Emotion Regulation Strategy Use and Posttraumatic Stress Disorder: Associations Between Multiple Strategies and Specific Symptom Clusters Daniel J. Lee, Tracy K. Witte, Frank W. Weathers & Margaret T. Davis
Journal of Psychopathology and Behavioral Assessment ISSN 0882-2689 J Psychopathol Behav Assess DOI 10.1007/s10862-014-9477-3
1 23
Your article is protected by copyright and all rights are held exclusively by Springer Science +Business Media New York. This e-offprint is for personal use only and shall not be selfarchived in electronic repositories. If you wish to self-archive your article, please use the accepted manuscript version for posting on your own website. You may further deposit the accepted manuscript version in any repository, provided it is only made publicly available 12 months after official publication or later and provided acknowledgement is given to the original source of publication and a link is inserted to the published article on Springer's website. The link must be accompanied by the following text: "The final publication is available at link.springer.com”.
1 23
Author's personal copy J Psychopathol Behav Assess DOI 10.1007/s10862-014-9477-3
Emotion Regulation Strategy Use and Posttraumatic Stress Disorder: Associations Between Multiple Strategies and Specific Symptom Clusters Daniel J. Lee & Tracy K. Witte & Frank W. Weathers & Margaret T. Davis
# Springer Science+Business Media New York 2015
Abstract A growing literature suggests that emotion regulation (ER) is associated with posttraumatic stress disorder (PTSD). However, most of the studies in this literature have one or more important limitations, including examining only a single ER strategy (e.g., thought suppression) rather than multiple strategies simultaneously, examining PTSD at the syndrome level rather than by symptom cluster, and failing to control for negative affect. The present study sought to address these limitations by using latent variable modeling to examine the associations between multiple ER strategies and individual PTSD symptom clusters while controlling for negative affect. Of the four measurement models of ER strategy use examined, the best-fitting model allowed items corresponding to each included strategy to load onto their independent factors. Of the four measurement models of PTSD symptoms examined, the best-fitting model was the fivefactor dysphoric arousal model. Results of structural models indicated that thought suppression and experiential avoidance were associated with most PTSD symptom clusters, even after controlling for negative affect. However, most other included ER strategies were not associated with any symptom clusters. A number of issues regarding measurement of ER and PTSD are discussed, and several suggestions for future research are provided.
Keywords Emotion regulation . PTSD . Experiential avoidance . Thought suppression . Trauma D. J. Lee (*) : T. K. Witte : F. W. Weathers : M. T. Davis Department of Psychology, Auburn University, Auburn, AL 36849-5214, USA e-mail:
[email protected]
Emotion Regulation Strategy Use and Posttraumatic Stress Disorder: Associations Between Multiple Strategies and Specific Symptom Clusters A growing literature suggests that emotion regulation (ER) – or efforts to affect the intensity, duration, or likelihood of experiencing a particular emotion (Gross, 1998) – is associated with posttraumatic stress disorder (PTSD; see Kring & Sloan, 2010 for a review). Indeed, this association is formalized in the diagnostic criteria for PTSD: two of the core symptoms of PTSD, avoidance of trauma-related thoughts and feelings and avoidance of trauma-related cues in the environment, are conceptualized as ER strategies and are thought to play a prominent role in the development and maintenance of the disorder (Badour et al. 2012; Foa & Kozak, 1986; Foa et al. 1989; Pineles et al., 2011). Further, a number of longitudinal studies (Ehring et al. 2008 Kumpala et al. 2011; Marx & Sloan, 2005; Nightingale & Williams, 2000) have demonstrated that avoidance and other ER strategies such as expressive suppression and rumination, play important roles in the etiology of PTSD. As currently used, ER is an umbrella label encompassing a number of related constructs. Several theoretical models have attempted to categorize ER strategies. Among the most influential is Gross’s process model (Gross, 1998), which proposes that ER strategies fall into five general categories, following the temporal sequence of the development and management of an emotion: (a) situation selection (e.g., not attending an anxiety-provoking situation to avoid becoming anxious), (b) situation modification (e.g., steering a conversation away from emotionally provocative topics), (c) attentional deployment (e.g., diverting attention away from a distressing idea), (d) cognitive change (e.g., trying to think about a distressing topic in a different way), and (e) response modulation (e.g.,
Author's personal copy J Psychopathol Behav Assess
