Evidence for a unique PTSD construct represented by PTSD's D1–D3 symptoms
Research highlights
▶ PTSD symptoms of sleep, concentration and irritability problems represent a unique PTSD factor. ▶ A 5-factor PTSD model fit significantly better than four-factor models. ▶ The Emotional Numbing model fit better than the Dysphoria model.
Introduction
The large volume of recent studies examining the factor structure of posttraumatic stress disorder (PTSD) has revealed that two four-factor models best represent the PTSD construct: (1) King, Leskin, King, and Weathers’ (1998) model entailing reexperiencing, effortful avoidance, emotional numbing and hyperarousal; and (2) Simms, Watson, and Doebbeling's (2002) model entailing reexperiencing, effortful avoidance, dysphoria and hyperarousal (reviewed in Elhai et al., 2009b, Shevlin et al., 2009). These models differ only in the placement of three PTSD symptoms: difficulty sleeping (PTSD's symptom D1), irritability (D2), and difficulty concentrating (D3). Symptoms D1–D3 are part of the King et al. model's hyperarousal factor, but part of the Simms et al. model's dysphoria factor. This study is the first to test whether PTSD's D1–D3 symptoms represent a unique construct within PTSD's factor structure, which may clarify questions about the role of depression-related symptoms in posttraumatic reactions.
Since the 1990s, factor analytic research has demonstrated that DSM-IV's (American Psychiatric Association, 1994) tripartite PTSD model (reexperiencing, avoidance/numbing, and hyperarousal) does not adequately account for PTSD's factor structure (reviewed in Asmundson, Stapleton, & Taylor, 2004). King et al. (1998) developed and tested the Emotional Numbing PTSD model, separating DSM-IV PTSD's avoidance and numbing factors, resulting in a four-factor model: reexperiencing (B1–B5), avoidance (C1–C2), numbing (C3–C7) and hyperarousal (D1–D5). The model reflects evidence demonstrating that avoidance and numbing are differentially related to psychopathology measures and post-treatment outcomes, and differentially predict poor treatment response prior to treatment initiation (reviewed in Asmundson et al., 2004). Numerous confirmatory factor analytic (CFA) studies have found empirical support for the Emotional Numbing model in various trauma-exposed samples of adults (most recently in Elhai et al., 2008, Elhai et al., in press, Grubaugh et al., 2010, Mansfield et al., 2010, McDonald et al., 2008, Naifeh et al., 2008, Palmieri et al., 2007a, Palmieri et al., 2007b) and adolescents (Saul, Grant, & Carter, 2008).
However, conceptual problems have been noted with the Emotional Numbing PTSD model. Simms et al. (2002) argued that several of PTSD's symptoms are examples of general emotional distress common to other anxiety and mood disorders. Simms et al. proposed (a) separating PTSD's D1–D3 symptoms from the Emotional Numbing model's hyperarousal factor and (b) combining them with symptoms C3–C7 to form an eight-item dysphoria factor to reflect this large distress/dysphoria construct. The Dysphoria PTSD model has received empirical support in various trauma-exposed samples of adults (most recently, in Armour and Shevlin, 2010, Carragher et al., 2010, Elhai et al., 2009a, Elhai et al., 2009b, Engdahl et al., in press, Naifeh et al., 2010, Palmieri et al., 2007b, Pietrzak et al., 2010, Shevlin et al., 2009) and adolescents (Elhai, Ford, et al., 2009).
