Abstract
Traumatic experiences and dissociative symptoms seem to be intrinsically related. There are two main, not mutually exclusive, theories about this relationship. One theory proposes that some people have the capacity to dissociate and will use this capacity to ward off the impact of horrible experiences [1]. In this view, dissociation is a coping mechanism with negative side-effects, because non-integrated traumatic experiences – that is, aversive sensorimotor and highly affectively charged experiences – tend to intrude consciousness. According to Nijenhuis et al. [2], these intrusions relate to rudimentary or more complex dissociative emotional parts of the personality. The emotional parts manifest themselves in dissociative symptoms such as dissociative flashbacks, nightmares and re-experiencing traumatizing events. The emotional parts can intrude into the consciousness of those parts of the personality dedicated to functioning in daily life, or take over executive control of consciousness and behaviour for some time. As for those aspects associated with functioning in daily life, survivors use dissociative skills to avoid traumatic memories and the emotional parts – manifesting in symptoms such as dissociative amnesia, depersonalization, emotional and bodily numbing – but these dissociative skills are fallible. In other theories [2–4], dissociation arises because severe stress can interfere with normal integrative mental processes, notably when the individual's integrative capacity is limited because of factors such as immaturity of the brain and prior stress exposure. Because of a lack of integration, cues that are saliently related to the traumatizing event will elicit different psychobiological reaction patterns for different dissociative parts of the personality [5]. For emotional parts, these cues almost automatically trigger more or less complete traumatic memories, involving a lack of inhibition. However, parts dedicated to functioning in daily life inhibit emotional reactivity to trauma-related cues, but are depersonalized and may have incomplete declarative memories of the traumatizing event(s). Posttraumatic stress disorder (PTSD) and acute stress disorder (ASD) are categorized as anxiety and clearly stress-related disorders in the DSM-IV, but there are sufficient arguments to consider them as essentially dissociative in nature [6]. This perspective is supported by theories that describe traumatic memories of PTSD patients as dissociated imprints of the sensory and affective elements of traumatic experiences [7]. The associated visual, olfactory, affective, auditory and kinaesthetic experiences can all be categorized as somatoform dissociative symptoms.
In line with these theories, higher levels of dissociation are reported in groups of traumatized individuals compared to non-traumatized control groups [8]. Chronicity and severity of trauma were also found to predict the level of dissociation in abused children [9]. Furthermore, an association between traumatization and dissociation is supported bymoderate correlations between ameasure of traumatization and dissociation in groups of traumatized individuals [10]. In a student sample of 312 patients, the correlation between self-reported physical abuse and dissociative symptoms was 0.18, and between self-reported sexual abuse and dissociative symptoms 0.21 [11]. Combining data from 26 studies on patients and non-patients in a meta-analysis resulted in a correlation of 0.25 between a measure of physical or sexual abuse and dissociation [12]. The modest size of these correlations is not surprising, given the difficulty in capturing the severity and impact of traumatization reliably in an index. Furthermore, correlations are a measure of linear association between an index of traumatization and a measure of dissociation. A model that captures the notion that a certain threshold has to be passed before dissociation occursmight be more realistic [Nijenhuis et al.: unpublished data].
Apart from this issue of the reliability of an index of the severity of a traumatization, there are other factors that may influence the magnitude of the correlation between traumatization and dissociation, notably in non-clinical samples. Several investigators noted that we can expect a restricted range of dissociative symptoms in student samples [10], [13] that may restrict the magnitude of the correlation coefficient. The same argument applies to the measurement of traumatization. Because non-students will generally be older, their risk of having experienced potentially traumatizing events will be greater. They will therefore probably report these events more often and are also more likely to endorse different types of such events.
A conceptual flaw in most studies on the relationship between traumatization and dissociation is that dissociation is investigated by measuring the presence of dissociative phenomena by the Dissociative Experiences Scale (DES) [14] or the Dissociation Questionnaire (DISQ) [15]. These instruments capture psychoform dissociative phenomena such as amnesia, loss of control, identity confusion and fragmentation, and absorption [15]. For decades, researchers focused on psychoform phenomena and only recently has attention been redirected to the somatoform manifestations of dissociation that were described by French psychiatrists in the nineteenth century such as Janet, and by twentieth century psychiatrists such as Myers and Rivers, who studied World War I soldiers [6], [16]. The renewed interest in somatoform dissociation has resulted in a new instrument, the Somatoform Dissociation Questionnaire (SDQ-20), which evaluates the severity of a range of somatoform dissociative symptoms, such as analgesia, anaesthesia, motor disturbances, alternating preferences of tastes and smells, pain and loss of consciousness [17]. In patients with conversion disorder, those who reported multiple types of traumatization had higher scores on the SDQ-20, but not on the DES, indicating a relationship between the severity of reported traumatization and the level of somatoform dissociative symptoms [18]. An assessment of both types of dissociation may give a more complete picture of the relationship between traumatization and dissociation.
