A Socio-interpersonal Perspective on PTSD: The Case for Environments and Interpersonal Processes

Transcription

1 Clinical Psychology and Psychotherapy Clin. Psychol. Psychother. 20, (2013) Published online 22 June 2012 in Wiley Online Library (wileyonlinelibrary.com). DOI: /cpp.1805 A Socio-interpersonal Perspective on PTSD: The Case for Environments and Interpersonal Processes Andreas Maercker* and Andrea B. Horn Department of Psychology, University of Zurich, Zurich, Switzerland Post-traumatic stress disorder (PTSD) is a common reaction to traumatic experiences. We propose a sociointerpersonal model of PTSD that complements existing models of post-traumatic memory processes or neurobiological changes. The model adds an interpersonal perspective to explain responses to traumatic stress. The framework draws from lifespan psychology, cultural psychology and research into close relationships and groups. Additionally, clinical knowledge about PTSD is incorporated. This involves knowledge about shame, guilt, estrangement feelings and protective factors, such as social support and forgiveness. Three levels are proposed at which relevant interpersonal processes can be situated and should be adequately researched. First, the individual level comprises social affective states, such as shame, guilt, anger and feelings of revenge. Second, at the close relationship level, social support, negative exchange (ostracism and blaming the victim), disclosure and empathy are proposed as dyadic processes relevant to PTSD research and treatment. Third, the distant social level represents culture and society, in which the collectivistic nature of trauma, perceived injustice, and social acknowledgement are concepts that predict the response trajectories to traumatic stress. Research by the current authors and others is cited in an effort to promote future investigation based on the current model. Methodological implications, such as multi-level data analyses, and clinical implications, such as the need for couple, community or larger-level societal interventions, are both outlined. Copyright 2012 John Wiley & Sons, Ltd. Key Practitioner Message: The socio-interpersonal model proposes an interpersonal view of the processes that occur in the aftermath of a traumatic experience. At the individual level, the model integrates the social affective phenomena that clinical research identifies in PTSD patients, including shame, guilt, anger, revenge and the urges or reluctance to disclose. At the level of close relationships, there is an emphasis on the role of the individuals partner, family or social support in the development or maintenance of PTSD and its recovery. At the distant social level, societal and cultural factors, e.g., individualistic versus collectivistic or other human value orientations, are acknowledged as contributing to the severity and course of PTSD. Increasing attention should be given to new approaches of PTSD treatment that refer to an interpersonal view of PTSD, e.g., communication training, PTSD-specific couples therapy or community programs. Keywords: Post-traumatic Stress Disorder, Interpersonal Processes, Social Context, Disclosure, Social Sharing Post-traumatic stress disorder (PTSD) is triggered by life-threatening events or other dramatic events, such as experiences of violence, accidents or disasters. Although the traumatic experience is commonly considered the main cause, etiological research into PTSD has identified a large number of biological and psychosocial factors that contribute to the development and maintenance of this disorder. Indeed, an important question underlying *Correspondence to: Andreas Maercker, Department of Psychology, Division of Psychopathology and Clinical Intervention, University of Zurich; Binzmuhlestr. 14/17, CH 8050 Zurich. maercker@psychologie.uzh.ch all research in this area is why this disorder does not develop in every individual who experiences a trauma. In the last decade, theoretical models of traumatic stress, in particular psychological trauma models (Brewin, 2001; Ehlers & Clark, 2000; Foa & Kozak, 1986; Rubin et al., 2008) and psychobiological models (Heim & Nemeroff, 2009; Yehuda, 2006), have made a fundamental contribution to our understanding of this disorder. Moreover, they have also sown the seeds for other areas of psychological research, such as memory, emotion and lifespan developmental research (e.g., Cicchetti & Walker, 2001; Conway & Pleydell-Pearce, 2000; Damasio, 2000). These research areas are integrated with, and applied to, PTSD research in the current paper. Copyright 2012 John Wiley & Sons, Ltd.