Across the literature, the Emotional Numbing and Dysphoria models generally are the best fitting PTSD models in relation to other similar models (including the three-factor DSM-IV model), with few exceptions (reviewed in Elhai et al., 2009b, Shevlin et al., 2009). However, neither model has emerged as the best fitting model across a clear majority of studies. Recent investigations have attempted to elucidate measurement conditions under which one of these models fits better than the other. Palmieri, Weathers, et al. (2007) found in a sample of disaster workers that the Dysphoria model fit best when a self-report PTSD instrument was used, while the Emotional Numbing model fit best when an interviewer-administered instrument was used. Furthermore, Elhai, Engdahl, et al. (2009) found that among trauma-exposed college students, the Dysphoria model fit best when respondents were instructed to rate PTSD symptoms from their most distressing traumatic event, while the Emotional Numbing model fit best when PTSD symptoms were rated from one's overall trauma history. Finally, Armour et al. (in press) found in war-exposed adolescents that the Emotional Numbing model fit best when rating PTSD symptoms from a traumatic event that met PTSD's criterion A2 (intense fear, helplessness or horror during the event), and neither model fit best when rating symptoms from a traumatic event not meeting criterion A2.
Even though most studies find that either the Emotional Numbing model or Dysphoria model fits best among competing PTSD models, the resulting fit indices do not always demonstrate an excellent fit. “Excellent fit” in CFA studies is traditionally determined by empirically-defined benchmarks, including the Comparative Fit Index (CFI) and Tucker-Lewis Index (TLI) ≥ .95, root mean square error of approximation (RMSEA) ≥ .06, and standardized root mean square residual (SRMR) ≥ .08 (Hu and Bentler, 1998, Hu and Bentler, 1999). Yet in numerous studies, these requirements for excellent fit were not satisfied across the Emotional Numbing or Dysphoria four-factor PTSD models, albeit fitting better than other models (e.g., recently found in Boelen et al., 2008, Carragher et al., 2010, Elhai et al., 2009a, Naifeh et al., 2010, Ullman and Long, 2008).
In this investigation, we explored the possibility of improving PTSD's model fit using a simple model alteration. We used the Emotional Numbing and Dysphoria models as the starting point. We begin by discussing the primary difference between the Emotional Numbing and Dysphoria models – the placement of PTSD's D1–D3 symptoms. In the Emotional Numbing model, D1–D3 symptoms are placed within the hyperarousal factor, while in the Dysphoria model they are placed within the dysphoria factor. Yet it can be argued that the D1–D3 symptoms are conceptually different from both hyperarousal and dysphoria.
Watson (2005) discussed a distinction between the D1–D3 symptoms on one hand and PTSD's remaining hyperarousal symptoms: D4 (hypervigilance) and D5 (exaggerated startle response). The distinction lies in the fact that D1–D3 symptoms involve general distress or dysphoria (albeit, in an agitated/restless manner); D4–D5 involve the anxious arousal that is characteristic of fear-based disorders such as panic disorder (Watson, 2005). And D1–D3 differ conceptually from dysphoria symptoms because D1–D3 involve restlessness and agitation (e.g., sleep difficulty, irritability) that is different from the dysphoria factor's remaining symptoms that involve a numbing of responsiveness (e.g., anhedonia, social isolation). In fact, Shevlin et al. (2009) discovered that D1–D3 symptoms loaded on both the hyperarousal and dysphoria factors, but the factor loadings were not particularly high (ranging from only .28 to .57, and much lower than other factors’ loadings). Shevlin et al. concluded that these three symptoms were not clear indicators of either factor.
One problem with Simms et al.’s (2002) Dysphoria model is that the authors simultaneously modified the Emotional Numbing model in two ways: (1) separating symptoms D1–D3 from the Hyperarousal factor, and (2) combining D1–D3 items with the Emotional Numbing model's Numbing symptoms to form the dysphoria factor (Simms et al., 2002). Thus we do not know which modification improved model fit.
In this study, we test whether (a) separating symptoms D1–D3 from the Emotional Numbing model's hyperarousal factor is empirically supported, and (b) whether separating D1–D3 from the Dysphoria model's dysphoria factor is empirically supported. We used a 5-factor model as an alteration to both the Emotional Numbing and Dysphoria models for comparison purposes: (a) reexperiencing (B1-B5), (b) avoidance (C1–C2), (c) emotional numbing (C3–C7), (d) the D1–D3 symptoms which we will call dysphoric arousal, and (e) anxious arousal (D4–D5) discussed by Watson (2005) and Simms et al. (2002) (item mappings for these models are found in Table 1).