Research on predisposing personality characteristics has focused on absorption, but has also incorporated fantasy proneness and hypnotizability. Absorption might be a risk factor for the development of dissociative symptoms, but evidence for this hypothesis is lacking to date [19]. Absorption is usually defined as a disposition for having episodes of ‘total’ attention that fully engage one's representational resources [20]. Fantasy proneness is a related concept but refers to the extent to which an individual displays a personal history of intense involvement in imaginative activities (Wilson and Barber, 1981 in [21]). Whereas absorption is essentially interactive, triggered by external events, one might consider fantasy proneness to be more ‘self-involved’ and stimulated by internal as well as external sources [21].
Fantasizing may actually serve a purpose in traumatized patients when they use it as a means to escape from reality. From such a point of view, fantasizing can be considered a coping mechanism [22]. Merckelbach and Muris do not comprehend dissociation as related to highly stressful experience, but suggest a radically different perspective [23]. They propose that fantasy proneness can give rise to both self-reports of traumatizing events and of dissociative experiences and thus might explain the relationship between these events and symptoms. Merckelbach and Muris' finding that there is a correlation between fantasy proneness, as measured by the Creative Experiences Questionnaire (CEQ) and dissociation, as measured by the DES [14] is consistent with this hypothesis. However, it also fits the hypothesis of fantasy as a coping skill regarding trauma among individualswho also dissociate, and DES scores in a normal population may assess absorption rather than true dissociation. As Merckelbach and Muris' study did not include a correlation between traumatization and fantasy proneness, and did not distinguish between absorption and true dissociation, a direct test on the influence of fantasy proneness on the relationship between traumatization and dissociative symptoms is still lacking. Moreover, causality obviously cannot be deduced from correlational relationships.
The primary aim of this study is to assess the relationship between a self-report measure of traumatization and psychoform dissociation as well as somatoform dissociation in a non-clinical population, while accounting for the influence of fantasy proneness. A secondary aim is to compare the extent to which students and non-student adults report potentially traumatizing events (for brevity, henceforth described as traumatizing events) and dissociative symptoms.
Method
Participants
The student sample consisted of 23 men and 50 women. The age range was from 17 to 29 years (mean=21.8 years, SD=2.3). The students received €5 or course credits. Students were recruited through an advertisement in our university magazine and through flyers distributed throughout the campus. The non-student sample consisted of 88 men and 59 women. Their age range was from 19 to 77 years (mean=46.4 years, SD=14.5). The non-students were recruited by inviting a random sample from residents of the city of Nijmegen to participate in an experimental study on suggestibility. As part of the study, participants were asked to fill out some questionnaires. About one in five subjects who received a letter of invitation agreed to participate and received a €5 gift voucher.
Measures
Psychoform dissociation
Psychoform dissociation was measured with the DES [14]. The DES is a 28-item self-report questionnaire that requires the patients to indicate to what extent presented statements apply to them. The DES measures psychoform dissociative phenomena including amnesia, loss of control, identity confusion and fragmentation, and absorption. The statements include, for example, the experiences of having done something without knowing when and how, or of having been somewhere without knowing how the place was reached. This widely used screening instrument for dissociative symptoms in clinical samples was found to have good reliability and clinical validity.
Somatoform dissociation
Somatoform dissociation was measured with the SDQ-20 [17]. The SDQ-20 is a 20-item questionnaire thatmeasures analgesia, anaesthesia, motor disturbances, alternating preferences of tastes and smells, pain and loss of consciousness. Five-point scales are used to indicate to what degree presented statements apply. Statements include: ‘It sometimes happens that I feel pain while urinating’ and ‘It sometimes happens to me that I grow stiff for a while’. The total score ranges from 20 to 100. The reliability of the scale is high and the construct validity is good [24].
Traumatization
Traumatization was assessed using the initial version of the Traumatic Experiences Checklist (TEC) [25]. Recently, good psychometric properties for a slightly extended version of the TEC were reported [26]. The TEC assesses various forms of potentially traumatizing events. The TEC total score involves the endorsed number of potentially traumatizing events (0–25). Trauma area composite scores (range: 0–12) can be calculated for emotional neglect, emotional abuse, physical abuse, sexual harassment and sexual abuse. A score of 1 is added to the incidence score for each of the following aspects: the duration of the traumatic experience, the relation to the person causing it, the indicated severity of the experience and the person's age at the start and end of the experience. A trauma area composite total score results from summing up these five subscores (0–60).
Absorption
The Tellegen Absorption Scale (TAS) [20] was used to measure absorption. Absorption is considered to be a reliable index of fantasizing [27]. The TAS consists of 34 contentions that require the patients to indicate whether presented statements apply to them.
