Attachment and Posttraumatic Symptomatology Following Physical Injury Accidents

This study investigates the association between attachment and posttraumatic symptoms in adults immediately after a physical injury accident (assessed after 1 week and 3 months). We expect to find higher percentage of attachment dysregulation and higher symptomatic response (BSI and PDS) of disorganized individuals. We introduce Personal Experience as a representational marker for high attachment anxiety and preoccupation with one’s own distress that blurs and potentially dissolves the capacity to maintain selfother boundaries under stress. 65 patients were administered the Adult attachment Projective Picture System. Results show that attachment disorganization appears more frequent than in healthy individuals and is related to stronger symptomatology. Personal Experience showed good prediction for symptoms after 3 months. This could be important to identify individuals in need for crisis intervention support.

Moreover attachment theory and research have demonstrated that adult attachment insecurity is associated with poor emotion regulation strategies, chronic distress and PTSD symptoms, following severe traumatic events (e.g.Benoit et al., 2010;Kaninnen et al., 2003;Renaud, 2008).This study aims to find a better understanding of the relations between attachment and recovery/ adjustment following less severe traumatic events that might result in adjustment disorders but are unlikely to result in a PTSD.Kessler et al. (1995) reported that human manmade disaster (e.g.rape) is most likely to be associated with the development of a PTSD (50% of cases), whereas less than 5% of the individuals develop PTSD after experiencing a natural disaster (e.g.earthquake, accident).Beside this aspect, intensity of reaction in response to a traumatic event varies based on the individual's specific perception and affectivity and the ability to regulate emotions in the immediate aftermath (Yehuda, 2004).
Yehuda ( 2004) described posttraumatic stress reaction as a failure to recover from a specific event, rather than an emergence of a universal set of symptoms in response to it.Symptoms that immediately follow the exposure to the event were consequently considered as adaptive (Yehuda, 2004).So traumarelated pathology was defined as the persistent presence of symptoms after a fixed time period (e.g.four weeks).This view is supported by studies reporting that up to 94% of trauma survivors initially exhibit some degree of acute PTSD symptoms (Rothbaum & Foa, 1993;Vasterling, Daly, & Friedman, 2011).

Attachment representation, emotion regulation and trauma
Inner working models of attachment (secure or insecure) serve to regulate, interpret and predict attachment related behaviour, thoughts and feelings in attachment related situations.These mental working models remain relatively stable throughout the life span.Affective change in inner working models can be triggered when confidence in a secure base is shaken to the core, leading to a reconstruction or modification (Bretherton & Munholland, 2008).
In some individuals experiencing an accident, the inner working model of attachment might be affected and disorganized at least for a short period of time.These individuals may evaluate the event as threatening and their attachment system might be highly activated, leading to strong symptomatic reactions (i.e.Benoit et al., 2010).
Attachment research in infants and adults has also shown that insecure attachment representations in adulthood were associated with the use of less adaptive strategies to regulate emotions in stressful situations (Mikulincer & Shaver, 2008).The term "disorganized" refers to a state that reflects the inability to respond in a contained and adaptive manner in attachment related contexts in order to achieve an emotional state of security (Solomon & George, 2011).In adulthood, disorganized attachment has been labeled "unresolved with respect to loss or trauma".In contrast, secure attachment is considered to be a buffer from toxic levels of stress.Attachment insecurity and attachment disorganization have been demonstrated to increase vulnerability and constitute a risk factor for the development of psychopathology (e.g.Bakermans-Kranenburg & van IJzendoorn, 2009;Dozier, Stovall-McClough, & Albus, 2008;Fonagy et al., 1996;van IJzendoorn et al., 1997).

