p<.001; 4.9% shared variance), and alcohol-related self-efficacy (r=−.32, p<.001; 10.2% shared variance). The pattern of correlations, indices of internal consistency, and evidence of convergent and discriminant validity were replicated in the subsample.

Incremental Validity

For the overall sample, hierarchical regression results revealed that Step 1 accounted for 19.5% of variance in PTSD symptom severity (F[1, 614]=148.53, p<.001), with DMQ-R coping motives emerging as a statistically significant predictor (b=1.56, t=12.19, p<.001). Step 2 accounted for an additional 10.0% of unique variance (F[2, 613]=128.19, p<.001), with TRAC emerging as a statistically significant predictor (b=0.51, t=9.33, p<.001). Regarding alcohol use severity as the criterion variable, Step 1 accounted for 24.3% of variance (F[1, 614]=197.72, p<.001), with DMQ-R coping motives emerging as a statistically significant predictor (b=0.48, t=14.06, p<.001). Step 2 accounted for an additional 12.0% of unique variance in alcohol use severity (F[2, 613]=174.82, p<.001), with TRAC emerging as a statistically significant predictor (b=0.15, t=10.73, p<.001).

For the PTSD/AUD subsample, results revealed that Step 1 accounted for 11.7% of variance in PTSD symptom severity (F[1, 127]=16.78, p<.001), with DMQ-R coping motives emerging as a statistically significant predictor (b=0.72, t=4.09, p<.001). Step 2 accounted for an additional 1.9% of unique variance but this was not statistically significant (F[2, 126]=9.85, p=.103). Regarding alcohol use severity as the criterion variable, Step 1 accounted for 8.1% of variance (F[1, 127]=11.14, p<.001), with DMQ-R coping motives emerging as a statistically significant predictor (b=0.30, t=3.34, p=.001). Step 2 accounted for an additional 9.9% of unique variance in alcohol use severity (F[2, 126]=13.77, p<.001), with TRAC emerging as a statistically significant predictor (b=0.11, t=3.89, p<.001).

Phase II: Study 2

Methods

Participants

Please see Table 1 for a summary of participant characteristics. Less than 5% of the data was missing across study variables and was handled via listwise deletion for descriptive and correlation analyses. Missing data in the TRAC was handled using the MLR estimator in Mplus. Distributions for all study variables approximated normality (skewness < ∣2.25∣; George & Mallery, 2003). The full sample included 510 trauma-exposed adults who endorsed PTSD symptoms and past-year hazardous drinking (52.5% women; M_age=39.5; SD=10.9). This sample was divided into two discrete subgroups of adults who met probable criteria for PTSD (n=385; 52.7% women; M_age=38.7; SD=10.5) per the recommended diagnostic cut-off score for the PCL-5 (Blevins et al., 2015) and those who did not (n=125; 52.0% women; M_age=41.7; SD=12.0). Among those meeting criteria for probable PTSD, the average number of traumatic event types endorsed was 11.5 (SD=6.5), compared to 8.4 (SD=5.9) among those who did not meet criteria for probable PTSD. The subsample included 356 trauma-exposed adults (50.0% women; M_age=38.3; SD=10.3) who endorsed symptoms consistent with PTSD and AUD per the recommended diagnostic cut-off scores for the PCL-5 and AUDIT, respectively (Blevins et al., 2015; Saunders et al., 1993). Among the subsample, the average number of traumatic event types endorsed was 11.6 (SD=6.5).

Procedure

Participants were recruited nationally to complete a survey-based study via Qualtrics Panels. Eligibility was determined via initial screening in Qualtrics, which required participants to be (1) between the ages of 18-65, (2) have access to a computer or mobile device on which they could complete the survey, (3) endorse current PTSD symptoms as determined by a score of ≥ 3 on the PC-PTSD-5 (Primary Care PTSD Screen for DSM-5; Prins et al., 2016), and (4) endorse current hazardous drinking per a score of ≥ 3 for women and 4 ≥ for men on the AUDIT Consumption subscale (items #1-3; Saunders et al., 1993). Participants were considered ineligible if they were unable or unwilling to provide consent and complete the self-report survey, or if they lacked English proficiency. The survey duration was estimated to be approximately 30 minutes. The study was conducted in accordance with the Declaration of Helsinki, approved by all relevant institutional review boards, and was not preregistered. Data and analysis code for the present study is available from the corresponding author upon request.