2 466 A. Maercker and A. B. Horn Lifespan developmental psychologists widely agree on the transactional or contextual models of psychological functioning. Individual factors interact with contextual factors over time and constitute risk and/or protective factors for the development of a disorder or psychological health (Bronfenbrenner, 1979; Cicchetti & Walker, 2001). According to these models, PTSD would be considered a disorder that can be facilitated or precluded by contextual factors. In contrast to lifespan developmental psychology, these views on contextual factors are still not prevalent within clinical psychology. Indeed, the area s perspective may be characterized by the idiom, the lone man against the elements. However, interpersonal and socio-ecological aspects are known to be highly relevant risk and protective factors in the aftermath of a trauma. Meanwhile, contextual approaches and changes in developmental and interpersonal processes over time provide promising and heuristically inspiring avenues for further research in the area of PTSD. The lack of conceptual coverage of social and interpersonal factors in most current PTSD models is even more surprising. Indeed, these very factors explained a large amount of variance in two milestone meta-analyses that explored the predictors of PTSD. These meta-analyses noted that the measures of social support were among the best predictors of PTSD (Brewin et al., 2000; Ozer et al., 2003). Both analyses implied that the social interpersonal realm comprises powerful mechanisms that can either prevent or cause PTSD-related suffering in the individuals. Moreover, in clinical work with PTSD patients, the patients struggles with shattered views of others, the world and with themselves all prominently relate to contexts and interactions. In summary, there is pronounced empirical and clinical evidence that indicates the importance of socio-ecological and interpersonal processes. It is worthwhile to examine concepts and results from research about interpersonal processes and contextual influences to build a better understanding of these processes. For this purpose, an integrative model is proposed below. This is designed to integrate contextual and interpersonal aspects into a scientifically up-to-date, clinically inspiring view of PTSD. INTERPERSONAL ASPECTS OF EXISTING POST-TRAUMATIC STRESS DISORDER MODELS: A SHORT OVERVIEW In current PTSD research, most existing psychological models of the disorder have arisen in the context of a particular type of therapeutic intervention, and most of these models also contain interpersonal aspects. In the following section, the interpersonal aspects of the classical models are highlighted. This overview relates perspectives that are either conceptually isolated or neglected. This is due to the focus being upon intrapersonal representations in the classical models described in the trauma literature. These findings will be contextualized in the new sociointerpersonal perspective proposed in this paper. Horowitz (1976) was the first researcher to describe the now generally accepted PTSD core symptom model. This comprises intrusions and avoidance together with typical schematic post-traumatic cognitive changes, such as shame and guilt. Shame and guilt are so-called social affects affective reactions that do not exist without a social reference. In this model, shame and guilt refer to the threat to the self and the person s moral attitudes. More recently, Horowitz (2007) highlighted revenge as being a composite of social affects that include anger at perpetrators, frustration about the injustices of the world and a belief that no rescuer can be trusted. According to Janoff-Bulman s model on traumatization (Janoff-Bulman, 1985), trauma survivors show interpersonal changes (shattered trust), as well as typical post-traumatic changes in general beliefs about the world s orderliness, meaningfulness and benevolence. These models reflect the influence of the more distant social reference systems, such as social groups or culture. Foa and colleagues (Foa & Kozak, 1986; Foa et al., 1992) proposed their emotional processing model of trauma. This centres on fear excesses that manifest following the trauma and mainly describes hyperactivated associative memory representations. The original version of the model accounts for post-traumatic changes in the views about the self, others and the world. Such emphasis has become more accentuated due to recent developments in assessment and treatment (Foa et al., 1999b). The well-established Post-traumatic Cognitions Scale, for instance, assesses negative cognitions about the world and self-blame. Self-blame refers to feelings of shame and guilt. As with the models listed above, these models examine social affective reactions in the response to a trauma at the level of the individual and interpersonal representations of both the closer and more distant social contexts. In treatment studies, Foa and Rauch (2004) observed improvement in the core PTSD symptoms as well as improvement in interpersonal representations in most of the cases. In their