breathing deeply to reduce anxiety). Another organizational model of ER proposed by Aldao et al. (2010) categorizes ER strategies as adaptive or maladaptive (i.e., as either protective factors against or risk factors for psychopathology, respectively). Other research has suggested ER strategies may be best accounted for by a single underlying latent construct (Aldao & Nolen-Hoeksema, 2010). Although research examining the latent factor structure of individual scales of ER strategy use exists, the larger factor structure underlying use of multiple strategies remains largely unexamined to date. The association between ER strategy use and PTSD has been investigated extensively. Several ER strategies have been established as being positively associated with PTSD symptom severity: thought suppression (e.g., Beck et al. 2006; Rosenthal et al. 2006; Shipherd & Beck 1999), expressive suppression (e.g., Moore et al. 2008; Nightingale and Williams, 2000), experiential avoidance (e.g., Ehring & Quack, 2010; Kumpala et al., 2011; Marx & Sloan, 2005), and rumination (e.g., Ehring et al., 2008; Michael et al. 2007). Thought suppression involves efforts to reduce the frequency of specific thoughts (Wegner et al. 1987). Expressive suppression involves actively reducing outward displays of subjective emotional experience (Gross, 1998). Experiential avoidance involves a general unwillingness to experience, and efforts to avoid, unwanted subjective events such as thoughts and emotions (Bond et al., 2011; Hayes et al., 1996). Rumination involves a passive pattern of perseverative focus on the cause and consequences of particular emotional experiences (McLaughlin & Nolen-Hoeksema, 2011). Conversely, two ER strategies have been established as being inversely associated with PTSD symptom severity: acceptance and positive reappraisal (e.g., Badour et al. 2012; Kashdan, Alvarez, & Gross, 2012; Vujanovic et al. 2009). Additionally, a number of acceptance-based interventions (e.g., mindfulness-based stress reduction, transcendental meditation) have been demonstrated to be effective in reducing PTSD symptom severity in both civilian and veteran populations (e.g., Kearney et al. 2012; Rosenthal et al. 2011). Acceptance involves the non-evaluative acknowledgment of affective experience (Baer et al. 2004; Hayes et al. 1996). Positive reappraisal involves thinking about an emotioneliciting stimulus in a way that affects subjective emotional response (Gross, 1998). To date, many studies investigating the association between ER strategy use and PTSD have examined individual strategies in isolation. This approach is limiting for two primary reasons. First, it precludes between-strategy comparisons regarding the relative magnitude of association with other constructs of interest. Second, it precludes examination of the larger factor structure of ER strategy use in general, i.e., the factor structure that would emerge from simultaneous consideration of multiple strategies. Further, much of the existing research examining the association between ER
strategy use and PTSD has examined PTSD at the syndrome level, rather than at the symptom cluster level. Confirmatory factor analyses have demonstrated consistently that PTSD consists of at least four, and possibly five, correlated but distinct factors (for a recent review see Elhai & Palmieri, 2011). Examination of PTSD at the symptom cluster level would provide valuable insight regarding which aspects of PTSD are most strongly associated with specific ER strategies. Lastly, an additional concern regarding the existing ER literature is that some of the observed associations between aspects of ER and PTSD may be inflated by negative affect (Tull et al. 2007). Negative affect encompasses a variety of negatively valenced emotions and moods such as anger, sadness, and guilt (Gross & Thompson, 2007; Watson et al. 1988; Watson & Pennebaker, 1989). Not surprisingly, negative affect is positively associated with PTSD symptom severity. However, it has been demonstrated to be associated with several aspects of ER as well (e.g., Tull et al., 2007). Accordingly, individuals experiencing greater PTSD symptom severity may be utilizing particular ER strategies due to experiencing greater degree of general negative affect. This suggests that negative affect may inflate the observed association between ER strategy use and PTSD. Accordingly, the unique association between ER strategy use and PTSD beyond degree of negative affect can be explored by controlling for individual differences in this construct. To date, few studies in this area have examined this association while controlling for negative affect. The purpose of the present study was to examine the association between ER strategy use and PTSD symptoms while addressing several of the limitations of previous studies. The first aim was to examine the larger factor structure underlying use of multiple ER strategies. Given the exploratory nature of these analyses, we did not make specific predictions about which model would best fit the data. The second aim was to examine the underlying factor structure of PTSD. Given the emerging trend of research finding support for the five-factor model of PTSD symptoms (Elhai et al. 2011), we hypothesized that this model would best fit the data. The third aim was to examine the association between the latent variables underlying ER strategy use and the latent variables corresponding to PTSD symptom clusters. We hypothesized that results would replicate existing research demonstrating a positive association between PTSD symptom severity and thought suppression, expressive suppression, experiential avoidance, and rumination, and an inverse association between PTSD symptom severity and acceptance and positive reappraisal. Lastly, the fourth aim was to examine these associations while controlling for negative affect. We hypothesized that the strength of observed associations would be substantially weakened after controlling for negative affect.