Measures

Please see the Measures section of Phase II-Study 1 for a summary of the demographic questionnaire, LEC-5, AUDIT, and PCL-5. For the current phase, an 18-item version of the TRAC was used, omitting item 11 due to improved model fit. Phase II-Study 2 utilized a short form version of the DMQ-R (DMQ-R-SF; Kuntsche & Kuntsche, 2009), in which the coping subscale consisted of only 3 items and the summed subscale score ranged from 3 to 9.

Satisfaction With Life Scale (SWLS; Diener et al., 1985)

The SWLS is a 5-item self-report measure used to measure an individual’s thoughts regarding the extent to which they are content with their life and is not designed to be a direct measure of positive or negative affect. Participants are asked to indicate the degree to which they agree with each statement (e.g., “In most ways my life is close to my ideal”) on a 7-point Likert scale (1=strongly disagree to 7=strongly agree), with higher scores indicating greater life satisfaction. The SWLS was used to examine discriminant validity with TRAC in the present study.

Data Analytic Plan

Data was first examined for missing data, multivariate outliers, and normality. Descriptive statistics were used to evaluate the overall national sample of trauma-exposed adults who endorsed PTSD symptoms and past-year hazardous drinking (N=510) and bivariate correlations were conducted across all study variables. Internal reliability was assessed via Cronbach’s alpha and interpreted using established guidelines (George & Mallery, 2003). Both the 19-item and modified 18-item versions of the TRAC was used considering the inter-item evaluation analysis conducted in Phase II-Study 1. A CFA was used to examine the latent factor structure confirmed in the initial CFA in Phase II-Study 1 and utilized MLR estimation across all models. Model fit was assessed using the same model fit indices outlined in Phase II-Study 1, including (1) chi-square goodness-of-fit statistics, (2) RMSEA, (3) SRMR, (4) CFI, and (5) TLI. Fit statistics were assessed using Hu and Bentler (1999) criteria for good model fit.

Akin to Phase II-Study 1 and to corroborate findings, zero-order correlations between study variables and the TRAC were conducted to demonstrate convergent and discriminant validity. Variables to evaluate convergent validity included measures of alcohol use severity (AUDIT total score), PTSD symptom severity (PCL-5 total score), and coping motives for alcohol use (DMQ-R-SF Coping subscale total score). The life satisfaction (SWLS total score) was used to evaluate discriminant validity. It was hypothesized that the TRAC total score would strongly correlate with alcohol use severity, PTSD symptom severity, and coping motives for alcohol use, and would weakly correlate with life satisfaction. Similar to Phase II-Study 1, hierarchical regression analyses was conducted to examine the incremental validity of the TRAC (total score) on PTSD symptom severity (PCL-5 total score) and alcohol use severity (AUDIT total score), above and beyond the variance accounted for by the DMQ-R-SF coping motives subscale. Additionally, model fit was examined among a subsample of participants meeting criteria for probable PTSD/AUD. Unique from Phase II-Study 1, a t-test was conducted comparing TRAC scores among participants who met criteria for probable PTSD, compared to those who endorsed subthreshold PTSD symptoms in order to assess known-groups validity.

Results

Confirmatory Factor Analysis

See Table 3 for a summary of standardized factor loadings and Table 4 for a summary of model fit indices. The hierarchical four-factor model (i.e., a four-factor model that included a single, higher-level factor, as well as four lower-level factors) assessed in Phase II-Study 1 was re-examined among the independent, nationally representative sample. Like results in Phase II, this model demonstrated excellent fit. CFAs were conducted with both the initial 19-item TRAC as well as the 18-item version to evaluate the best fitting model and results indicated that the 18-item TRAC demonstrated comparable model fit compared to the 19-item version, among both the full sample and the subsample of participants meeting criteria for probable PTSD/AUD.