Attachment and PTSD
A recent meta-analysis reporting on the distribution of attachment patterns in clinical and non-clinical groups concluded that clinical subjects show highly more insecure and unresolved attachment representations than healthy individuals (Bakermans- Kranenburg & van IJzendoorn 2009).This review reported that 68% of PTSD and abused patients (n = 271 in total) were classified unresolved while only 14% were classified secure.Studies investigating the association between attachment and PTSD revealed that attachment disorganization (unresolved trauma) was related to high levels of distress and PTSD symptoms in the aftermath of a traumatic event in adulthood (e.g.rape, stillbirth, war veterans) (Harari et al., 2009;Hughes, Turton, Hopper, McGauley, & Fonagy, 2004;Stovall-McClough & Cloitre, 2006).
In sum, research on attachment and PTSD suggest that attachment representations are an important response mediator under severe trauma conditions such as war and human threats (e.g.Renaud, 2008).There is a higher risk of severe symptomatology following traumatic events for insecure and particularly disorganized individuals.Research to date has mainly examined attachment in relation to extreme situations, whereas few studies have focused on the relation between attachment, posttraumatic symptomatology and coping factors in individuals after less severe trauma conditions, such as physical injury accidents that are intense but not life threatening and do not involve the loss of significant others.These events are unlikely to be followed by a manifest PTSD but might by followed by a number of symptoms.
A study of Benoit et al. (2010), investigating the link between posttraumatic symptoms and attachment in emergency room patients who were recruited in hospitals and not diagnosed with PTSD, reported that 42% of their sample (n=36) was unresolved with respect to trauma.Using the Adult Attachment Projective Picture System (AAP, George, West & Pettem, 1999;George & West, 2012) attachment representations were assessed within 72 hours of exposure to a traumatic event (following DSM-IV-TR criteria).Using a security index to measure relative security, it could be shown that a higher attachment security rating was associated with significantly fewer PTSD symptoms in the one and three months following the exposure.Despite this innovative aspect, the security scale was not able to capture the degree of organization of attachment.
The role of dissociation in attachment is evident especially in attachment dysregulation (e. g.Liotti, 1999).Within posttraumatic symptomatology dissociation and self-other boundary confusion are crucial and are associated with attachment (Panksepp, 1998).The role of dissociation in attachment appears observable in narrative interviews, widely used to assess mental representations of attachment.Signs of self-other boundary confusion that are marked by a kind of reality shift in an attachment interview might be considered as a precursor of dissociation (George & West 2012).
A recent study on anxiety patients (Buchheim & George, 2011) revealed that AAP pictures portraying individuals as alone (i.e.images that portrayed only one individual in an attachment relevant situation) elicited a higher amount of traumatic Personal Experiences in their stories (e. g. "this is me in that picture, I am desperately alone like always in my life, this is very disturbing to look at for me").The authors suggest that this finding demonstrated self-other boundary confusion in anxiety patients; these patients were prone to become distracted and to shift their attention away from the task of responding to the hypothetical character of the story.Instead, they described their own traumatic life events.Their stories temporarily or permanently focused on personal biographical aspects.This study constituted an interesting approach to explore anxiety arousal (signs of self-other boundary confusion) by evaluating the emergence of Personal Experience material in the AAP.

Aim of the study
The present study investigates the association between attachment and posttraumatic symptomatology in adults immediately after a physical injury accident (less severe traumatic event) and after a recovery period of three months.The responses to threatening events such as accidents may explain, from an attachment perspective, how inpatients in emergency hospitals experience and recover from these events in dependence of their inner working model of attachment.

Hypotheses
In our study we expect (1) to find a higher percentage of unresolved trauma (similar to PTSD population samples) in inpatients immediately after the accident in comparison to healthy populations, (2) individuals with attachment dysregulation (unresolved trauma) compared to those with organized attachment would show stronger symptomatic responses (assessed by the BSI and PDS) immediately after the event and less of a decrease in symptoms during the three months period, and (3) that high anxiety arousal (assessed by signs of self-and other boundary confusion) would be associated with less symptom decrease and therefore with stronger symptomatic response (assessed by the BSI and PDS) after 3 months.

Method
Participants.The sample comprised of 65 patients, ranging from 18 to 76 years (M = 43.82,SD = 16.75).25 of the total sample were female (38.5%) without age differences from male participants (t (63) = 0.62, p = .54).Inclusion criteria were (1) inpatient hospitalization due to accident related physical injury with varying severity but without being life threatening, (2) mental and physical condition that allowed the participation in an interview conducted shortly after the accident, and (3) that the accident did not result in the loss of a significant other.Patients were recruited in an emergency hospital unit.The participation rate was 91%.Written consent was given by each individual.In the three months follow up 38 patients responded our request (58%) and remained in the study.
Procedure.All questionnaire measures and the AAP attachment interview assessment were administered within approximately 6 days (M = 5.91, SD = 1.32) of inpatient treatment.The AAP-interviews were administered by a trained psychologist in a quiet room in a hospital in a one to one setting.Interviewees were asked to fill in the questionnaires after the AAP procedure.They were also asked for written agreement to contact them for follow-up.Hospital files with information concerning the accident and degree of injury were also available to the researchers.3 months later (M = 108.81days, SD = 18.84) all participants who agreed (89%) were contacted by telephone.Questionnaire measures were sent and returned about 2 weeks later.Participants who didn't return their questionnaires were contacted and reminded again.The completion rate for the second assessment was 58% (N = 38).Ethics approvals were given by the local committee.

