P1: FRC/GCR/GCX/GCZ Journal of Traumatic Stress

P2: GCO/GCP/GCQ PP047-293807

January 15, 2001

17:8

Style file version February 22, 1999

Journal of Traumatic Stress, Vol. 14, No. 1, 2001

The Relationship Among Personality Variables, Exposure to Traumatic Events, and Severity of Posttraumatic Stress Symptoms1 Dean Lauterbach2,4 and Scott Vrana3

In examining predictors of posttraumatic stress, researchers have focused on trauma intensity and devoted less attention to other variables. This study examined how personality and demographic variables are related to the likelihood of experiencing a trauma, and to the severity of posttraumatic symptoms in a sample of 402 college students reporting a wide range of trauma. Elevations in antisocial and borderline traits were significant predictors of retraumatization, accounting for 12% of the variance. Personality variables and trauma intensity were significant predictors of PTSD severity, accounting for 43% of the variance. Neuroticism interacted with trauma intensity in predicting Posttraumatic Stress Disorder (PTSD) severity. Among persons low in Neuroticism, there was a modest trauma intensity– PTSD relationship, whereas among persons high in neuroticism there was a strong relationship. KEY WORDS: PTSD; posttraumatic stress disorder; predictors; vulnerability.

Numerous models of traumatic stress have emerged that attempt to explain the genesis and maintenance of psychological symptoms following a traumatic experience (Chemtob, Roitblat, Hamada, Carlson, & Twentyman, 1988; Foa, Steketee, & Rothbaum, 1989; Horowitz, 1975; Keane, Fairbank, Caddell, Simering, & Bender, 1985; Kreitler & Kreitler, 1988; van der Kolk, 1988). Many of these 1 This

manuscript was part of the first author’s doctoral dissertation. The authors would like to thank Rebecca Davis Merritt, Ph.D., E. J. Capaldi, Ph.D., and Eliott Smith, Ph.D., for their thoughtful feedback on early versions. Portions of this manuscript have been presented at the 8th annual conference of the American Psychological Society and at the 12th annual conference of the International Society for Traumatic Stress Studies. 2 Psychology Department, Northwestern State University, Natchitoches, Louisiana 71497. 3 Virginia Commonwealth University, Department of Psychology, Richmond, Virginia. 4 To whom correspondence should be addressed at Psychology Department, Northwestern State University, Natchitoches, Louisiana 71497; e-mail: [email protected]. 29 C 2001 International Society for Traumatic Stress Studies 0894-9867/01/0100-0029$19.50/1 °

P1: FRC/GCR/GCX/GCZ Journal of Traumatic Stress

30

P2: GCO/GCP/GCQ PP047-293807

January 15, 2001

17:8

Style file version February 22, 1999

Lauterbach and Vrana

models have not not thoroughly examined the relationship among personality traits, demographic variables, risk of experiencing a trauma, and the severity of Posttraumatic Stress Disorder (PTSD) symptoms. Recently, investigators have begun examining the relationship between personality traits and trauma exposure (Breslau, Davis, & Andreski, 1995; Breslau, Davis, Andreski, & Peterson, 1991; Fontana & Rosenheck, 1993; Helzer, Robins, & McEvoy, 1987; Helzer, Robins, Wish, & Hesselbrock, 1979; Lee, Vaillant, Torrey, & Elder, 1995; Simons, King, & King, 1991). Preliminary findings suggest that personality traits related to impulse control may be associated with an increased probability of encountering a traumatic event. One personality trait that has received considerable attention is antisocial personality, and in particular one component of antisocial personality, early behavioral difficulties. For example, preservice antisocial behavior predicted Vietnam veterans’ assignment to a combat unit (Helzer et al., 1979) and level of exposure to traditional combat (King, King, Foy, & Gundanowski, 1996). The National Epidemiological Catchment Area Survey (Helzer et al., 1987) found that behavioral problems prior to age 15 predicted being beaten, mugged, or having seen combat if they served in Vietnam. Lastly, veterans who elected to remain in Vietnam for extended duty or a second tour scored higher than their one-tour counterparts on a composite measure of premilitary adjustment difficulties (Simons et al., 1991). These studies suggest that early behavioral difficulties may be related to trauma exposure. An alternative hypothesis is that sensation seeking, another component of antisocial personality, might be responsible for multiple traumatization. High levels of sensation seeking have been observed in a variety of populations including rock climbers, high-speed motorcyclists (Gridley, 1990), and downhill skiers (Bouter, Knipschild, Feij, & Volovics, 1988). A propensity to engage in high risk behaviors might increase the probability of experiencing a traumatic event. An important next step is to determine if other elements of antisocial personality, such as sensation seeking, might predict additional variance in trauma exposure. One personality trait that may be related to trauma exposure among women is borderline personality. Borderline personality may represent the same risk factor among women that antisocial personality does among men. Both disorders are characterized by impulsiveness, and engaging in high risk behaviors that might increase the likelihood of encountering some traumatic event. Further, whereas men are more frequently diagnosed with Antisocial Personality Disorder than women, the reverse is true for Borderline Personality Disorder (American Psychiatric Association, 1987; Widiger & Weissman, 1991; Wolfgang, Lichtermann, Klingler, Heun, & Mallmayer, 1992). For these reasons, it is essential that the preliminary findings regarding the relationship between various personality variables and trauma exposure be extended to include borderline personality. Personality variables are also related to symptom severity (Davidson, Kudler, & Smith, 1987; Lee et al., 1995; McFarlane, 1988a, 1988b; Schnurr & Vielhauer,