Construct Validity

Cronbach’s alpha scores for the overall measure as well as each TRAC symptom cluster ranged from good to excellent: full measure’s α=.98, Cluster B’s α=.94, Cluster C’s α=.85, Cluster D’s α=.94, Cluster E’s α=.92. Regarding convergent validity, the TRAC total score was positively and statistically significantly correlated with alcohol use severity (r=.64, p<.001; 41.0% shared variance), PTSD symptom severity (r=.64, p<.001; 41.0% shared variance), and coping motives for alcohol use (r=.61, p<.001; 37.2% shared variance). Regarding discriminant validity, the TRAC total score was not correlated with life satisfaction (r=.075, p<.10).

Incremental Validity

Hierarchical regression results revealed that Step 1 accounted for 14.8% of variance in PTSD symptom severity (F[1, 508]=88.49, p<.001), with DMQ-R-SF coping motives emerging as a statistically significant predictor (b=4.16, t=9.41, p<.001). Step 2 accounted for an additional 26.1% of unique variance (F[2, 507]=176.03, p<.001), with TRAC emerging as a statistically significant predictor (b=0.56, t=14.99, p<.001). Regarding alcohol use severity as the criterion variable, Step 1 accounted for 18.9% of variance (F[1, 508]=118.08, p<.001), with DMQ-R coping motives emerging as a statistically significant predictor (b=2.28, t=10.87, p<.001). Step 2 accounted for an additional 21.9% of unique variance in alcohol use severity (F[2, 507]=174.56, p<.001), with TRAC emerging as a statistically significant predictor (b=0.25, t=13.70, p<.001).

Known-Groups Validity

Comparing discrete groups of participants who endorsed criteria consistent with probable PTSD (n=385) to those who endorsed subthreshold PTSD symptoms (n=125), independent samples t-tests revealed equal variances across groups, and that participants with probable PTSD reported significantly higher mean scores on the TRAC total (t(508)=13.57, p<.001; mean difference [MD]=27.36) and subscales corresponding with PTSD clusters: TRAC-B (t(508)=12.78, p<.001; MD=5.95), TRAC-C (t(508)=12.46, p<.001; MD=2.95), TRAC-D (t(508)=13.39, p<.001; MD=11.64), and TRAC-E (t(508)=12.01, p<.001; MD=6.83).

Discussion

The current study aimed to develop and validate a novel measure of PTSD-related drinking to cope utilizing a multi-phase approach. Specifically, the TRAC was developed using expert review and consultation for item creation and reduction and was then empirically validated in two independent samples. Initial item evaluation led to the removal of one item in the TRAC and resulted in a modified, 18-item version of the measure, which was used in subsequent analyses.

The 18-item TRAC evinced excellent psychometric properties across both samples. Among the university-based sample of trauma-exposed students and a subsample of students with probable PTSD/AUD, the hierarchical four-factor model of the TRAC, reflective of the four DSM-5 PTSD symptom clusters, demonstrated excellent fit and internal consistency. Moreover, robust associations between variables of interest highlights the measure’s strong convergent and discriminant validity. Notably, the TRAC was moderately to strongly (r’s=.43-66) associated with alcohol use severity, PTSD symptom severity, coping motives for alcohol, and negative affectivity and weakly correlated (r’s=−.11−.32) with positive affectivity, distress tolerance, and alcohol-related self-efficacy. Similar results were found among the nationally representative sample of adults who met criteria for co-occurring probable PTSD and probable AUD regarding model fit, internal consistency, and construct validity (i.e., convergent and discriminant validity).

