Mechanistic Role of Emotion Regulation in the PTSD and Alcohol Association

Abstract

Objective

Posttraumatic Stress Disorder (PTSD) has been linked to problematic alcohol use. Those with PTSD have greater difficulty regulating emotions, which may help to explain the PTSD-drinking co-occurrence. However, emotion regulation as a mediator of PTSD-alcohol associations has been limited. In the present study, we examined this association.

Method

College students (N = 466, M_age = 19.5, 53% female) were assessed for PTSD by structured clinical interview, and were categorized into three groups: those who had not experienced a significant trauma (No Trauma, n = 182), those who had experienced a significant trauma but did not have current PTSD symptoms (Trauma Only, n = 171), and those with significant trauma and with current PTSD (partial or full; PTSD, n = 113). Alcohol use over the past six months and emotion regulation were assessed via self-report (DDQ; Collins, Parks, & Marlatt, 1985; DERS; Gratz & Roemer, 2004).

Results

Findings revealed that those with trauma and clinically significant PTSD reported greater difficulty with emotion regulation than those who had not been exposed to trauma, which in turn significantly predicted alcohol use. This mediating effect was not found in those with trauma exposure alone, suggesting an important role for PTSD in this pathway. Findings also indicated that the ability to control emotion-based impulses is a particularly relevant mediating facet of emotion regulation.

Conclusion

These results implicate emotion regulation as a potential explanatory link between PTSD and alcohol use, lending further support to the inclusion of emotion regulation training in PTSD treatment.

Introduction

Posttraumatic stress disorder (PTSD) is a constellation of symptoms (i.e., intrusive thoughts and memories, avoidance of trauma cues, alterations in cognitions and emotions, and hyperarousal) that may occur after experiencing a traumatic event (APA, 2013). PTSD is associated with a variety of harmful outcomes, including problematic alcohol use (e.g., Ouimette, Read, Wade, & Tirone, 2010). Indeed, PTSD and problem drinking commonly co-occur, and those with this co-occurrence have more severe symptom presentations in both PTSD and alcohol domains (Blanco, Xu, Brady, Pérez-Fuentes, Okuda, & Wang, 2013; Stappenbeck, Bedard-Gilligan, Lee, & Kaysen, 2013). PTSD also has been found to be a risk factor for the later development not only of diagnosable alcohol use disorders (Kline, Weiner, Ciccone, Interian, Hill, & Losonczy, 2014; Nickerson et al., 2014) but for problematic alcohol use more broadly (Read, Colder, Merrill, Ouimette, White, & Swartout, 2012). Yet at present, little is known about the mechanisms that may explain this association (Stewart, 1996). This is an important gap in the existing literature, as the elucidation of how PTSD and problematic alcohol are linked will have implications for clinical intervention. Knowledge of mechanisms of association can inform treatment, and potentially prevent those with PTSD from developing problematic alcohol use behaviors.

Multiple theories have been forwarded to explain the relationship between PTSD and alcohol involvement (e.g., Stewart & Conrod, 2003). One popular explanatory model of this co-occurrence, the self-medication hypothesis, is based in principles of affect regulation and reinforcement (Khantzian, 1999). This model conceptualizes the association between PTSD and alcohol as a function of negative reinforcement. In this framework, alcohol use and misuse co-occur with PTSD because the individual with PTSD may drink to cope with negative affect. Mediators of this association have been posited; however, to date, tests of these mediated pathways have focused primarily on coping mechanisms. Both coping style (e.g., emotion-focused coping) and coping motives have been found to function in this capacity (e.g., Kaysen et al., 2007; Lehavot, Stappenbeck, Luterek, Kaysen, & Simpson, 2014; O’Hare & Sherrer, 2011; Staiger, Melville, Hides, Kambouropoulos, & Lubman, 2009; Yeater, Austin, Green, & Smith, 2010). However, affect regulation more broadly, as it applies to both positive and negative affect, is fundamental to theoretical models of alcohol consumption (Curtin & Lang, 2007; Stritzke, Patrick, & Lang, 1995). Emotion regulation abilities have been found to predict alcohol use in those being treated for alcohol dependency, even after controlling for pertinent variables such as dependence severity, comorbidity, negative affect, and cognitive abilities (Berking et al. 2011). Furthermore, research suggests that individuals with trauma and posttraumatic stress may drink to reduce negative affect, as well as to evoke or enhance positive affect (Beseler, Aharonovich, & Hasin, 2011; Grayson & Nolen-Hoeksema, 2005; Kaysen et al., 2013). Thus, the ability to regulate affect may be especially important in the PTSD-alcohol association. The examination of a more general emotion-relevant factor such as emotion regulation, may elucidate a way of responding to one’s emotions that may play a role in PTSD-alcohol associations and could help to clarify self-medication pathways.