P1: FRC/GCR/GCX/GCZ Journal of Traumatic Stress

P2: GCO/GCP/GCQ PP047-293807

January 15, 2001

17:8

Style file version February 22, 1999

Correlates of Trauma Exposure and Responsiveness

31

2000; Sutker, Davis, Uddo, & Ditta, 1995). Elevated levels of two personality traits, Neuroticism and Introversion, have been found in trauma survivors with symptoms of PTSD. In a study of World War II and Korean conflict veterans with PTSD (Davidson et al., 1987), the trauma survivors scored higher on Neuroticism than either age-matched controls, or depressed psychiatric controls. Similarly, Lee et al. (1995) found that Neuroticism was associated with elevated levels of PTSD in a sample of World War II veterans. In a longitudinal study of fire fighters exposed to a major bushfire in South Australia, McFarlane found that persons diagnosed with chronic posttraumatic stress had significantly higher Neuroticism scores than persons with less severe traumatic stress (McFarlane, 1988a) or persons with no disorder (McFarlane, 1988b). Two of these studies (Davidson et al., 1987; McFarlane, 1988b) also found significantly elevated levels of Introversion among the trauma survivors with PTSD. Thus, it would appear that Neuroticism and Introversion are related to the development of PTSD symptomatology following a traumatic event. Possibly the most definitive findings regarding the relationship between PTSD severity and personality come from the National Vietnam Veterans Readjustment Study (NVVRS: Kulka et al., 1990). This study examined the influence of premilitary factors, such as childhood and family background variables, on the probability of PTSD developing among Vietnam combat veterans, Vietnam era veterans, and matched controls who were not in the military. Generally, the adjustment for pretrauma influences decreased the difference in PTSD prevalence rates between combat veterans and the comparison groups. After examining the influence of trauma intensity and premilitary factors, the NVVRS concluded that . . . these results are consistent with a model of PTSD that posits a role for individual vulnerability (potentially including biological, psychological, and sociodemographic predisposing factors) and a role for exposure to environmental factors (specifically, war-zone stressors) in determining who among theater veterans develops PTSD. (p. 77)

The research literature is in agreement that posttraumatic stress symptoms are strongly affected by the intensity of the trauma. Further, there is an emerging body of literature which suggests that personality factors are related to the number of traumas experienced and the severity of PTSD symptoms that emerge. However, no study has examined all of these variables together to assess the relative strength of association between each predictor and posttraumatic stress symptomatology. Further, most studies have focused on the category of persons experiencing one specific trauma, often combat, reducing the ability to generalize findings to a broad range of experiences. The focus on combat veterans is also problematic because women are often not studied. Although several studies have included women (Breslau et al., 1991, 1995; Helzer et al., 1979, 1987), they did not report data on trauma exposure for women broken down by severity of early behavioral