The TRAC also evinced incremental validity above and beyond variance explained by general coping-motivated drinking (i.e., DMQ-R and DMQ-R-SF coping motives subscale). Although TRAC did not significantly predict PTSD symptom severity relative to general coping-motivated drinking among the subsample of university students with probable PTSD/AUD, it served as an incremental predictor of PTSD symptom severity across all other samples and as an incremental predictor of alcohol use severity in all samples, including the nationally representative sample of adults with probable PTSD/AUD. Specifically, the TRAC accounted for an additional 10-26.1% of unique variance in PTSD symptom severity and an additional 9.9-21.9% of unique variance in alcohol use severity, in relation to the variance accounted for by general coping-motivated drinking. These findings underscore the potential utility of the TRAC in predicting the severity of PTSD and AUD symptomatology and provides an alternative to the use of broad measures of coping-motivated drinking; however, longitudinal research is needed to further support these results.

The current findings suggest that the TRAC provides an empirically viable approach to the assessment of drinking to cope with PTSD at both the overall and symptom cluster level. Regarding its clinical utility, the TRAC would allow clinicians to assess coping-oriented drinking in relation to specific PTSD symptoms as well as symptom clusters and promote measurement-based care. For example, a clinician may be interested in assessing the extent of alcohol use to cope with PTSD symptoms for a patient engaged in PTSD or AUD treatment and/or integrative PTSD/AUD treatment. An important future direction is to assess the TRAC’s test-retest reliability as it can potentially serve to capture increased drinking due to PTSD-related symptoms (i.e., hyperarousal) during treatment and create opportunities for adjunct care and/or additional tailoring of treatment and thus potentially reduce attrition. This may be particularly crucial in the early stages of PTSD treatment where there is a greater risk of dropout (Kehle-Forbes et al., 2016; Niles et al., 2018) Regarding research utility, the TRAC provides a more nuanced approach to disentangling PTSD/AUD associations. Specifically, the TRAC may allow researchers to directly identify which specific symptoms or clusters are associated with increased coping-oriented drinking and thus extend results from previous work to better understand the maintenance processes relevant to PTSD/AUD comorbidity and to refine and tailor evidence-based interventions accordingly (e.g., Smith & Cottler, 2018; Walton et al., 2018).

The present study has several limitations. First, although each sample was generally representative of the larger population by race and ethnicity, future research is needed to test measure invariance across individuals with different gender identities (e.g., Barboza et al., 2016; Reisner et al., 2016), as well as in larger samples of racial/ethnic groups. Second, cross-sectional data collection was sufficient to accomplish the aims of the current study; however, such a research approach offers a onetime snapshot of relations between the constructs of interest. Future research using longitudinal methodology would be a useful next research step to establish relations with TRAC over time with alcohol and PTSD symptoms in naturally occurring settings and during treatment. Third, further work is necessary to extend these findings utilizing clinician-administered assessments and biomarkers of alcohol consumption. This type of work would decrease method variance and permit a more robust evaluation of construct validity. Fourth, while the AUDIT can measure a participant’s typical (e.g., average) frequency of alcohol consumption, it is not an exact measure of drinking frequency, such as a Timeline Follow-Back. Thus, the specific, recent frequency and quantity of alcohol use by participants was not directly collected in the current study and underscores the need for future work to include more detailed and/or daily approaches to examining drinking behaviors. Lastly, it is necessary for extensions of this research to examine trauma-specific coping or attempts to regulate affect across different age groups (e.g., older and younger populations) and exploring sex differences within these groups.

Overall, the present study developed and tested a measure of alcohol use related to coping with PTSD symptoms. The TRAC measure demonstrated excellent internal consistency and convergent and discriminant validity with well-established measures of mental health. Moreover, it evinced incremental validity relative to general coping-motivated drinking across two, independent samples of adults, and demonstrated known-groups validity across two discrete groups of participants with and without probable PTSD.

Public Health Significance Statement:

This study highlights the development of a measure that assesses the use of alcohol to cope with specific facets of PTSD symptomatology. The measure evinced strong psychometric properties, including internal consistency, convergent and discriminant validity, known-groups validity, and incremental validity, and provides both research and clinical utility.