Emotion regulation is related to both PTSD and alcohol use and has been suggested as a possible explanatory link for the PTSD-problematic alcohol use relationship. According to Gross (1998), emotion regulation is “the processes by which individuals influence which emotions they have, when they have them, and how they experience and express these emotions.” Several studies have found differences in emotion regulation based on PTSD diagnosis (McDermott, Tull, Gratz, Daughters, & Lejuez, 2009; Tull, Barrett, McMillan, & Roemer, 2007; Weiss, Tull, Anestis, & Gratz, 2013), and on PTSD symptom severity (Ehring & Quack, 2010). In considering facets of emotion regulation in particular, those with PTSD report significantly more difficulty in accepting emotions, engaging in goal-directed behavior when upset, controlling behavioral impulses when distressed, and use of emotion regulation strategies (Weiss et al., 2013). Emotion regulation has been shown to differentiate those who have experienced a trauma but did not develop PTSD compared to trauma-exposed individuals who develop the disorder (Weiss, Tull, Davis, Dehon, Fulton, & Gratz, 2012). This is consistent with research that suggests that it is not trauma exposure itself, but the development of PTSD, that leads to problematic substance use (Breslau, Davis, & Schiltz, 2003). Notably, those with comorbid PTSD and alcohol dependence have been found to have more difficulty in emotion expression than those with PTSD or alcohol dependence alone (Blanco et al., 2013).

As noted, the emotion regulation construct is multi-faceted (Rottenberg & Gross, 2007). Some facets may be especially relevant to alcohol use in those with trauma and posttraumatic distress. For example, a core feature of PTSD is avoidance of trauma cues and trauma-related negative affect (i.e., experiential avoidance). This avoidance of emotional experience thought to maintain PTSD symptomotology has been linked to myriad behavior problems, including substance use (Hayes, Wilson, Gifford, Follette, & Stosahl, 1996). This may be because individuals seeking to avoid emotional experiences may be drawn to alcohol for its real or perceived affect-regulation properties (e.g., expectancies of the effects of alcohol) (Brown, Goldman, & Christiansen, 1985; Cooper, Frone, Russell, & Mudar, 1995; Christiansen, Smith, Roehling, & Goldman, 1989). For example, non-acceptance of emotions is a facet of emotion regulation that may correspond to experiential avoidance in that it would indicate rejection of one’s emotional experience or self-directed negativity about feeling upset (Gratz & Roemer, 2004). Awareness is another facet of the emotion regulation construct that signifies attention and validation of a negative emotional experience (Gratz & Roemer, 2004). These facets may be implicated in the avoidance of trauma-related thoughts and feelings, serving as mediators of the PTSD-alcohol association.

The ability to control emotion-based impulses is another emotion regulation facet which likely is relevant to the PTSD-alcohol association. According to Koole (2009), emotion regulation strategies can serve a variety of functions (i.e., need-oriented, person-oriented, and goal-oriented). The need-oriented function of emotion regulation prioritizes the immediate alleviation of negative affect with the secondary goal of promoting positive affect at the cost of long term consequences. Thus, difficulties controlling behavioral impulses when experiencing PTSD-related negative affect may result in alcohol use and related consequences. Although impulsivity has been examined in its association to both PTSD and alcohol, only one study to our knowledge has tested it as a mediator of the PTSD-substance use association (Bornovalova, Ouimette, Crawford, & Levy, 2009). Though this study found evidence for impulsivity as a mediator in females, the PTSD construct in this study was not well explicated; it was unclear if these individuals were trauma-exposed or met criteria for PTSD.