influential model, Ehlers and Clark (2000) postulated dysfunctional cognitions, particularities of the trauma memory and persistent anxiety as being the three central etiological components of development and maintenance of PTSD. The dysfunctional cognitions suggested in this model included intrapersonal representations of perceived, ubiquitous danger in close and distant social contexts, anger in response to unfairness and feelings of shame. Additionally, the pivotal maladaptive safety behaviours listed in the Ehlers and Clark model are partly behavioural responses, including dysfunctional interaction within close relationships (e.g., withdrawal from others). Many other models have been developed, mostly with a focus on optimizing therapeutic strategies (e.g., Brewin, 2001; Resick

3 Socio-interpersonal Perspective on PTSD & Schnicke, 1992; van der Kolk, 2006; Rubin et al., 2008). However, the role of interpersonal factors in those models is not qualitatively different from the previously discussed models. Apart from mistrust, shame, guilt, anger and beliefs about the world as being unfair, no further social and contextual processes were considered in these models. Intrapersonal cognitive and affective processes are indisputably important to the establishment and maintenance of PTSD in the individual. The questionable element within these models is the mostly exclusive focus on the inner world of the patient. Conversely, in the classic models, there is an implicit position, whereby the patient is viewed as living isolated on a deserted island. The common, individual-centred perspective might lead to the conclusion that the individual s well-being is solely the product of his or her own cognitive representations of the world. Admittedly, most would agree that it would be clinically imprudent to assume that the reactions of proxies, peers and society are of no influence to the individual s symptoms. Accordingly, empirical findings, such as the strong association between social support and symptoms after the experience of a trauma, also indicate the importance of social contexts (Brewin et al., 2000;Ozeretal.,2003).Additionally,theclassicmodels mostly neglect interactional aspects, such as the partner s or other relatives reactions to the trauma and the PTSD. Recent findings from different perspectives support a new conceptualization of current PTSD models, with a more pronounced focus on interactional and contextual components. Investigating childhood trauma, Charuvastra and Cloitre (2008) reviewed research findings that pointed to the importance of attachment behaviours for short-term and long-term outcomes of trauma. They consistently identified perceptions of social support, both before and after a traumatic event, as an important factor in determining the subsequent vulnerability to the development of PTSD. Cloitre et al. (2006) have developed a treatment manual specifically for the group of victims of childhood abuse that suffer from developmental disturbances in their attachment behaviours. This manual includes the training of interpersonal skills and thus has a particular focus on the current social reality of the patient. Furthermore, the authors report that clinical work with couples led to the development of treatment models that focused on couple-specific effects of trauma on the individual and dyadic level. The Couples Adaptation to Traumatic Stress model (Goff & Smith, 2005) and the cognitive-behavioural interpersonal theory of PTSD (Monson et al., 2010) are prominent examples of important contributions to the research on interpersonal aspects in PTSD. In the latter model, close relationship problems are regarded as the main agent of dysfunctional adaptation to trauma. However, the models mainly focus on couple relationships and do not include the more distant social environment, such as cultural and societal influences. Beyond the PTSD models that have been developed according to different therapeutic approaches, current PTSD research examines the conceptualizations of resilience and post-traumatic growth (e.g., Bonanno et al., 2006; Calhoun Tedeschi, 2006; Zoellner & Maercker, 2006). Indications of the importance of social and interpersonal factors have frequently emerged in this research area, such as in the context of the terrorist attacks of 11 September 2001 (Bonanno et al., 2006; Hobfoll et al., 2006) or community disasters (Norris et al., 2008; Kaniasty & Norris, 2008). It has been shown that collectively experienced trauma is generally processed more easily than individually experienced trauma. This and other research findings from related areas, which will be discussed later in this article, are strong evidence for the importance of considering social context. Consequently, there is a need for a model of PTSD that integrates the role of social contexts in its description of the emergence of this disorder. The current model is designed to extend our view of PTSD from a solely intrapersonal individualistic perspective towards an integration of social, contextual and interactive processes. GENERAL STRUCTURE OF THE SOCIO- INTERPERSONAL