We investigated these research questions in a multi-site study of women victims of recent domestic violence, an “at-risk” sample for developing mental health problems including PTSD. In analysis 1 we compare the Emotional Numbing model to this 5-factor model; this comparison will help determine whether dysphoric arousal is statistically different from anxious arousal. In analysis 2 we compare the Dysphoria model with the 5-factor model; this comparison will help determine whether dysphoric arousal is statistically different from the dysphoria factor. We hypothesized, based on the theoretical and empirical work by Watson (2005) and Shevlin et al. (2009), that the 5-factor model would fit significantly better than the 4-factor Emotional Numbing and Dysphoria models, demonstrating that the dysphoric arousal factor is independent of both hyperarousal and dysphoria. Results will help inform our understanding of PTSD's factor structure and the role of depressive symptoms, as part of the current effort to redefine PTSD's characteristics, and more generally will have implications for our understanding of posttraumatic reactions.
Section snippets
Procedure
Two research assistants invited women, aged 18–70 years old, from five Midwestern publicly-funded domestic violence shelters to participate in this study. These shelters provide short-term residential care, legal, educational, and healthcare services to women domestic violence victims. Participants were recruited at shelters on varying days of the week and times of day, during spring and summer of 2009. During those times, all shelter residents, in the shelter at that time, were approached for
Results
Regarding the domestic violence incident leading to shelter assistance, most participants reported that it occurred in the prior week (n = 101, 40.4%), 1–4 weeks ago (n = 93, 37.2%), or 1–6 months ago (n = 38, 15.2%). The most prevalent (non-mutually exclusive) previous traumatic events included adult physical abuse (n = 231, 91.7%), child physical abuse (n = 137, 54.4%), completed rape (n = 134, 53.2%), attempted rape (n = 128, 50.8%), losing a very close family member or friend in an accident, homicide or
Discussion
We found that symptoms we refer to as dysphoric arousal represent a separate construct from the Emotional Numbing model's arousal factor, and the Dysphoria model's dysphoria factor. Separating dysphoric arousal symptoms into its own factor resulted in a significantly better fitting (5-factor) model than when implementing the 4-factor Emotional Numbing or Dysphoria models. And using this 5-factor model resulted in uniformly large factor loadings that were at least as large as the largest of the
References (43)
- et al.
The factor structure of posttraumatic stress disorder symptoms among bereaved individuals: a confirmatory factor analysis study
Journal of Anxiety Disorders
(2008) - et al.
Factor structure of posttraumatic stress disorder symptoms in the Australian general population
Journal of Anxiety Disorders
(2010) - et al.
An examination of the construct validity of posttraumatic stress disorder with veterans using a revised criterion set
Behaviour Research and Therapy
(2010) - et al.
The PTSD Symptom Scale's latent structure: an examination of trauma-exposed medical patients
Journal of Anxiety Disorders
(2008) - et al.
Structure of posttraumatic stress disorder symptoms and psychosocial functioning in Veterans of Operations Enduring Freedom and Iraqi Freedom
Psychiatry Research
(2010) - et al.
Reconciling the differences between the King et al. (1998) and Simms et al. (2002) factor models of PTSD
Journal of Anxiety Disorders
(2009) Diagnostic and statistical manual of mental disorders
(1994)- American Psychiatric Association—DSM-5 Development. (2010). Proposed draft revisions to DSM disorders and criteria....
- Armour, C., Layne, C. M., Naifeh, J. A., Shevlin, M., Duraković-Belko, E., Djapo, N., et al. (in press). Assessing the...
- et al.
Testing the dimensionality of PTSD and the specificity of the dysphoria factor
Journal of Loss and Trauma
(2010)