Data analysis
The restriction of range was tested using Levene's test for the equality of variances. To enable comparison with previous studies, this test was performed on both raw scores and transformed scores. A test on differences between samples on the scores on the questionnaires TEC, DES and SDQ-20 was performed using the Mann–Whitney U-test on raw scores. As scores on TEC, DES and SDQ-20, all deviated significantly from a normal distribution they were all transformed before calculating correlations between these measures. Traumatic Experiences Checklist scores were raised by 1 and subsequently transformed using the natural logarithm. Dissociative Experiences Scale scores were also log-transformed. Somatoform Dissociation Questionnaire scores were transformed by taking the square root of the score minus 20. Differences between the correlations of the two samples were tested after application of Fisher's r to z transformation.
Results
Traumatic experiences
Levene's test for the equality of variances indicated that the range of the number of types of traumatic experiences was greater in adults than in students (F(72,146) =10.91, p=0.001). Similarly, the range of the indices of physical abuse (F(72,146) =3.90, p=0.05) of sexual harassment (F(72,146) =17.85, p<0.001) and of the trauma area total composite score of the TEC (F(72,146) =5.25, p=0.023) was greater in adults than in students. These ranges were indeed restricted in the student sample. Variances of emotional neglect, emotional abuse and sexual abuse did not differ between samples. Transformation of TEC total scores successfully removed these differences in range, as Levene's test on the transformed scores was not significant.
Table 1 presents the various aspects of reported traumatizing events in the two samples. The mean number of types of traumatizing events as measured by the TEC is equally large in non-students as in students, 3.25 versus 2.27, respectively (z=−1.706, p=0.088).
Means and standard deviations of scores for the Traumatic Experiences Checklist (TEC), Dissociative Experiences Scale (DES) and Somatoform Dissociation Questionnaire (SDQ-20) for students and non-students
In students, 79.5% reported events that are included in one of the five categories of the TEC, whereas in non-students this proportion is 85%. The trauma area composite scores of the other categories did not differ significantly from each other. The proportion of students reporting emotional neglect did not differ significantly from the proportion of non-students, 19.2% compared with 28.6%. Similar proportions of students and non-students reported emotional abuse, 18.6% and 28.6%, respectively. Physical abuse was reported by 12.3% of the students and 15.0% of the non-students. Although 12.3% of the students and 20.4% of the non-students reported sexual harassment, this difference was not significant. Sexual abusewas reported by 5.5% of the students and 8.2% of the non-students.
Dissociative experiences
The variance of psychoform dissociative phenomena as measured by the DES was significantly smaller in adults than in students (F(72,146) =5.36, p=0.02). The variance of somatoform dissociative phenomena as measured by the SDQ-20 did not differ between the two samples. Students had higher DES scores compared to non-students and higher SDQ-20 scores. This difference between samples remained intact after transformation of DES and SDQ-20 scores. A score on the DES of 25 is the recommended cut-off score in the screening for DSM-IV dissociative disorder [28]. Among the students there were four individuals with such a high score, and among the non-student sample, five individuals. Similarly, a cut-off score of 30 is recommended for the SDQ-20 [29]. Five students and seven non-students displayed this level. Only two non-students displayed scores above the cut-off level on both DES and SDQ-20.
Relationship between traumatic experiences and dissociation
As shown in Table 2, there were significant zero-order Pearson correlations between the total score of the TEC, that is, the number of types of traumatizing events and psychoform dissociation in both samples. This correlationwas significantly lower in adults than in students (p<0.001). When the effect of absorption was partialled out, this correlation only remained significant in the student sample. Again, the correlation was significantly higher in students than in adults. The zero-order Pearson correlation between the total score of the TEC and somatoform dissociationwas also significant in both samples and did not differ significantly between samples. The correlation remained significant in both samples after partialling out the effect of absorption. Again, no significant difference between the magnitude of the correlation in the samples was observed.
Correlations in the two samples between traumatic experiences and cognitive and somatoform dissociation, adjusted for absorption
Discussion
In line with many other studies, the findings of the current study indicate that a reported history of traumatization is to some extent related to psychoform dissociation as measured by the DES in two non-clinical samples. The study also documents an association between reported traumatization and somatoform dissociation, as measured by the SDQ-20, in these samples. However, the magnitude of these correlations and the number of statistically significant correlations decreased when controlling for absorption as an index of fantasy proneness.
Because most studies on non-clinical patients gathered data from students, such studies probably suffer from a restricted range in dissociative phenomena and traumatic experiences [10]. In our study, the range of the number of reported traumatizing events is indeed significantly smaller in students than in normal adults. In contrast, the range of psychoform dissociation is higher in students than in adults, while the range of somatoform dissociation did not differ. One therefore needs to be cautious in generalizing results from studies of traumatization in student samples to the general population.
