2. Female Dissociative Responding to Extreme Sexual Violence in a Chronic Crisis Setting: The Case of Eastern Congo
2.1 Abstract
The present cross-sectional study aimed to examine relationships between the number of traumatizing events, degree of shutdown dissociation, posttraumatic stress disorder (PTSD), and depression. Fifty-three female survivors of the ongoing war in Eastern Congo, who sought medical treatment, were interviewed. A path-analytic model was created with paths to PTSD via shutdown dissociation, the number of self-experienced and witnessed traumatizing event types. Cumulative trauma exposure and shutdown dissociation were associated with increased PTSD severity. PTSD and witnessing predicted depression when depression was modelled as a consequence of PTSD. Moreover PTSD mediated the correlation between dissociation and depression. The findings emphasize the power of shutdown dissociation in predicting PTSD and suggest differential effects of threat to oneself as opposed to witnessing trauma.
2.2 Introduction
Dissociative responding during and in the aftermath of trauma exposure has been ob-served in survivors who were exposed to serious threat with a proximity to danger, such as sexual assaults (Marx, Forsyth & Lexington, 2008). Persistent dissociation has been shown to be a powerful predictor of posttraumatic stress disorder (PTSD) (Briere, Scott & Weathers, 2005). In the present study, dissociation is regarded as a defence that typically appears in response to bodily proximity to danger. It includes partial or complete functional sensory shutdown, classified as negative dissociative symptoms (van der Hart, Nijenhuis & Steele, 2005). From an evolutionary perspective these shutdown dissociation might be adaptive in life-threatening conditions in which neither flight nor fight are viable options for survival (Bracha, 2004). The parasympathetically dominated shutdown of sensory and functional systems reoccurs in confrontation with intrusions, minor stressors or during trauma exposure therapy (for a detailed description of this concept see Schauer & Elbert, 2010).
PTSD commonly co-occurs with depression. In the National Comorbidity Survey, Kessler Sonnega, Bromet, Hughes, and Nelson (1995) found that approximately half of the patients with PTSD had a comorbid major depression. Almost 80% of those patients reported that depression symptoms followed PTSD. In contrast, patients who reported depression prior to trauma exposure reported a higher PTSD symptom severity in response to a single
event, challenging the direction of comorbidity (Shalev et al., 1998). There is a documented relationship between the number of traumatizing events to which an individual has been ex-posed and the likelihood of developing PTSD (Neuner et al., 2004).
In sum, PTSD, depression, dissociation and the number of traumatizing events seem to be intercorrelated. The present study utilized a path-analytic approach to disentangle par-tial relationships, modelling depression as a consequence of PTSD. A sample of women from a war-torn area was chosen because the exposure to traumatic stressors is variable, and rape necessarily implies maximal proximity to the perpetrator, and thus would favour later shutdown dissociation (Schauer & Elbert, 2010).
2.3 Method
2.3.1 Settings and Procedure
Respondents were survivors of the ongoing organised violence in the Eastern Congo who sought medical treatment for gynaecological problems secondary to sexual violence (va-ginal infection or Fistula) at one of the health centres (Panzi Hospital, SOSAME psychiatric centre in Bukavu and health centre Bunjakiri). Six clinical psychologists and one psychiatric nurse from the University of Konstanz carried out interviews between January and March 2009. Prior to assessment the interviewers, who all had prior work experiences in crisis re-gions had been specifically trained.
Respondents were informed about the goal of the study and that they would not re-ceive any financial remuneration for their participation. Respondents were randomly selected from a list of survivors of organised violence provided by the hospitals’ stuff. Everyone who was asked agreed and gave the informed consent. Under-aged participants as well as their legal representatives also signed the form. The University of Konstanz’s ethics committee approved the protocol. Our group trained the personnel of the respective clinical units to provide trauma-focussed treatment. Respondents were referred to these experts and further treatment was supervised.
2.3.2 Instruments
The structured interviews lasted approximately 2.5 hours and were carried out in Eng-lish with the help of trained local translators in Swaheli or Mashi.
To assess shutdown dissociation we used a newly developed 12-item questionnaire based on our concept of shutdown dissociation (Schauer & Elbert, 2010), adapted to the
re-Female Dissociative Responding 25
spective cultural context. We asked for the frequency of fainting, transitory deafness, changed acoustic perception, transitory blindness, changed visual perception, numbness, tran-sitory paralysis, analgesia, tension, feeling of nausea, out-of-body experiences, and inability to speak. Responses to all items were given on a scale ranging from 0 = not at all, 1 = once a week or less, 2 = 2-4 times a week, to 3 = 5 or more times a week. Factor analysis provided evidence for unidimensionality. The questionnaire showed sufficient internal consistency;
Cronbach’s alpha = .75.