P1: FRC/GCR/GCX/GCZ Journal of Traumatic Stress

P2: GCO/GCP/GCQ PP047-293807

January 15, 2001

17:8

32

Style file version February 22, 1999

Lauterbach and Vrana

difficulties. However, Helzer et al. (1987) report data indicating that women with a diagnosis of PTSD are more likely to be diagnosed with Antisocial Personality Disorder. That the number of studies including both men and women is limited, is unfortunate because recent research indicates that men and women differ in the number of traumatic events they experience, the types of events they are likely to experience, and their responses to those events (Davidson & Smith, 1990; Kessler, Sonnega, Bromet, Hughes, & Nelson, 1995; Vrana & Lauterbach, 1994). In addition, the failure to include both men and women prevents researchers from examining potential interactions between sex and personality variables. The current study examined the relationship between various personality/ demographic variables and the number of traumas experienced and the severity of PTSD symptoms. It used a nonclinical sample of men and women experiencing a broad range of traumatic events. It was hypothesized that several personality variables (antisocial personality traits, borderline personality traits, and sensation seeking) would predict variance in trauma exposure. Antisocial personality is a complex group of symptoms consisting of an acting out component and a sensation seeking component. This study also examined if sensation seeking explains additional variance in trauma exposure beyond that explained by the acting out component. In addition, several interactions were tested. Specifically, would antisocial personality raise the risk for trauma exposure equally for men and women or would this effect be stronger for men. Conversely, would borderline personality raise the risk for trauma exposure equally for men and women or would this effect be stronger for women. It was also hypothesized that the intensity of the trauma would be the most potent predictor of the severity of posttraumatic stress symptoms. Once a trauma has been experienced, another group of variables (neuroticism and introversion) are hypothesized to predict variance in severity of posttraumatic stress symptoms. Several interactions were also tested. It was hypothesized that elevations in neuroticism and introversion would be associated with higher PTSD scores for men and women, but that this effect would be more potent for women. Method Participants The research participants were 402 undergraduate students (225 men and 177 women). The majority were in their freshman or sophomore year (78.6%). Approximately 91.3% reported that their racial heritage was European American, for 3.5%, it was Hispanic; for 3.2%, it was Asian; and for 2.0%, it was African American. The majority were single (93.3%), with considerably fewer persons

P1: FRC/GCR/GCX/GCZ

P2: GCO/GCP/GCQ

Journal of Traumatic Stress

PP047-293807

January 15, 2001

17:8

Style file version February 22, 1999

Correlates of Trauma Exposure and Responsiveness

33

reporting that they were living with their partner (3.7%), married (2.0%), or separated/divorced (1.0%). Measures The Traumatic Events Questionnaire (TEQ). The TEQ (Vrana & Lauterbach, 1994) assesses experiences with nine specific types of traumatic events (accidents, natural disasters, crime, child abuse, rape, adult abusive experiences, witnessing the death/mutilation of someone, being in a dangerous/life-threatening situation, receiving news of the unexpected or sudden death of a loved one) reported in the empirical literature as having the potential to elicit posttraumatic stress symptoms. In addition, two residual categories are included, allowing respondents to report any other very traumatic event not listed and events that are too traumatic to discuss. Table 1 lists the specific questions used to screen for trauma exposure. To reduce the likelihood of false negatives in reporting sexual assaults, the term rape was Table 1. Screening Questions Used on the Trauma Events Questionnairea Item No.

Question

1

Have you been in or witnessed a serious industrial, farm, or car accident, or a large fire or explosion? Have you been in a natural disaster such as a tornado, hurricane, flood or major earthquake? Have you been a victim of a violent crime such as rape, robbery, or assault? As a child, were you the victim of either physical or sexual abuse? As an adult, have you had any unwanted sexual experiences that involved the threat or use of force? As an adult, have you ever been in a relationship in which you were abused either physically or otherwise? Have you witnessed someone who was mutilated, seriously injured, or violently killed? Have you been in serious danger of losing your life or of being seriously injured? Have you received news of the mutilation, serious injury, or violent or unexpected death of someone close to you? Have you ever had any other very traumatic event like these? Have you had any experiences like these that you feel you can’t tell about (note: you don’t have to describe the event).

2 3 4 5 6 7 8 9 10 11

If you answered “Yes” to one or more of the questions above, which was the MOST traumatic thing to have happened to you? Fill in the number of the question (e.g., No. 2 for natural disaster). Did you answer Yes to more than one question above while thinking about the same event? Yes ❑ No ❑ If yes, which items refer to the same event? If you answered “No” to all questions, describe briefly the most traumatic thing to happen to you. a Persons

wishing to receive a copy of the TEQ can contact the first author.