Author's personal copy J Psychopathol Behav Assess
Method Participants & Procedure Participants were undergraduate students enrolled in psychology courses at a large southeastern university who completed an online survey for course credit. Determination of Criterion A (the trauma exposure criterion for PTSD) status was based on a review of brief narratives participants were asked to provide describing the index event and confirmed by the Traumatic Life Event Questionnaire (TLEQ; Kubany et al., 2000) rating for the corresponding event category. Narratives were reviewed by the third author (FW), an expert in assessment of trauma exposure and PTSD. A conservative threshold was used such that participants were included only if their index event unequivocally met Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-5; APA, 2013) Criterion A. In particular, in keeping with a notable change in Criterion A for DSM-5, the sudden, unexpected death of a loved one was considered to meet Criterion A only if it involved a close family member or close friend and was due to an accident or violence. Of the 322 participants who completed the survey, 213 (66 %) described an index event that met DSM-5 Criterion A, 35 (11 %) described an index event in the narrative that did not meet DSM-5 Criterion A, 59 (18 %) did not provide adequate information in the narrative to determine Criterion A status, and 15 (5 %) did not provide a narrative. Time since index event occurrence ranged from less than 1 to 19 years (median=3 years). A summary of index event categories endorsed is provided in Table 1. Sixty-two participants (29.12 %) had PTSD Checklist – Specific version (PCL; Weathers et al. 1993) scores at or above 44, indicating
Table 1 Events
DSM-5 Criterion A Traumatic Events Endorsed as Index
Traumatic Event Category
Participants (% of sample)
Transportation Accident Sudden accidental death Sudden violent death
41 30 26
19.2 14.1 12.2
Life threatening illness or injury Sexual assault Natural Disaster Assault with a weapon Physical assault Serious accident at home, work or during recreational activity Other stressful experience Other unwanted sexual experience Combat or exposure to warzone Fire/Explosion
23 22 19 13 11 9
10.8 10.3 8.9 6.1 5.2 4.2
9 7 2 1
4.2 3.3 0.9 0.5
they meet provisional criteria for PTSD (Blanchard et al. 1996; Ruggiero et al. 2003). Of the 213 participants retained, the majority identified as female (n=166, 77.93 %). Mean age was 20.77 (SD=2.24, minimum=18, maximum=38). The majority of participants identified as Caucasian (n=194, 91.08 %), and the rest identified as African American (n=14, 6.57 %), Asian American (n=2, 0.94 %), or other (n=3, 1.41 %). With regard to ethnicity, nine participants (4.23 %) identified as Hispanic or Latino. All procedures used were approved by the university Institutional Review Board (IRB). Measures A demographics questionnaire was used to collect information regarding gender, age, race, ethnicity, student, work, and marital status, and current prescription medication use information. Trauma exposure history was assessed with the TLEQ (Kubany et al., 2000), which assesses frequency of exposure to a wide range of traumatic events, including natural disasters, combat, and assault. The TLEQ demonstrated adequate temporal stability and convergent validity with a clinicianadministered version of the measure (Kubany et al., 2000). Additionally, participants were asked to provide a brief written account of their identified worst event. This narrative was used to rate whether index events met Criterion A. PTSD symptom severity was assessed with the 17-item PCL. The PCL instructs respondents to rate how much they have been bothered by PTSD symptoms in the past month, using a 5-point scale ranging from not at all to extremely. Higher PCL scores indicate greater PTSD severity. PCL scores have strong reliability and validity (for reviews, see McDonald & Calhoun, 2010; Wilkins et al. 2011). Negative affect was assessed with the 10-item negative affect subscale of the Positive and Negative Affect Schedule (PANAS-NA; Watson et al., 1988). The PANAS instructs respondents to rate the degree to which they have experienced a variety of positive and negative emotions during the past few weeks on a 5-point scale ranging from very slightly or not at all to extremely. Higher PANAS-NA scores indicate greater negative affect. Thought suppression was assessed with the 15-item White Bear Thought Suppression Inventory (WBSI; Wegner & Zankos, 1994). The WBSI instructs respondents to rate their agreement with a list of statements on a 5-point scale ranging from strongly disagree to strongly agree. Higher WSBI scores indicate greater use of thought suppression. Rumination was assessed with the 22-item Ruminative Responses Scale (RRS; Nolen-Hoeksema & Morrow, 1991), which measures two distinct dimensions of rumination: brooding and reflection (Treynor et al. 2003). The RRS instructs respondents to rate how often they utilize these strategies when feeling depressed on a 4-point scale ranging from