To assess the trauma exposure we used a newly developed checklist consisting of traumatizing events with focus on different experiences of sexual violence. We assessed the number of different traumatic event types that were self-experienced and the number of trau-matic event types that were witnessed. A trautrau-matic event type was judged as self-experienced if the participant was the victim and as witnessed if the participant had observed the traumatic event while someone else was threatened.
For the PTSD diagnosis and score, we used the PSS-I (PTSD-Symptom-Scale-Interview), which follows the DSM-IV. The PSS-I has high reliability and validity and is comparable to more complex instruments for diagnosing PTSD (Foa & Tolin, 2000). In order to assess depression symptoms, we used the depression part of the Hopkins-Symptom-Checklist-25 (HSCL-25). This scale demonstrated satisfactory psychometric properties (Glass, Allan, Uhlenhut, Kimball & Borinstein, 1978). Apart from the shutdown dissociation questionnaire, all protocols were tested for their usability and validity in resource-poor set-tings (Ertl et al., 2010). Moreover, we had extensive discussions about the instruments with local psychologists and medical doctors prior to the assessment.
2.3.3 Data Analysis
Analyses were performed using SPSS 17.0; alpha level was set at .05 and two-sided t-tests were used to analyse significance. Scores with missing data were partially excluded from further analysis (n = 3). We used forced-entry linear regression analysis to evaluate direct and indirect effects on PTSD and depression. In order to constrain the paths in the model, the directions were predetermined by the theoretical assumptions and temporal pre-cedence. Therefore shutdown dissociation, the number of self-experienced and witnessed traumatizing events were considered antecedent to PTSD. Again, shutdown dissociation, the number of self-experienced and witnessed traumatic event types and PTSD, were modelled to predict depression. Associations among variables were taken into account. This model
pos-its that depressive symptoms in this sample arise primarily as a result of trauma exposure and as a consequence of PTSD.
2.3.4 Participants
The age of the 53 female respondents ranged from 14 to 56 years (mean = 30.0, SD = 11.3). Twenty-one of these survivors were currently married, 10 divorced, 4 widowed and 18 were single (never married). On average, they had two children (range 0 to 11). The educa-tional level varied widely among respondents with a range of 0 to 12 years of education com-pleted. Twenty-nine of the women were illiterate.
2.4 Results
On average, the respondents had experienced 6.9 (range 2 to 15) traumatic events and witnessed 5.5 traumatic events (range 0 to 11). The following self-experienced events were reported most frequently: sexual assault (96.2%), natural disaster (62.3%) and abduction (60.4%). The majority witnessed physical assaults (73.6%), 69.8% witnessed armed assault and 62.3% homicide. The time elapsed since the worst event was variable (Mdn = 3 years;
IQR = 4). Thirty-seven respondents fulfilled all criteria for PTSD (for diagnosis criteria see Foa, Riggs, Dancu & Rothbaum, 1993). Approximately 67% of respondents with and 32%
of those without PTSD met the DSM-IV criteria for depression (Bolton, Neugebauer &
Ndogoni, 2002). The reported PTSD severity (M = 21.40, SD = 8.81) and HSCL-depression score (M = 2.51, SD = 0.64; n = 51) were high. On average, respondents reported a mean sum-score of shutdown dissociation of 8.48 (SD = 6.31).
Guided by theoretical assumptions and previous empirical research on the effects of cumulative trauma and dissociation, we built a path-model to uncover potential associations between the variables. Linear regression coefficients were used as path-coefficients. Shut-down dissociation, the number of self-experienced and witnessed traumatizing events were modelled as being correlated and having both direct and indirect influence on PTSD and de-pression. Figure 2.1 illustrates the results, with standardized linear regression coefficients.
Model-I indicated two paths to PTSD. Shutdown dissociation and the number of self-experienced traumatic event types influenced PTSD severity. Sufficient power was obtained for the shutdown dissociation, whereas the power of the number of self-experienced event types turned out to be less than .8 (compare Cohen, 1992). The number of witnessed event types was indirectly related to PTSD via the correlation with the number of self-experienced
Female Dissociative Responding 27
traumatizing events (r = .53, p < .001). Table 2.1 provides linear regression coefficients, and respective significance levels.