In short, despite strong conceptual support for testing emotion regulation as a mediator of the PTSD-alcohol association, few studies have tested this this explicitly. Among these, some studies have found evidence supporting the mechanistic role of emotion regulation. Yet, there have been a number of limitations to this work. For example, studies have varied in the alcohol outcome variables examined (e.g., alcohol-related consequences or coping motives for drinking is being predicted rather than drinking behavior itself, or impulsive behaviors broadly, rather than alcohol use specifically; Tripp, McDevitt-Murphy, Avery, & Bracken, 2015; Vujanovic, Bonn-Miller, & Marlatt, 2011; Weiss, Tull, Viana, Anestis, & Gratz, 2012). Critically, studies have varied substantially in their measurement and conceptualization of emotion regulation. Although the mediating role of some specific aspects of emotion regulation have been tested (e.g., non-judgmental acceptance, see Vujanovic, Bonn-Miller, & Marlatt, 2011), in general, studies examining the full, multi-dimensional emotion regulation construct have been sparse and findings have been somewhat mixed. In one of only a handful of such studies, authors found that some facets of emotion regulation (i.e., difficulty with impulse control and goal-directed behavior when upset) mediated the association between PTSD symptoms and alcohol-related consequences in a sample of college students, but assessment of trauma and PTSD were self-report and analyses did not allow for comparisons between trauma exposure only and PTSD (Tripp et al., 2015). Findings from two other studies using military samples did not find emotion regulation to mediate the PTSD-alcohol association (Klemanski, Mennin, Borelli, Morrissey, & Aikins, 2012; Tripp & McDevitt-Murphy, 2015). According to authors of one study, this may have been due to restricted range of alcohol behaviors (Klemanski et al., 2012). The other study found evidence of mediation for the impulse control and emotional clarity facets only when examining solely men (Tripp & McDevitt-Murphy, 2015). Thus, more research examining multi-dimensional aspects of emotion regulation as a mediator of PTSD-alcohol associations is needed, as findings have been mixed and have been unable to parse out the effects of trauma and PTSD in considering these mediated pathways. Such research may inform which aspects of emotion regulation are most pertinent to target in skills training incorporated in the treatment of PTSD.

Current Study and Hypotheses

The current study tested the mediating role of emotion regulation in the PTSD-alcohol association by examining relations among PTSD status, emotion regulation, and alcohol use in a large sample of alcohol-using college students (N = 466). Based on prior literature, we expected that those with PTSD would report significantly more difficulty in emotion regulation in all domains and more alcohol use than those without PTSD, regardless of trauma exposure history. We hypothesized that emotion regulation and specific facets of emotion regulation related to experiential avoidance (i.e., awareness, non-acceptance, and ability to control emotion-based impulses) in particular, would mediate the association between PTSD status and alcohol use in that PTSD would be related to greater difficulty in emotion regulation which would also be associated with greater alcohol use.

Method

Participants

Sample

The current sample was drawn from a larger experimental study investigating implicit associations in trauma-exposed, alcohol-using college students (Bachrach, Wardell, Coffey, & Read, 2012; Read, Bachrach, Wardell, & Coffey, in preparation). Given the larger aims of the study, participants were selected for having PTSD symptomatology (see Procedure) and thus are not representative of rates of PTSD in the college population. Participants were recruited locally from introductory psychology courses and posted flyers. All participants were full-time students in a 4-year university program, ages 18 to 25 (M_age = 19.5, SD = 1.42). In order to participate, these students had to be English-speaking, have consumed alcohol at least once in the past three months, and report no colorblindness or hearing impairment. The sample (N = 466, 53% female, 72% Caucasian) was fairly evenly distributed across education level and included freshman (35%), sophomores (25%), juniors (22%), and seniors (17%).