P1: FRC/GCR/GCX/GCZ Journal of Traumatic Stress

34

P2: GCO/GCP/GCQ PP047-293807

January 15, 2001

17:8

Style file version February 22, 1999

Lauterbach and Vrana

not used. Instead, persons were asked to indicate whether they had any sexual experiences that involved the threat or use of force and were against their will. This questioning is consistent with the recommendations of previous investigators (Kilpatrick et al., 1989). To assess trauma intensity, respondents provide information for each event that was experienced, including (1) severity of injuries, (2) if they felt their life was in danger, (3) how traumatic the event was at the time, and (4) how traumatic the event is currently. Each item was measured on a 7-point Likert scale anchored by “Not at All” and “Severely/Extremely.” These four items were summed for the worst event experienced to produce a trauma intensity score ranging from 4 to 28. Finally, to obtain a trauma exposure score, the number of events the individual reports experiencing was summed. Personality Diagnostic Questionnaire-Revised (PDQ-R).5 The PDQ-R (Hyler & Rieder, 1987) is a 152-item true–false questionnaire and it was used to assess borderline personality. The diagnostic criteria contained in the PDQ-R precisely match those listed in the DSM-III-R. Respondents are asked to think about how they have tended to feel, think, and act over the last several years when responding to the individual items. A 3-week test-retest study (Hurt, Hyler, Frances, Clarkin, & Brent, 1984) yielded kappa values of .50 for borderline. The PDQ-R possess high sensitivity (typically greater than .80) and moderate specificity (.60–.90) for most personality disorders. The strongest relationship between clinicians’ diagnoses of personality disorder and the PDQ diagnosis was for borderline personality (κ = 0.46, r = .51) (Hyler et al., 1989). Gough Socialization Scale (So Scale). The So Scale (Gough & Peterson, 1952) of the California Psychological Inventory is a 54-item true–false measure that assesses four general domains of antisocial personality. This measure has been used extensively in research on antisocial personality (see Schalling, 1978 for a review) and was included as a measure of this personality trait. Zuckerman Sensation Seeking Scale (SSS). The SSS-Form V (Zuckerman, 1979) is a 40-item scale that was used to assess one component of antisocial personality. For each item, respondents are presented with two choices labeled “A” and “B” and are asked to “choose the one which better describes (their) likes or feelings.” The SSS correlates with measures of Antisocial Personality Disorder, such as the Psychopathic Deviant Scale of the MMPI (Zuckerman, Bone, Neary, Mangelsdorff, & Brustman, 1972). Eysenck Personality Questionnaire-Revised (EPQ-R). The EPQ-R is a 100item true–false questionnaire and was used to assess two personality traits, Neuroticism and Extroversion (Eysenck, 1967; 1982), that were expected to relate 5 The

present study incorporated two minor changes in the PDQ-R. Item 5b, which originally read “I hurt myself accidently,” was changed to read “I have attempted to harm myself” to more accurately reflect the question’s intent. To increase the range of scores within each personality type, the response format was changed from forced choice to a 1–7-Likert scale anchored by “Not at All” and “A Lot.”