Author's personal copy J Psychopathol Behav Assess
almost never to almost always. Higher RSS subscale scores indicate greater use of rumination. Expressive suppression and positive reappraisal were assessed with respective subscales of the 10-item Emotion Regulation Questionnaire (ERQ; Gross & John, 2003). The ERQ instructs respondents to rate their agreement with a list of statements related to each included construct on a 7-point scale ranging from strongly disagree to strongly agree. Higher scores on each subscale indicate greater use of the respective strategy. Experiential avoidance was assessed with the 10-item second edition of the Acceptance and Action Questionnaire (AAQ-II; Bond et al., 2011). The 7-item one-factor scoring supported by the most recent psychometric data (Bond et al., 2011) was used in the current study. The AAQ-II instructs respondents to rate each item on a 7-point scale ranging from never true to always true. For the current study, the AAQ-II was scored such that higher scores indicate greater experiential avoidance. Acceptance was assessed with the 9-item accepting present experience without judgment subscale of the Kentucky Inventory of Mindfulness Skills (KIMS; Baer et al., 2004). The KIMS is a 39-item measure of several dimensions of mindfulness. However, only the accepting present experience without judgment subscale was used in the current study. This measure instructs respondents to rate items on a 5-point scale ranging from never or very rarely true to very often or always true. Higher scores on this subscale indicate greater use of acceptance.
Table 3 Item Mapping for Measurement Models of DSM-IV Posttraumatic Stress Symptom Clusters Model PTSD Symptom
DSM- Numbing Dysphoria Dysphoric IV Arousal
B1: Intrusive thoughts B2: Nightmares B3: Reliving trauma B4: Emotional cue reactivity B5: Physiological cue reactivity C1: Avoidance of thoughts C2: Avoidance of reminders C3: Trauma-related amnesia C4: Loss of interest C5: Feeling detached C6: Feeling numb C7: Hopelessness D1: Difficulty sleeping D2: Irritability D3: Difficulty concentrating D4: Overly alert
R R R R
R R R R
R R R R
R R R R
R
R
R
R
A/N
A
A
A
A/N
A
A
A
A/N
N
D
N
A/N A/N A/N A/N H H H
N N N N H H H
D D D D D D D
N N N N DA DA DA
H
H
H
AA
D5: Easily startled
H
H
H
AA
R=Reexperiencing; A=Avoidance; Avoidance and Numbing; H=Hyperarousal; N=Numbing; D=Dysphoria; DA=Dysphoric Arousal; AA= Anxious Arousal
Data Analytic Strategy All analyses were conducted in Mplus version 6.0 (Muthén & Muthén, 1998–2011). Measurement models of ER and PTSD symptom clusters presented in Tables 2 and 3 were tested using confirmatory factor analysis (CFA). Four measurement Table 2
Measurement Models of Emotion Regulation Model
Scale Items
A
B
C
D
KIMS-Acceptance ERQ-Positive Reappraisal RRS-Reflective Rumination RRS-Brooding Rumination WBSI-Thought Suppression ERQ-Expressive Suppression AAQ-II-Experiential Avoidance
1 1 1 1 1 1 1
1 1 2 2 2 2 2
1 2 3 3 3 1 1
1 2 3 4 5 6 7
AAQ-II=Acceptance and Action Questionnaire, 2nd edition; ERQ= Emotion Regulation Questionnaire; KIMS =Kentucky Inventory of Mindfulness Skills; RRS=Ruminative Responses Scale; WBSI=White Bear Suppression Inventory
models of ER strategy use were tested and compared. The first hypothesized model (Model A; Aldao & Nolen-Hoeksema, 2010) grouped all ER strategies into a one-factor solution. This model was evaluated against three alternative models. The first alternative (Model B; Aldao et al., 2010) was a twofactor model in which acceptance and positive reappraisal loaded onto the latent construct of adaptive ER strategy use and thought suppression, expressive suppression, experiential avoidance, and two dimensions of rumination loaded onto the latent construct of maladaptive ER strategy use. The second alternative model (Model C; Gross, 1998) grouped strategies according to Gross’ original process model of ER, which categorizes both forms of rumination and thought suppression as forms of attentional deployment; positive reappraisal as a form of cognitive change; and acceptance, expressive suppression, and experiential avoidance as forms of response modulation. The third alternative model (Model D) examined each strategy as its own first-order factor. In addition, four measurement models of PTSD symptom severity were tested and compared (see Table 3). The DSM-IV three-factor model (American Psychiatric Association, 2000) was compared