Measures.
Attachment Representation: All participants were administered the Adult Attachment Projective Picture System (AAP, George et al., 1999;George & West, 2012).The AAP is a reliable and valid measure of attachment representations for adults.The measure consists of eight drawings, including one neutral scene and seven attachment scenes.The picture set depicts events, which are expected to activate attachment, such as illness, separation, solitude, death, and threat, as determined by attachment theory (Bowlby, 1973).The interviewee is asked to describe what is happening in the picture, what led up to the scene, what the depicted persons are thinking or feeling, and what will happen next.The AAP coding system evaluates content and process elements of the attachment stimuli responses (see George & West, 2001; 2012 for a comprehensive description of the AAP coding and classification system).Transcripts are classified into one of the four standard adult attachment categories: secure (F), insecuredismissing (Ds), insecure-preoccupied (E) and unresolved (U).Organized attachment (F, Ds, E) is characterized by the capacity to contain dysregulated fear.Disorganized/Unresolved attachment is identified by attachment dysregulation characterized by the failure of integrated and functional representations of the self to contain or organize segregated systems.
Personal Experience is defined as the inclusion of personal life-experience memories while responding to the AAP stimulus.The AAP task asks individuals to tell a hypothetical story about the characters portrayed in the pictures; individuals are never asked to describe their own experience as an element of their response.Personal Experience is a representational marker for high attachment anxiety and preoccupation with one's own distress that blurs and potentially dissolves the capacity to maintain self-and other boundaries under stress (George & West, 2012).Personal Experience can be found in any of the classification groups, as the experience of heightened distress and self-other boundary blurring is a sign of high anxiety and is not a classification criterion.Psychometric properties of the AAP are excellent (George & West, 2001, 2004, 2012).The comparison with the Adult Attachment Interview (George et al., 1985, Main & Goldwyn, 1985-1996), an established measure to assess attachment representation (see George & West, 2012;Buchheim & George, 2011) showed high convergent validity for the four major attachment groups (F, Ds, E, U) of 90% (kappa = .84,p <.000).Convergent agreement for two group (organized (F, Ds, E) vs. disorganized (U)) classifications was 97% (kappa = .88,p <.000).In the present study two independent certified reliable judges double-coded 20% (n = 13) of the AAPs of the total sample (n = 65).There was a 92% classification group agreement in the four attachment categories (kappa = 0.88).Personal Experience is coded as either present or absent per story.100% agreement for coding personal experience was achieved in our study by two independent judges.
Posttraumatic Stress Symptomatology.Acute (time 1, assessed while hospitalized) and long-term (time 2, 3 months follow up) impact of the (traumatic) event was measured using the German version of two self-report measures.The Brief Symptom Inventory (BSI -a 54 items short form of the SCL-90) (Franke, 2000) provides the level of subjective impairment caused by somatic and psychic symptoms including scores for somatization, obsessivecompulsive thoughts or actions, interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety, paranoid ideation and psychoticism.Internal consistency for the 9 primary symptom dimensions is between alpha=.71and alpha=.85.Test-retest reliability for the 3 Global Indices is between .80 and .90(Derogatis & Melistaratos, 1983).We used the Global Severity Index (GSI) as a general measure of symptomatic distress, which combines information on the number of symptoms and the intensity of perceived distress.
The Posttraumatic Stress Diagnostic Scale (PDS, Foa, Cashman, Jaycox, & Perry, 1995) consists of 49 items, indicating potentially traumatizing events experienced by the participant and is used to identify symptoms of posttraumatic stress disorders.The participants rate their response to the indicated event (e. g. severe accident, abuse etc.) at the time of its occurrence along the DSM-IV criteria.The PDS includes a symptom severity score, calculated as the sum of the individual's responses.Symptom severity scores range from 0 to 51.The cut offs for symptom severity rating are 0 = no rating, 1-10 = mild, 11-20 = moderate, 21-35 = moderate to severe and>36 = severe.Alpha is .92for Total Symptom Severity (Foa et al., 1997)