P1: FRC/GCR/GCX/GCZ Journal of Traumatic Stress

P2: GCO/GCP/GCQ PP047-293807

January 15, 2001

Correlates of Trauma Exposure and Responsiveness

17:8

Style file version February 22, 1999

35

to reactivity to traumatic events. Reliability data on the revised version of the EPQ are favorable (Eysenck, Eysenck, & Barrett, 1985). Purdue PTSD Scale-Revised (PPTSD-R). The PPTSD-R (Lauterbach & Vrana, 1996) is a 17-item scale that assesses symptoms of Posttraumatic Stress Disorder, and was used as the trauma response measure.6 Each item of the PPTSD-R corresponds to one of the DSM-III-R PTSD symptoms. Respondents rate the frequency of occurrence of each symptom for the preceding month on a 1–5 scale anchored by “Not at all” and “often” with an intermediate rating labeled “sometimes.” The PPTSD-R is highly internally reliable (α = .91), its scores are stable over time (2-week test-retest r = .91), it can distinguish between persons who were and were not traumatized, and can distinguish between persons experiencing different types of traumas (Lauterbach & Vrana, 1996). Procedure The research participants were enrolled in an introductory psychology course and were required to participate in research projects as a part of their course requirement. They were provided with general information on all current experiments and were advised to select from among the available options. They were told that this study involved completing questionnaires asking about their experiences with various traumatic events and their reactions to those events. In addition, they were told that they would be completing several questionnaires asking about their personality. The participants first completed a demographics questionnaire followed by the four personality measures (So Scale, EPQ-R, SSS, and PDQ-R). They then completed the TEQ, assessing the presence/absence and severity of various traumatic events. Those experiencing multiple events judged which was the most traumatic, and those experiencing none of the events briefly described the most traumatic thing to ever happen to them. These latter persons were classified as experiencing “no event.” Participants then completed the measure of traumatic stress (PPTSD-R) based on the most traumatic event they had experienced. Data Screening The procedures outlined by Tabachnick and Fidell (1989) for screening data for multiple regression were closely followed. Data were tested for normal distribution of independent and dependent variables, univariate and multivariate outliers, and multicollinearity and singularity of independent variables. Analyses were run with and without appropriate transformations for normality and with and without 6 The

Civilian version of the Mississippi Scale was also administered. However, this scale appears to be more a measure of general distress than of posttraumatic symptoms (Lauterbach, Vrana, King, & King, 1997), and will not be discussed here.

P1: FRC/GCR/GCX/GCZ Journal of Traumatic Stress

P2: GCO/GCP/GCQ PP047-293807

January 15, 2001

17:8

36

Style file version February 22, 1999

Lauterbach and Vrana

outliers. All analyses produced identical results; therefore, for convenience and ease of interpretation, results will be presented on nontransformed data, with outliers included. To assist in interpreting the regressions, Table 2 lists the correlations between all variables. Results Prevalence of Trauma and PTSD About 80% (n = 322) of the participants reported experiencing at least one event of sufficient intensity to potentially elicit symptoms of PTSD. Multiple traumatic events were frequently reported in this sample. Forty percent (n = 162) reported experiencing 0–1 events, 34% (n = 174) reported experiencing 2–3 events, 20% (n = 50) reported experiencing 4–6 events, and 6% (n = 16) reported experiencing 7–12 events. To determine the prevalence of PTSD in this sample, symptom ratings on the PPTSD-R were used. Persons who were classified as PTSD positive were those who reported experiencing a significant traumatic event on the TEQ and who maximally endorsed one or more re-experiencing symptom, three or more avoidance symptoms, and two or more arousal symptoms. When these criteria were used, 3.2% (9 women, 4 men) would receive a presumptive PTSD diagnosis. A variety of trauma were reported by this group, including crime, childhood abuse, rape, being abused as an adult, being in a dangerous life-threatening situation, experiencing the violent or very unexpected death of a loved one, experiencing an event too difficult to discuss, and experiencing some other event. An additional 5% (14 women, 6 men) were one symptom short of the full diagnostic criteria. Relationship Between Personality Variables and Trauma Exposure The first analysis explored the relationship between several personality variables and trauma exposure (No. of events experienced). A simultaneous multiple regression was performed in which trauma exposure was regressed onto antisocial traits, borderline personality traits, and sex. In addition, the interaction terms7 between sex and each of these variables were added. The R for the regression was significantly different from zero, F(5, 396) = 10.72, p < .001. To obtain the final estimates, nonsignificant predictors were omitted and the analysis rerun with significant main effects, interactions, and components of interactions, F(4, 397) = 13.10, p < .001. Table 3 lists the results of the regression. The squared semipartial correlation (sr 2 ) is the unique contribution of each independent variable to the total variance of the dependent variable. These variables accounted 7 Interaction

terms were generated by computing the cross product of the scale score and the dummycoded variable.

.34∗∗ .134∗ .23∗∗ .26∗∗ −.23∗∗ .06 .51∗∗ .16∗∗ .09 .25∗∗ −.31∗∗ .19 .38∗∗ −.19∗∗ .02 −.09 .31∗∗ .03 .55∗∗ −.22∗∗ −.08 .39∗∗

−.54∗∗ .31∗∗ .38∗∗

−.50∗∗ −.32∗∗

.10

17:8

< .01. ∗∗ p < .001.