Author's personal copy J Psychopathol Behav Assess
against the four-factor emotional numbing model (King et al. 1998), the four-factor dysphoria model (Simms et al. 2002), and the five-factor dysphoric arousal model (Elhai et al., 2011). For all ER and PTSD measurement models examined, individual items were used as indicators for latent variables. Because each of the included measures uses ordinal scales and were not normally distributed (Wirth & Edwards, 2007), parameters were estimated using Mean- and VarianceAdjusted Weighted Least Squares (WLSMV), which provides a robust χ2. Missing data were handled using pairwise deletion. This approach was taken for three reasons: only a small portion of the data was missing (i.e., covariance coverage ranged from .96 to 1.00 for all item pairs), no techniques exist for pooling most fit indices across estimates in multiple imputation (Enders, 2010), and full information maximum likelihood (FIML) procedures are not available for the WLSMV estimator. Model fit was assessed using multiple indices: χ2, Tucker-Lewis index (TLI), comparative fit index (CFI), and root mean square error of approximation (RMSEA). Proposed fit statistics cutoffs outlined by Hu and Bentler (1999) and Kline (2005) were used (RMSEA≤.05, CFI and TLI≥.95). However, overall fit of each model was interpreted by taking all fit statistics into account (Brown, 2006). An a priori power analysis was conducted according to criteria established by MacCallum and colleagues (1996) and each of the examined models was determined to be adequately powered given the obtained sample. Specifically, the obtained sample exceeded identified minimum sample sizes to achieve power of .80 using both the test of close fit and test of not-close fit given the degrees of freedom in each of the specified models. For ER strategy use, measurement models A, B, and C were each compared against its parent model D. For PTSD symptom severity, the DSM-IV three-factor model, numbing four-factor model, and dysphoria alternative four-factor model were each compared against the parent dysphoric arousal fivefactor model. These nested models were compared using the DIFFTEST command in Mplus (Muthen & Muthen, 2006). The associations between the latent variables of the best fitting measurement models of ER strategy use and PTSD symptom clusters were examined in structural models with and without negative affect included as an exogenous variable.
Results Measurement Models Descriptive statistics for measures in the current study are provided in Table 4. The first set of analyses examined the factor structure of ER strategy use by comparing four proposed models (see Table 2). None of the fit indices indicated
Table 4
Descriptive Statistics of Included Measures Observed
Measure
Mean SD
PCL-Total PCL-Reexperiencing PCL-Avoidance PCL-Emotional Numbing PCL-Dysphoric Arousal PCL-Anxious Arousal PANAS-Negative Affect WBSI RRS-Brooding RRS-Reflection ERQ-Expressive Suppression ERQ-Positive Reappraisal AAQ-II KIMS-AC
36.61 15.04 17 11.61 5.01 5 5.10 2.38 2 9.43 4.93 5 6.28 3.51 3 4.20 2.50 2 20.71 7.44 10 49.46 12.64 15 10.99 3.97 5 9.92 3.28 5 14.62 5.23 4 27.23 6.27 6 22.89 9.36 7 28.09 7.05 9
Possible
Min Max Min Max 80 25 10 24 15 10 44 75 20 19 27 42 48 45
17 5 2 5 3 2 10 15 5 5 4 6 7 9
85 25 10 25 15 10 50 75 20 20 28 42 49 45
AAQ-II=Acceptance and Action Questionnaire, 2nd edition; ERQ= Emotion Regulation Questionnaire; KIMS =Kentucky Inventory of Mindfulness Skills; PANAS=Positive and Negative Affect Schedule; PCL=PTSD Checklist; RRS=Ruminative Responses Scale; WBSI= White Bear Suppression Inventory
close fit for the one-factor model A (χ2 =4,785.17, df=1,224, p < .001; CFI = .74; TLI = .73; RMSEA = .12, 90 % CI [.12–.12]), two-factor model B (χ2 =4,404.69, df=1,222, p < .001; CFI = .77; TLI = .76; RMSEA = .11, 90 % CI [.11–.12]), and three-factor model C (χ2 =2,838.59, df=1, 221, p