Procedure

Data were collected over two sessions; the first session served to determine study eligibility based on PTSD status. In the first session, diagnostic clinical interviews were used to assess for trauma exposure and PTSD symptoms using the Life Events Checklist and the Clinician Administered PTSD Scale (CAPS; Blake, Weathers, Nagy, Kaloupek, Charney, & Keane, 1998). Participants were compensated for this interview via $20 cash or credit through the psychology research subject pool. Eligible participants (n = 490) based on PTSD status (i.e., those who had not experienced a significant trauma and did not endorse PTSD symptoms, those who had experienced a significant trauma but did not have current PTSD symptoms, and those with significant trauma and with current clinical or subthreshold PTSD) were invited to participate in the second portion of the study. These groups were created to allow for clear group comparisons in an effort to better understand the impact of trauma and PTSD. Those who did not fall into those three categories were deemed ineligible (n = 131). During the second session (approximately one week later), participants completed a computer task about alcohol beliefs and self-report questionnaires. Participants were able to choose compensation for this second portion of the study, either $30, or credit through the psychology research subject pool. Twenty-four participants (6 with no criterion A event, 13 with a criterion A event but no PTSD, and 5 with either clinical or sub threshold PTSD) who were eligible for the second session failed to attend and thus did not contribute data to the present study. All procedures were approved by the Institutional Review Board.

Measures

Trauma and PTSD

PTSD status was determined by clinical assessment interviews conducted by trained student clinicians using the Life Events Checklist and the Clinician Administered PTSD Scale (CAPS; Blake et al., 1998). The LEC is a list of 16 potentially traumatic events that may have been personally experienced, witnessed, or learned about happening to someone close. Participants were asked to report on PTSD symptoms over the past 30 days related to any DSM IV-TR criterion A events to assess for current PTSD (APA, 2000). If a criterion A event was not experienced, participants were asked to report on their most stressful or upsetting event. A number of methods have been used to assess PTSD symptoms with the CAPS diagnostic interview. In this study, we used a method that yields a conservative estimate of PTSD based on both the frequency and the intensity of symptom presentations (Blake et al., 1995). Each PTSD symptom was rated by the interviewer on both “frequency” and “intensity” dimensions of the 17 possible PTSD symptoms; only those that occurred at least 1–2 times per week in the past month (score of 2 or greater on CAPS symptom Frequency items) and were of moderate intensity (score of 2 or greater on CAPS symptom Intensity items) were counted as “present”. Using this scoring method, each participant received a symptom count consistent with DSM-IV-TR criteria.

Based on this interview data, participants were categorized into three groups: (No Trauma, n = 182) those who had not experienced a significant trauma, (Trauma Only, n = 171) those who had experienced a significant trauma but did not have current PTSD symptoms, and (PTSD, n = 113) those with significant trauma and with current PTSD (partial or full; n = 70, n = 43, respectively). Full PTSD criteria were based on DSM-IV-TR (i.e., reporting at least 1 B cluster symptoms, 3 C cluster symptoms, and 2 D cluster symptoms in response to a criterion A event); partial PTSD was defined as the endorsement of at least 1 B cluster symptom, 1 C cluster symptom, and 1 D cluster symptom in response to a criterion A event. Graduate interviewers met for weekly supervision. Twenty-seven percent of the CAPS interviews were randomly selected to be coded for inter-rater reliability, both within the site-based research group and by an outside Psychologist, which was high (Kappa = .85).

Emotion regulation

In the present study, we conceptualized this construct consistent with work by Gratz and Roemer (2004), and accordingly, used the Difficulties in Emotion Regulation Scale (DERS; Gratz & Roemer, 2004) as a measure of emotion regulation. The DERS is a 36-item measure assessing the multifaceted construct of emotion regulation. Items are rated on a five-point Likert Scale (1 = almost never; 5 = almost always), with higher scores indicating greater difficulties in emotion regulation. In addition to a total score that is obtained by summing the items, there are six subscales: 1) Nonacceptance of Emotional Responses (NONACCEPT, α = .90), 2) Difficulties Engaging in Goal-Directed Behavior (GOALS, α = .89), 3) Impulse Control Difficulties (IMPULSE, α = .88), 4) Lack of Emotional Awareness (AWARE, α = .84), 5) Limited Access to Emotion Regulation Strategies (STRATEGIES, α = .90), and 6) Lack of Emotional Clarity (CLARITY, α = .85). Full scale reliability for the current study α = .94 was consistent with reliability found by the measure developers (Gratz &Roemer, 2004; α = .93).