.15∗∗ −.03 .12∗∗ .23∗∗ .06 .12∗ −.16∗∗ .23∗∗

January 15, 2001

∗p

Trauma intensity Trauma exposure Extroversion Neuroticism Borderline personality Socialization Scale Sensation Seeking Scale Purdue PTSD Questionnaire

Trauma intensity Trauma exposure Extroversion Neuroticism Borderline personality Socialization Scale SSS

PP047-293807

Sex

Journal of Traumatic Stress

Table 2. Correlations Among Measures Assessing Trauma Intensity, Trauma Exposure, Exposure Personality Variables, and Response Personality Variables

P1: FRC/GCR/GCX/GCZ P2: GCO/GCP/GCQ Style file version February 22, 1999

Correlates of Trauma Exposure and Responsiveness 37

P1: FRC/GCR/GCX/GCZ Journal of Traumatic Stress

P2: GCO/GCP/GCQ PP047-293807

January 15, 2001

17:8

Style file version February 22, 1999

38

Lauterbach and Vrana Table 3. Simultaneous Multiple Regression of Personality Variables on Trauma Exposure Scores Variables Socialization Gender Borderline Borderline × Gender

B

SE B

β

sr2

−.10 1.43 .04 .04

.02 .82 .02 .02

−.26 .31 .19 .34

.05∗∗∗ .01 .01∗ .01∗

Note. Intercept = 4.41, R 2 = .12∗∗∗ , Adjusted R 2 = .11, R = .34. ∗ p < .05. ∗∗ p < .01. ∗∗∗ p < .001.

for 12% of the variance in trauma exposure. Antisocial traits (as assessed by the So scale) and the Borderline × Sex interaction both made a significant contribution to the regression equation. Examination of the bivariate correlations between socialization scores and trauma exposure (Table 2) reveals that persons lower in socialization (higher in antisocial traits) were more likely to experience one or more traumas. Examination of the regression lines for the Borderline × Sex interaction indicated that higher levels of borderline personality traits increase the risk of multiple traumatization for both men and women. However, this effect was more pronounced for men. These results provide support for the notion that certain personality variables, especially those reflecting antisocial traits, are related to experiencing multiple traumatic events. Antisocial personality is, however, a complex group of symptoms consisting of both an acting-out component and a less contentious thrill and adventure seeking component. Therefore, an alternative hypothesis is that sensation seeking may be the element of antisocial personality that is related to trauma exposure. To determine if sensation seeking contributes additional variance beyond that afforded by the exposure personality variables, a hierarchial multiple regression was performed. The significant exposure variables found earlier, and the components of significant interactions, were entered first, followed by scores on the Sensation Seeking Scale. If sensation seeking is related to trauma exposure, it should add predictive power after accounting for the other personality variables. The R for the regression was significantly different from zero, F(5, 396) = 10.55, p < .0001, R = .34. However, sensation seeking did not make a significant contribution to the prediction of trauma exposure scores after accounting for differences in antisocial traits (sr = .03). Therefore, it appears that it is the more hostile component of antisocial personality that is associated with multiple traumatization. Relationship Between Personality Variables and PTSD Symptomatology The next analysis examined whether personality variables are related to symptom severity after accounting for variance attributable to trauma intensity. A hierarchical multiple regression was performed in which PPTSD-R scores were

P1: FRC/GCR/GCX/GCZ Journal of Traumatic Stress

P2: GCO/GCP/GCQ PP047-293807

January 15, 2001

17:8

Style file version February 22, 1999

Correlates of Trauma Exposure and Responsiveness

39

Table 4. Hierarchical Multiple Regression of PTSD Scores on Trauma Intensity and Personality Variables Variables Block 1a Trauma intensity Total number of events Block 2b (N ) × (Trauma intensity) N

B

SE B

β

1.67 1.33

.12 .26

.43 .23

.10 −.48

.02 .24

.80 −.20

a Intercept = 12.5, R 2 = .31∗∗∗ , Adjusted b Intercept = 20.9, R 2 = .43∗∗∗ , Adjusted ∗ p < .05. ∗∗ p < .01. ∗∗∗ p < .001.

sr2 .16∗∗∗ .05∗∗∗ .02∗∗∗