FRAMEWORK MODEL OF POST-TRAUMATIC STRESS DISORDER 467 This model offers a framework for research of the contextual and interpersonal processes that occur in the aftermath of a traumatic experience. In the proposed model, the individual is nested in different levels of social contexts. We propose three levels to structure the contextual impacts on and interactions with the individual. These levels are based on concepts from social psychological research. The first level is the individual level, which consists of intrapersonal features or impairments. This level is consistent with previously mentioned intrapersonal models of PTSD and can be integrated with those models. We also include an interpersonal focus at this level by taking into account social affective phenomena, such as shame, guilt, anger and social cognitions. In sum, these are the thoughts and feelings that relate to the social reality of the patient. The second level comprises the context of close relationships. Close relationships are constituted by a high level of psychological intimacy. Typically, in adulthood, this would be the romantic partner, close family members and friendships. Including this level offers a framework for trauma-relevant processes that occur during interaction with the people who share a close relationship with the victim. Trauma-relevant processes, such as disclosure, social support or negative exchange, require interaction with close relations and are thus interactive phenomena. The third level comprises processes that occur at more social distances and is termed distant social level. Its

4 468 A. Maercker and A. B. Horn processes are defined by the factors that reflect the cultural and societal influences on the individual processing the trauma. This level represents belonging to a certain culture, religion or society with a certain justice or health system based on a shared cultural value system. In contrast to the second level where interactions happen among acquainted individuals, the interaction is between the group and the trauma victim. In social psychology, it is suggested that processes regarding group membership have a different quality than interactional processes between psychologically close individuals. Furthermore, clinically it seems evident that belonging to a certain social group (e.g., veterans and political prisoners) can be an important feature in therapy. This is distinct from the issue as to whether the trauma victim is, e.g., in a satisfying relationship. Thus, the model proposes the acknowledgment of interpersonal and social processes, without downplaying the importance of intra-individual processes in the development of PTSD (see above). The current model s new contribution to the literature is the perspective that the individual is nested within different contexts, with all of which the individual interacts. These interactions are highly relevant to the individual s chances of developing PTSD, and their effects are predicted according to different principles than those that are presented in the intrapersonal, individualistic models. In other domains of psychology, such as social or developmental psychology, this interpersonal logic has been conceptualized and researched. Following a contextual view (cf. Bronfenbrenner, 1979), the model suggests transactional or mutual relationships between the different levels. For example, intrapersonal reactions of shame on experiencing rape might have different impact on symptom extend if the rape victim is in a romantic relationship that allows disclosure of shamerelated content. Furthermore, both the reactions of shame and the disclosure within the couple, regarding traumarelated content, will occur differently, depending on the degree to which sexual assaults are taboo in the society in which the couple belongs. Methodologically, the hierarchical structure of these mentioned influences suggests a multi-level approach that allows taking into account characteristics of each level (see more detailed below). Furthermore, cross-level interactions can be assessed in that high levels of disclosure within the couple might buffer the negative impact of pronounced shame reactions. Accordingly, we propose an integrative approach that benefits from both the heuristic value of the contextual view and the methodological approaches that have already been introduced. Moreover, this new approach also has implications for clinical use, a point that will be discussed in more depth at the end of this article. Figure 1 illustrates the different levels and variables that are relevant to post-traumatic processes. The suggested examples of processes in Figure 1 situated at the different levels are related to the research described in the following paragraphs, although some of these processes can happen on different levels. For instance, disclosing trauma-related content to the partner would be allocated to level 2, whereas discussing traumarelated issues in a public setting from a support group to a formalized truth commission should be allocated transformed Traumatic experiences interpersonal (man made) or accidental shaped Distant social contexts: Culture & society Collective experience of trauma Societal acknowledgment / Injustice Cultural value orientations received Close social relationships Disclosure Social support / negative exchange Empathy induced perceived Individual: Social affective response Shame Guilt Anger Revenge provided Outcomes Individual Symptoms Well being Close relationships quantity quality Distant social contexts Social integration Societal segregation Figure 1. The socio-interpersonal model