Alcohol use

To assess typical past six month alcohol consumption, participants reported the average number of standard drinks consumed on a typical Monday, Tuesday, Wednesday, etc. in the past 30 days. Respondents were given Standard Drink Conversion charts including a definition of a standard drink to enhance reporting accuracy. Summing the number of drinks across all seven days provided a measure of typical weekly alcohol quantity over the past six months. Summing the number of days on which they reported consuming any alcoholic beverages provided a measure of typical weekly alcohol frequency. This methodology follows that of the Daily Drinking Questionnaire (DDQ; Collins, Parks, & Marlatt, 1985), a measure designed to assess typical quantity, frequency, and volume of alcohol consumption.

Data Analytic Approach

Data Management

Descriptive information for the putative mediator (DERS) and outcome (alcohol use) variables were examined for skew and kurtosis, as well as a critical assessment of any outliers for possible exclusion from analyses or necessary adjustments. Although all variables were in acceptable range for skewness and kurtosis, we identified three outliers in alcohol quantity signified by large gaps in the histogram and a Z-score greater than 3.33. These outliers were addressed by adjusting the scores to remain at the extreme ends of the distribution but with a score that is more representative of the sample (Tabachnick and Fidell, 2000). Of the participants who participated in both sessions, no data were missing.

Analysis

Group differences in emotion regulation and drinking were assessed using one-way analysis of variance tests and follow up Tukey’s HSD (Honestly Significant Difference) tests. Due to the high correlation between alcohol quantity and frequency (r = .73), the mediation models were only tested with alcohol quantity. A mediational analysis was conducted to test the mediating role of emotion regulation using the DERS total score. A second mediation analysis tested the unique effects of proposed mediators (i.e., the six subscales of the DERS). The independent variable remained the same for each of the analyses (i.e., PTSD status). Mediation analyses were conducted based on procedures recommended by Hayes and Preacher (2013) and Hayes (2013) using the SPSS macro MEDIATE. This program was chosen based on its intended use for multicategorical independent variables and testing multiple mediators. Indirect effects (a₁b and a₂b) were tested using stratified bootstrapping with a sample size of 5,000 and bias corrected confidence intervals (95%). Indicator coding was used to make two comparisons: 1) those who had never experienced a criterion A trauma (No Trauma) to those who experienced a criterion A trauma but did not have PTSD (Trauma Only), 2) those who had never experienced a criterion A trauma (No Trauma) to those who met criteria for full or partial PTSD (PTSD). Thus, No Trauma served as the reference group in these analyses. These comparisons allowed us to evaluate group differences based on trauma exposure and symptom development. Gender was controlled for in this analysis due to gender differences found in the emotion regulation, alcohol use, and PTSD literatures. See Figure 1 for mediation model analyses.

Figure 1.

Results

Descriptive Statistics

PTSD

Of the possible 17 PTSD symptoms, the number endorsed ranged from 0 to 16 with an average of 1.91 (SD = 3.32) symptoms. Across the whole sample, the most commonly reported “worst or most upsetting” events according to the participants were the sudden unexpected death of someone close, life threatening illness or injury, transportation accident, and physical or sexual assault. Of those in the PTSD group, the majority reported more than one Criterion A trauma (78%). The most commonly reported trauma types were physical assault, sexual assault, assault with a weapon, and transportation accident. These are comparable to rates of exposure endorsed in other, non-college PTSD samples, including combat veterans (Breslau et al., 1998; Kaltman, Greene, Mete, Sharaet, & Miranda., 2010, Fikretoglu, Brunet, Schmitz, Guay, & Pedlar, 2006; Marshall-Berenz, Vujanovic, Bonn-Miller, Bernstein, & Zvolensky 2010). As such, the participants in this sample with PTSD are comparable to other PTSD samples, both with respect to trauma type and trauma severity.

Emotion regulation

DERS Total for the whole sample ranged from 36 to 157 (M = 81.77, SD = 22.60). See Table 1 for sample descriptive information on DERS subscales. One-way analysis of variance (ANOVA) revealed that each of the DERS subscales including NONACCEPT, F(2, 463) = 16.35, p < .01, GOALS, F(2, 463) = 7.96, p < .01, IMPULSE, F(2, 463) = 23.31, p < .01, AWARE F(2, 463) = 3.01, p = .05, STRATEGIES, F(2, 463) = 24.07, p < .01, CLARITY, F(2, 463) = 160.50, p < .01, as well as the DERS total score F(2, 463) = 25.45, p < .01, differed significantly based on PTSD status as hypothesized. Follow-up tests indicated that those in the Trauma Only group relative to the No Trauma group did not differ in emotion regulation; however, the PTSD group was significantly higher on every emotion regulation subscale and the DERS total score, reflecting greater difficulties in emotion regulation, than the No Trauma group. The PTSD group also reported significantly more difficulty with emotion regulation than the Trauma Only group on NONACCEPT, IMPULSE, STRATEGIES, and CLARITY subscales, as well as on the DERS total score.