Results
Out of the entire sample, 19 patients (29.2%) sustained a traffic accident, 35 (53.8%) a recreational accident (e.g., skiing or sports related injury), and 11 (16.9%) had a work related accident.Accident type and degree of injury were not related (χ 2 (4, N = 65) = 2.94, p = .56).We proceeded a follow-up assessment (time 2) after 3 months (M = 108.81days, SD = 18.84), in which 38 patients (58%) remained in the study.No significant age-difference between females (35.8%) and males occurred (t (63) = 1.25, p = .22).Responders and non-responders did not differ significantly in any of the variables: gender, age, attachment or symptomatology at time 2. Table 1 summarizes the symptom intensity of our sample at T1 (GSI and PDS global scores).We first hypothesized that individuals would show a higher percentage of unresolved trauma (similar to PTSD population samples), immediately after the accident, in comparison to healthy populations.The 4-way distribution of our sample showed the following results: 10.8% were classified secure (F), 15.4% dismissing (Ds), 30.8% preoccupied and 43.1% unresolved (U).In order to evaluate our data in comparison to distributions in the literature, we examined the frequency distribution of two-way attachment classifications (organized versus disorganized) in our sample in comparison to healthy populations and PTSD/abuse samples reported by the meta-analysis of Bakermans-Kranenburg & van IJzendoorn (2009), and to the sample from Benoit et al. (2010).We calculated an exact binominal test.Our sample showed significantly fewer disorganized individuals than PTSD samples, but more than in the healthy samples.This result demonstrated that there is an impact of the experience of a physical injury accident on the immediate organization of the inner working model of attachment.Results are summarized in table 2. Our second hypothesis predicted that individuals with attachment dysregulation (unresolved trauma) would show stronger symptomatic response immediately after the event and less decrease in symptom intensity within 3 months.The results in table 1 demonstrated an overall symptom intensity on a mild to moderate level.To test our second hypothesis, a 2 x 2 mixed analyses of variance with attachment organization (versus disorganization) as betweenfactor and time as within-factor were conducted to evaluate differences on (A) general distress (GSI-score) and (B) symptom intensity (PDS-score) shortly after the accident (time 1) and after 3 months (time 2).To explore cross sectional differences we calculated one-way t-tests (as planned contrasts are not recommended in mixed designs).To control for multiple testing significance level was adjusted for the two dependent measures (Bonferroni correction; α = .025).Means and standard deviations are reported in table 3.
We can conclude, that disorganized patients with respect to attachment suffer from slightly higher symptom intensity immediately after the accident but not after three months.Our third hypothesis stated, that individuals with high anxiety arousal and difficulty modulating attachment distress as evidenced in Personal Experience material in the AAP response, would show higher levels in symptomatology (GSI and PDS intensity score) that would not dissipate over the time of 3 months after the accident.We added the total number of stimulus responses in which personal experience occurred (scores ranged from 0 to 3).At time 1, 43.1% of our participants scored above 0 in the Personal Experience score (PE).The average amount of PE was about 1 (M = 0.94; SD = 1.36).
Beside the main effect of time (decrease within 3 months) in the GSI score (F (3,37) = 15.36,p = .001)and in the PDS score (F (3,35) = 5.18, p = .02)we found -according to our hypothesis -significant interaction of time X Personal Experience score in our data for the GSI score F (3,36) = 7.62, p = .01),and for the PDS score (F (3,36) = 6.63, p = .01). Figure 2 shows decrease in individuals with Personal Experience versus those without Personal Experience.At time 2 differences were not significant (t (36) = 0.67, p = .51)for the BSI scores but for the PDS scores (t (36) = 2.76, p = .02).The results show, that patients with higher anxiety arousal show less decrease in symptomatology.