5 Socio-interpersonal Perspective on PTSD to level 3. As a further example, social ostracism can take place in direct interaction but also in interaction with a whole group, e.g., excommunication in Catholicism. Socio-interpersonal processes reflect interaction of the individual with others. As the term interaction suggests, the model also includes corresponding perspectives on each level, and these comparisons constitute a third dimension (see the spatial dimension in Figure 1). The competing perspectives at the individual level, perceived versus induced, provide an outcome of either selfperceived guilt in the survivor or guilty feelings induced in the survivors close relations. At the close relationship level, the competing terms are perceived or provided, which are well documented in social support research (Maisel & Gable, 2009). At the distant social level, the terms shaped versus transformed reflect the two possible directionsofactiononthislevel.theperspectivepairingsateach level of the model add an additional dimension to the focus on the interactional nature of the involved processes. The third dimension of the model opens the space of observation and data assessments of corroborating data from sources other than the direct trauma survivors. Recent criticisms of self-report data note that self-report results can be limited by the reporting individual s retrospection biases and response sets (Fahrenberg et al., 2007; Baumeister et al., 2007). These concerns should be taken seriously. In PTSD research, only a few reported responses at the individual level have been accompanied by corroborating investigation of their particular reference antipodes or peers, such as sleep disorders (Germain et al., 2006) and post-traumatic growth (Zoellner & Maercker, 2006). 1 Defining Outcomes The model suggests the outcome variables that characterize each socio-ecological level. Clearly, symptoms or dysfunctions at the individual level are the focus of clinical psychology. Additionally, the social affective reactions that are 1 The interactional nature of interpersonal processes indicates the need for integration of the multiple perspectives. However, although conceptually plausible, the methodological demands of researching interactive processes in an adequate manner were often too challenging to be successfully achieved. In recent years, research on close relationships has developed important new methodologies in response to this challenge. The Actor Partner Interdependence Model for dyads (APIM, Kenny & Cook, 1999) and the Social Relations Model for groups (SRM, Cook & Dreyer, 1984) both offer methodological tools to capture multiple perspectives in social interactions. These models include the possibility of mutual interdependencies. The recent paradigm shifts in the research of social support (e.g., Bolger, Zuckerman & Kessler, 2000) are good examples of the benefits of including multiple perspectives in interpersonal processes research. Therefore, we include an interactional perspective at each level of contexts within the current model. 469 outlined in detail in the model are related to well-being and successful coping. At the level of close relationships, relationship duration or the cohabitant status of couples represents important variables at the dyadic level, in that the measurement accounts for both members of the dyad. Furthermore, relationship satisfaction and psychological intimacy are important outcome variables that reflect the social reality regarding the individuals close relationships. For instance, recent studies about couples in the PTSD literature have included relationship satisfaction as an outcome and have analysed whether satisfaction moderates individual symptom levels (Fredman et al., 2010; Lev-Wiesel & Amir, 2001; Nelson & Goff, 2005;). One of these studies showed an inverse effect of partner satisfaction. It was found that traumatization does not necessarily lead to psychopathology but indeed that it can even lead to higher relationship satisfaction and thus be observed as a protective factor. Fredman et al. (2010) found that some aspects of the trauma s magnitude (e.g., extent of loss) were positively associated with relationship satisfaction, whereas other aspects (e.g., extent of threat) were not. This finding illustrates support for including outcome variables, not only at the level of individual symptoms but also at the level of close relationships. Similarly, the number of individuals belonging to the survivors social network or group is an important reflection of interpersonal functioning at the dyad and group level. At the