Table 1.

Descriptive Statistics

	Overall Sample				PTSD Groups (Means & SD)
	Mean	SD	Min	Max	No Trauma	No PTSD	PTSD
PTSD Symptoms
	1.91	3.32	0	16	0.14 (0.41)	0.36 (0.61)	7.11 (2.99)
Emotion Regulation
DERS Total	81.77	22.60	36	157	76.12 (20.08)	79.68 (21.15)	94.02 (24.07)
NONACCEPT	12.76	5.66	6	30	11.86 (5.04)	12.04 (5.27)	15.33 (6.40)
GOALS	14.92	4.93	5	25	14.00 (4.75)	14.97 (4.80)	16.32 (5.11)
IMPULSE	11.22	4.73	6	29	10.26 (4.24)	10.58 (4.46)	13.73 (5.01)
AWARE	15.18	4.99	6	30	14.68 (4.85)	15.09 (4.85)	16.12 (5.33)
STRATEGIES	16.33	6.55	8	37	14.74 (5.96)	15.74 (6.15)	19.80 (6.83)
CLARITY	11.35	4.16	5	24	10.58 (3.73)	11.27 (4.07)	12.72 (4.62)
Alcohol Behaviors
Quantity	15.65	11.80	0	57	14.23 (10.71)	15.63 (11.78)	17.99 (13.15)
Frequency	3.11	1.59	0	7	2.84 (1.45)	3.06 (1.57)	3.62 (1.71)

Note. N = 466, No Trauma n = 182, No PTSD n = 171, PTSD n = 113

Alcohol use

On average, the sample drank alcohol on 3.11 days (SD = 1.59) and 15.65 drinks in a typical week over the past six months (SD = 11.80). ANOVA tests revealed that alcohol quantity F(2, 463) = 3.58, p = .03, and frequency F(2, 463) = 8.82, p < .01, significantly differed based on PTSD status. The PTSD group reported significantly more drinking days than both the No Trauma and Trauma Only group, and significantly more alcohol quantity than the No Trauma group. The Trauma Only group did not significantly differ in their alcohol quantity or frequency from the No Trauma group. See Table 1 for whole sample descriptive information and for PTSD group descriptive information. See Table 2 for variable correlations.

Table 2.

Correlations of DERS total and subscales and alcohol variables

	1	2	3	4	5	6	7	8
NONACCEPT	1
GOALS	.45**	1
IMPULSE	.58**	.55**	1
AWARE	.120**	.01	.11*	1
STRATEGIES	.66**	.63**	.72**	.16**	1
CLARITY	.49**	.33**	.47**	.52**	.51**	1
DERS TOTAL	.79**	.69**	.79**	.44**	.87**	.74**	1
ALCOHOL FREQUENCY	.07	.04	.20**	.05	.14**	.09*	.14**	1
ALCOHOL QUANTITY	.034	−.02	.14**	.08	.10*	.05	.09	.73**

Note.

^**= p<.01

^*= p<.05

Mediation Analyses

The initial mediation analysis was run with alcohol quantity as the outcome variable, examining the DERS total score as the mediator. Findings revealed that the PTSD group had a significant direct (c₁’) effect on alcohol quantity (b = 4.64, t = 3.24, p < .05) relative to the No Trauma group. This contrast also produced a significant indirect effect (a₁b) observed through emotion regulation (a₂b TOTAL = 0.78, 95% CI = 0.04 to 1.75) indicating that the PTSD group reported more difficulty with emotion regulation as a function of PTSD status, which was in turn associated with higher alcohol consumption. The second mediation analysis examined the six subscales of emotion regulation as mediators between PTSD status and alcohol use. Results showed a significant direct (b = 4.15, t = 2.90, p < .05) and indirect (a₂b IMPULSE = 1.45, 95% CI = 0.42 to 2.88) effect when comparing the PTSD group to the No Trauma group. That is, relative to the No Trauma group, those in the PTSD group reported more difficulty with controlling impulses when upset (as measured on the IMPULSE subscale) as a function of PTSD status, which in turn was associated with greater alcohol consumption. No other indirect effects were found. See Table 3 for detailed results of this analysis.