The absolute number of traumatizing events did not differ significantly between samples. Although people have a risk of experiencing a potentially traumatic event every day, some of the experiences that are captured in the TEC pertain to experiences which are generally seen before the age of 18 years. This applies specifically to emotional neglect, emotional abuse and physical abuse. Therefore, the total score will only increase in few people after they have turned 18 years. Students reported, however, more psychoform dissociation and more somatoform dissociation than non-students. This finding suggests that younger people dissociate more. Another possibility is that these figures reflect that people get over dissociative symptomatology with age, with or without professional help. Only a longitudinal study could, however, provide conclusive data on this topic.
The relationship between traumatization and the two measures of dissociation differs between the samples. In adults, the relation between reported traumatization and psychoform dissociation is significantly weaker than in students. The correlation of reported traumatization and somatoform dissociation is of similar magnitude in the two samples. This difference can most likely be attributed to the differences between the two samples, the most prominent distinction being age, followed by gender. For the patients in the non-student sample, probably more time has elapsed since the reported events happened than for the students.
We used the TAS as a ‘close cousin’ [27] of fantasy proneness. How exactly absorption and fantasy proneness are related, and how they relate to distorted memories is not completely clear. A plausible hypothesis that is suggested in the recovered memory debate states that absorption is a capacity that fosters fantasizing, which in turn leads to distorted memories. However, a strong argument against this hypothesis comes from an experimental study showing that individual differences in memory accuracy for autobiographical events were significantly related to absorption but not to fantasy proneness [30]. The measure of fantasy proneness in this study was the Inventory of Childhood Memories and Imaginings, a measure of involvement in fantasy, both as a child and as an adult that was also used to construct the Creative Experiences Questionnaire (CEQ) [27]. The authors speculate that those who score high on absorption are more inclined to incorporate post-event information of any sort into the memory reconstruction process [30]. This tendency resembles the capacity to focus more on internal experiences, while at the same time relatively neglecting external events that is seen in experimental research in highly hypnotizable individuals [31]. These findings suggest that highly absorbed individuals can have difficulty to distinguish internal and external events while memorizing experiences and that, therefore, information can be easily mixed with internal events and not corrected by external events. In this view, absorption may foster errors in details of a memory, which is completely different from making up false memories that have no relation with reality at all. A more accurate model of the relation between absorption and fantasy proneness and distortion in memories will therefore contain a pathway from absorption to fantasizing and a second, separate pathway from absorption to memory distortion.
Absorption is also correlated with dissociation [12]. The instrument that we used to measure psychoform dissociation, the DES, contains a subscale that measures absorption-like phenomena that are commonly seen in mentally healthy people and that manifest in a wide range of mental disorders. Taxometric studies have indicated that it is useful to distinguish between ‘normal’ and ‘pathological’ dissociation. Individuals in the ‘pathological’ dissociative class (taxon) can be identified with a brief, eight-item questionnaire, called the DES-T [32]. The remaining items may not measure true dissociation – defined as a lack of integration of mental and behavioural phenomena that an individual experiences and memorizes – but measure absorption-like phenomena, that is, alterations of consciousness. These alterations are conceptually different from dissociative phenomena [33]. A factor analysis on DES scores of a non-clinical sample suggested a structure of only one dimension of dissociation [34], but it could be that true dissociation is uncommon in the normal population. For these reasons, we have partialled out absorption from the relation between traumatic experiences and dissociation.
As with most other studies, our study is cross-sectional in nature and therefore does not allow any conclusions with respect to a causal relationship between reported traumatization and dissociation. However, the study documents that when the relationship between reported traumatization and psychoform dissociation is partially mediated by absorption in both non-clinical samples, the relationship between reported traumatization and somatoform dissociation in non-clinical adults is not a function of absorption. The findings suggest that the SDQ-20 may be a more reliable indicator of dissociative phenomena than the DES. In a normal population, psychoform dissociation as measured by the DES mainly seems to coincide with absorption.
Several studies of psychiatric patients have indicated that a history of traumatization does not necessarily lead to lasting dissociative symptomatology [35]. Our data support the notion that many people who experience a potentially traumatizing event are able to deal with it and will not develop dissociative symptoms. Most likely, dissociation is particularly associated with specific types of overwhelming events, notably a threat to one's body from another person, and with unfavourable combinations of unnerving events, features of the exposed individual (e.g. young age, prior history and insecure attachment) and characteristics of the situation (e.g. lack of support) [36].
Footnotes
Acknowledgements
The authors thank J. Verspaandonk and E. Geensen for their efforts in the data collection for this project. Näring presented parts of this study at the 24th European Conference on Psychosomatic Research in 2002 in Lisbon and at the VIII European Conference on Traumatic Stress in Berlin in 2003.