Discussion
Trauma is defined as an overwhelming experience, exceeding the individual mental capacities and is usually associated with threatening and anxious feelings.This leads to support seeking, flight or dissociative behaviour, depending on what is (unconsciously) estimated as most successful for the individual to re-establish an emotional state of security and control.This subjective evaluation depends on several factors including the quality and intensity of the threat, subjective expectations based on former experiences, the availability of social support and the activation of the attachment system.Inner working models of Attachment are considered to play a crucial role in this process (see e.g.Fizke et al., 2013).In our study we explored the meaning of attachment disorganization in less severe events and introduced Personal Experience (shifts from hypothetical story to reality) as a marker of the capacity to maintain self-and other boundaries under stress.This might be considered as a precursor of dissociation (see also Juen et al., 2013) The comparison of the distribution of attachment classifications in our study with reported distributions in healthy and risk samples, revealed that the amount of individuals classified as disorganized in our sample was significantly higher than in healthy populations and comparable to other clinical risk samples recruited in hospitals (Benoit et al., 2010).This is beside the fact that sample characteristics might be different in the compared samples as attachment distribution is reported to be rather independent of age or gender (Bakermans-Kranenburg & van IJzendoorn, 2009).
Some individuals might show a disorganized inner working model of attachment even before the exposure.As mentioned in the introduction this group might already bring in a pre-traumatic risk for a PTSD development.
A major limitation of our study concerns the lack of information about the individual's attachment classification before the accident, but the assessment of attachment before the event would need a prospective design, which is nearly impossible to plan.Our suggestion for explaining the amount of disorganization is that in some individuals experiencing an accident, the inner working model of attachment is affected and disorganized, as the confidence in a secure base is shaken to the core at least for a short period of time.The question arises how Global Severity Index (GSI) Symptom Intensity (PDS) this can happen as internal working models usually are rooted at their core in procedural memory.We are not able to answer this question with this study but these individuals seem to evaluate the event as more threatening (maybe based on contents within procedural memory) and their attachment system is highly activated and overwhelmed, leading to strong symptomatic reactions.
Our data suggest that the exposure to the accident in some individuals might trigger an affective change in the inner working models leading to disorganization.We assume that the inner working model of the 'triggered group' mentioned above, might re-establish their usual functioning without stronger support after a period of time, whereas it does not do so in a group already disorganized beforehand.It might be a highly interesting future research question to differentiate these groups.Nevertheless, we have some evidence with our first results, that individuals who remain organized with respect to their attachment after stressful events tend to be less vulnerable for clinical trauma related symptoms especially when they do not show signs of self-others boundary confusion.
Attachment disorganization seems to be a rather unspecific predictor in less severe events in contrast to severely traumatic events.We introduced a promising methodological approach to identify individuals with respect to their self-and other boundaries under stress, assessed by Personal Experience in the AAP.The idea is based on the assumption that the impact of accidents is not strong enough for the majority of individuals to show clear dissociative behaviour.Although dissociation is low in our sample as shown in our PDS data, we consider the breakdown of the capacity to maintain self-and other boundaries under stress as a mild form of dissociation (see also Juen et al., 2013).The lack of an organized inner working model of attachment immediately after the event and signs of self-others boundary confusion seem related to symptom intensity after 3 months especially in trauma specific symptomatology, assessed by the PDS.To be explored in future studies we suggest that the assessment of these individual resilience factors immediately after the event-even when no clear signs of a PTSD risk are given-might be a way to capture vulnerability in crisis intervention in order to promote selective support.
Our study is limited by some methodological problems.As mentioned before, we do not have information on the attachment status before the accident.This problem is difficult to solve, as a screening of a large population in a prospective design means high effort and costs.Moreover, measuring attachment 3 months after the event would bring additional information on the recovery but it was unfortunately not possible to interview our participants again.We had a rather high amount of non-responders in our sample.After three months participants usually didn't show physical impairments from the accident and their homes were spread so that no personal contact beside telephone was possible.The sample size at T2 makes it difficult to detect differences between groups so especially our T2 results are preliminary.In future studies we should add more detailed socio demographic information as some data are potentially confounding.
Finally, we should enlarge our sample (and include a control group) and compare our data with other samples with diverse traumatic experiences in order to examine the role of intensity of trauma and inner working models of attachment to be able to cope with attachment relevant events.In this study we offer a first step in exploring the meaning of attachment and related aspect in the recovery from less severe traumatic events without life threatening and without the loss of a significant other.Overall individuals seem to be able to recover from such events without big effort but there is a risk group with problems in recovering without showing strong signs of a manifest PTSD diagnosis.In a further study we intend to identify these individuals and assess attachment related aspects in order to improve and support their recovery.

Figure 1 .
Figure 1.Severity Index (GSI) and Symptom Intensity (PDS) for organized and disorganized individuals at time 1 and time 2

Figure 2 .
Figure 2. Severity Index (GSI) and Symptom Intensity (PDS) for individuals showing Personal Experience at time 1 and time 2

Table 1 .
Symptom intensity of the sample at T1 (GSI and PDS global scores).

Table 2 .
Relative frequencies of attachment organization/disorganization in the compared sample ** p<0,001

Table 3 .
Severity Index (GSI) and Symptom Intensity (PDS) for organized and disorganized individuals at time 1 and time 2