Table 3.

Mediation analyses for the PTSD status and alcohol quantity association controlling for gender

	a1	a2	b	a1b	a2b
TOTAL	3.34	16.7*	0.05	0.16	0.78*
NONACCEPT	0.13	3.23*	−0.15	−0.02	−0.47
GOALS	0.91	1.97*	−0.27	−0.24	−0.53
IMPULSE	0.28	3.30*	0.44*	0.12	1.45*
AWARE	0.43	1.53*	0.09	0.04	0.14
STRATEGIES	0.97	4.84*	0.14	0.13	0.66
CLARITY	0.63	1.82*	0.02	0.01	0.03

Note. a1 and a2 are the mean differences in emotion regulation domains between the No PTSD and PTSD groups, respectively, relative to No Trauma; b is the coefficient for each emotion regulation domain holding PTSD status constant; a1b and a2b are the relative indirect effects of PTSD status;

^*=p<.05.

Discussion

The aim of the current study was to test the mediating role of emotion regulation in the PTSD-alcohol association. By comparing across levels of trauma exposure and PTSD symptoms, we were able to isolate specific effects of PTSD above and beyond trauma exposure alone. This study also built on prior work by using an Emotion Regulation measure that tapped multiple facets of this construct, and by studying a sample reporting a broader range of drinkers than studies of treatment-seeking samples. Study findings also significantly contribute to our knowledge of the association between trauma exposure and PTSD and alcohol use.

Consistent with our hypotheses and with previous findings in the literature, those with PTSD reported more difficulty in emotion regulation and greater alcohol use than those without PTSD (Weiss et al., 2012; Weiss et al., 2013). Importantly, we found that emotion regulation mediated the relationship between PTSD status and alcohol use when comparing those in the PTSD group to the No Trauma group, but not when comparing those in the Trauma Only group to the No Trauma group. From this, we conclude that differences in the mediating role of emotion-regulation are PTSD specific as they were not found for those with trauma exposure alone. This is an important advancement on prior work, and suggests that the path to drinking is different for those with PTSD relative to those with only trauma exposure. This may in part be due to difficulty in regulating the challenging emotions that are a component of the PTSD construct (Miller, Greif, & Smith, 2003).

Findings also offered evidence that emotion regulation serves as a mediating mechanism in associations between PTSD and alcohol use. Of the facets of emotion regulation tested in our models, difficulty controlling impulse behavior when emotionally distressed (i.e., the IMPULSE subscale) mediated the pathway from PTSD to alcohol consumption. This result was consistent with our hypothesis and previous findings (Tripp et al., 2015), and may speak to Koole’s (2009) conceptualization of need-oriented emotion regulation, in which reducing negative affect is prioritized, even at the cost of negative consequences. In short, according to Koole’s model, a person's drinking may be driven by a perceived need for immediate relief from negative affect.

Hypotheses pertaining to the subscales thought to be more related to experiential avoidance (i.e., NONACCEPT and AWARE) were not supported. Though as predicted, those with PTSD relative to the No Trauma group reported more difficulty in these domains of emotion regulation, these subscales did not function as mediators in the PTSD-alcohol association,

Implications

Several studies have examined cognitively-based constructs such as drinking to cope as a mediator of the PTSD-substance use association, yet there has been far less examination of other potential mediators of this association. This is a significant contribution of this study and provides more information about emotion regulation deficits that may be targeted in intervention. Further, in this study, we assessed PTSD status using gold-standard interview measures (i.e., CAPS). This enabled us to examine unique effects of trauma and PTSD, and had a large sample with which to test pathways of interest.