Model-II proposes two pathways to depression. One path included the PTSD se-verity. Moreover, PTSD served as a mediator between dissociation and depression. The sec-ond path to depression comprised the number of witnessed traumatizing event types. In this model, both variables achieved satisfactory power. Taken together, the analysis revealed an adjusted R2 value of .42 in model-I and .43 in model-II.
Table 2.1. Summary of Simultaneous Regression Analyses
Table 2.1a
Summary of Simultaneous Regression Analysis for Variables Predicting PTSD Symptom Severity (N=52)
Model 1 B SE B ß t
constant 1.62 3.59 0.45
shutdown dissociation 0.76 0.15 .56 5.01***
number of traumatic event types self-experienced
0.80 0.37 .28 2.19*
number of traumatic event types witnessed
-0.28 0.38 -.09 -0.74
Note. Unstandardized (B) and standardized (ß) regression coefficients. *p < .05, **p < .001.
Table 2.1b
Summary of Simultaneous Regression Analysis for Variables Predicting Depression Score (N=50)
Model 2 B SE B ß t
constant 1.19 0.27 4.39***
shutdown dissociation 0.01 0.01 .11 0.80
number of traumatic event types self-experienced
-0.06 0.03 -.26 -1.97
number of traumatic event types witnessed
0.08 0.03 .37 2.95*
PTSD symptom score 0.05 0.01 .63 4.47***
Note. Unstandardized (B) and standardized (ß) regression coefficients. *p < .05, **p < .001.
Figure 2.1: This figure presents a path analytic model for developing PTSD and depression symptoms after traumatizing events and demonstrates one possible model among others. The values in the model are standardi-zed linear regression coefficients. In order to constrain the number of paths in the model, effects between vari-ables that did not reach conventional levels of significance (p > .05) were deleted.
2.5 Discussion
We used a path-analytic approach to examine relationships between the exposure to traumatizing events, shutdown dissociation, PTSD and depression, modelling (for this sam-ple) depression as a consequence of PTSD. Consistent with many other studies, the number of self-experienced traumatizing events significantly predicted the severity of PTSD (Neuner et al., 2004). However, prior studies did not consider witnessed and self-experienced trau-matizing events together in a composite model. Surprisingly, witnessing the trauma of others had no direct effect in our model, but seemed to be associated with PTSD via the number of self-experienced traumatizing events.
The second path to PTSD leads through shutdown dissociation. These results are in line with those of Briere and co-workers (2005), who reported that dissociation in the after-math of the trauma predicts PTSD. Additionally it is consistent with the findings of Nijen-huis, van der Hart, Kruger and Steele (2004), who found a high correlation between somato-form dissociation and PTSD. In addition to the effect of cumulative exposure to traumatizing events, participants who suffered more from shutdown dissociation reported a higher PTSD score. On the one hand, PTSD symptoms can co-occur with shutdown dissociation. Re-experiencing the trauma could evoke the same defensive responses as during the trauma, and reported dissociative responding seems to be more common in participants who experienced
Female Dissociative Responding 29
sexual assaults (Griffin, Resick & Mechanic, 1997). On the other hand, it has been argued that shutdown dissociation may be conditioned as a defensive response to cope with stressful PTSD symptoms (Schauer & Elbert, 2010).
PTSD symptom severity had the highest common variance with depression. This finding argues for co-morbidity. In predicting depression, witnessing traumatizing events seems to play an essential role. Many studies have found a positive relationship between depression and trauma exposure (Monroe & Reid, 2009).
Methodological limitations of this study include the limited sample size. However, even when this is taken into account, shutdown dissociation was still a powerful predictor of PTSD. These results were obtained for female survivors of a war-torn area, who experienced extreme sexual violence and may be, in part specific for this setting. Moreover, the cross-sectional as well as retrospective nature of this study limits attributions about the direction of causality and this path-model represents one possibility for interpreting these results and prospective studies are necessary to provide further support for this outcome. Finally, the newly developed structured interview that we used to measure shutdown dissociation has not yet been validated and inter-reliability wasn’t examined.
We conclude that present observations emphasize the relevance of shutdown dissoci-ation for predicting PTSD. Moreover, the findings suggest differential effects of threat to oneself as opposed to witnessing trauma.
2.6 Acknowledgements
We thank the respondents who participated in the study with great effort and openness. The authors thank the translators Aline Iragi, Benedict Bashimbe, Francine Nsindabagoma, Priska Cinogerwo and Roger Buhendwa from Bukavu and Anna Mädl, Anett Pfeiffer, Elisabeth Schauer, Heike Riedke and Nina Winkler for supporting the interviews. Funded by the Deutsche Forschungsgemeinschaft (DFG) and the European Refugee Fund.