These findings have implications for treatment. In particular, the finding that difficulty controlling behavioral impulses when upset (IMPULSE) mediated the relationship between PTSD status and alcohol use for those with PTSD relative to the No Trauma group is support for the self-medication hypothesis. That is, those with PTSD exhibit greater difficulty in emotion regulation and thus, may drink to cope with negative affect. In this sense, alcohol use can be thought of as an emotion regulation process or strategy (Aldao & Dixon-Gordon, 2014). Thus, although this research shows that those with PTSD have emotion regulation deficits broadly, difficulty in controlling impulses when upset specifically mediates the association between PTSD and alcohol use. Treatment interventions focusing on either controlling impulses when upset or increasing one’s ability to manage negative affect may be particularly important in this population. Constructs such as distress tolerance (e.g., see Marshall-Berenz, Vujanovic, & MacPherson, 2011) and negative urgency (e.g., see Cyders & Smith, 2008) may be particularly important in moving forward with research in this area and clinical application. Such intervention efforts are consistent with data showing that emotional intelligence including regulatory processes, may be protective against urges to use alcohol in response to negative affect (de Sousa Uva et al., 2010).

Limitations and Future Directions

This study also had limitations. Predominant among these is that the cross-sectional design of this study precludes us from establishing temporal precedence among these variables, and does not allow for us to examine the unfolding of mediated processes as they are posited to occur according to self-medication theory. Further, both theory (Chicoat & Breslau, 1998) and empirical data (Read, Wardell, & Colder, 2013) suggest multiple paths of association between PTSD and alcohol beyond just self-medication associations. Cross-sectional data cannot tease out other plausible orders of temporal association. Thus, though our study offers a first step in helping to delineate the role of emotion regulation in PTSD-alcohol associations, future research using longitudinal design will offer a stronger test of these pathways and their direction of temporal association.

This study also was limited in that measurement was based on self-report of conscious efforts to regulate emotion. Furthermore, because the wording of most of the DERS items asks participants to report on difficulty in emotion regulation in the context of negative affect (i.e., “when I’m upset…”), it is possible that this may have resulted in greater endorsement of emotion regulation processes in those with greater distress (PTSD). This is a potential confound. Additionally, this assessment of emotion regulation may also limit our ability to evaluate the scope of the relationship between PTSD and alcohol use, as individuals with PTSD may rely on alcohol use to moderate affect even when not in distress due to general deficits in emotion regulation. Emotion regulation has been conceptualized and operationally defined in many ways, and is thought to occur both consciously and unconsciously; thus, perhaps future research may also want to consider processes that occur outside of the conscious awareness. Future studies may further expand on the present study by exploring more nuanced conceptual differentiations of emotion regulation as it occurs in real time. For example, such studies would allow for differentiation between antecedent and response-focused approaches as described by Gross (1998). Moving forward with research in this area, it may be particularly important to consider how to measure emotion regulation, types of emotion regulation, the role of context in the emotion regulation process, and in a way that it is not confounded with assessments of negative emotions themselves (Aldao, 2013; Aldao & Dixon-Gordon, 2014; Berking & Wupperman, 2012; Gross, 1998; Hassija, Luterek, Naragon-Gainey, Moore, & Simpson, 2012). Of note, alcohol use was also assessed via self-report and future studies examining alcohol use in vivo in relation to emotion regulation could provide more detailed insight into the association between drinking and the regulation of emotions.

Finally, the current sample consisted of college students who were recruited for being alcohol users rather than meeting diagnostic criteria for an alcohol use disorder. Thus it is possible that results may not generalize to those dependent on alcohol or to those with greater or more chronic PTSD symptomatology. Though the average participant in the PTSD group endorsed 7.11 symptoms in the past month which reflects clinically significant pathology, 62% of this group was comprised of those with subthreshold PTSD (i.e., partial). An important next step will be to test whether associations observed here generalize to other samples (e.g., clinical samples, non-college, veterans, etc.).

Still, despite these limitations, this study adds to the growing knowledge base of the function of emotion regulation in the association between PTSD and alcohol. Most importantly, use of gold-standard assessment procedures for PTSD allowed for the comparison of mechanistic pathways for those with no criterion A trauma, criterion A trauma without PTSD, and criterion A trauma with PTSD bringing focus to the critical role of PTSD symptomatology. Knowledge from these findings may be applied to interventions geared toward the amelioration of problem drinking among those with PTSD.