third level of culture and society, members of a region affected by a natural disaster may show altered social integration or segregation parameters (e.g., the New Orleans region after Hurricane Katrina). It is also worthwhile to examine markers of societal coherence and value systems. All of these variables describe the contexts in which the individual is nested. On the basis of theoretical considerations and empirical findings, we propose that the characteristics of each level are relevant in the development of PTSD. Such multi-level analyses allow researchers to account for the contextual dependencies of the nested individuals. These analyses may lead to clinically relevant findings about important context characteristics that could alter the adaption trajectory that occurs after a traumatic experience. Of particular interest is the interaction of those markers across levels, from the individual to close relationships and further to distal group membership. In terms of methodology, multi-level models permit researchers to study multi-level phenomena as such. In a multi-level model, the individual being researched can be considered nested within different contexts, and each of these contexts and attributes can be modelled at a separate level, e.g., individual, close relationship or group. Admittedly, most current methodologies still lack a broad variety of tools for the measurement of critical contextual factors. Accordingly, the forthcoming examples of the relevant aspects from different contextual levels mainly focus on

6 470 A. Maercker and A. B. Horn the impact on and perceptions of the individual. This focus is reasonable in so far as the targets of the research are the post-traumatic symptoms of the individual. EMPIRICAL EVIDENCE FOR THE MODEL AND ITS LEVELS The First Level: Social Affective Changes Post-traumatic stress disorder research has shown that many different affective reactions play a crucial role in the response to a traumatic experience. Affective reactions that relate to other persons, groups or communities are termed social affects (Hareli & Parkinson, 2008). The affective reactions that possess social connotations are of particular importance to the current model because they focus on the interplay between the individual and his or her environment. The pronounced impact of interpersonal traumatic experiences on social affects has long been noted. Even accidental trauma, such as injuries or disasters, may lead to these social affective reactions (Horowitz, 1976; Janoff- Bulman, 1985). A whole array of social affects is reported in the aftermath of a trauma, including shame, anger, disgust, guilt and vengefulness. Social affect is necessarily characterized by a representation of the mental states of others (e.g., of the perpetrators in the case of interpersonal trauma), a point that has yielded increasing interest in social affective neuroscience (Burnett et al., 2008). Accordingly, a social affective reaction is considered a mental state that refers to both the self and others (e.g., Orth et al., 2010; Tracy, Robins, & Tagney, 2007). Recent studies of traumatic stress provide increasing evidence for the new proposal that affective responses should be included as symptom criteria for PTSD in the ICD-11 or DSM-5 (Hathaway et al., 2010). On the basis of these new approaches to PTSD treatment, this array of affects was consequently added to the PTSD treatment manuals, and so adjusting the previously solitary focus on posttraumatic fear and anxiety responses (Rizvi et al., 2009; Zöllner et al., 2005). The first description of PTSD cited guilt as one of the defining symptoms of the disorder (Horowitz, 1976). The term survivor guilt (Niederland, 1961) was particularly appropriate for the Holocaust survivors. Such individuals frequently provided self-referential statements, such as, Why did I survive the Holocaust and not the others/my friend/my wife? and Had I done... back then, he/she would still be alive. Guilt is generally considered to function for its own immediate relief. However, chronic guilty feelings contribute to the maintenance of psychopathology in the long run (Rachman, 1993). Feelings of shame are also frequently experienced by PTSD patients and in particular occur following humiliating or sexually violent traumas. The functionality of shame in the context of PTSD is still unclear. Shame has been demonstrated to be strongly related to intrapersonal avoidance (Street et al., 2005). Furthermore, it is plausible that shame is related to social withdrawal, but that possibility has not yet been investigated. Shame-related and guilt-related cognitions are captured together in the commonly used Post-traumatic Cognitions Inventory (Foa et al., 1999b), where they regularly predict PTSD symptoms. Anger phenomena are frequent in the context of traumatic stress. Traumatized persons indicate signs of anger, rage, hatred and revenge towards the persons, groups or institutions that caused or did not prevent the trauma. Compared with non-traumatized persons, persons with PTSD generally show increased levels of all facets of anger (Orth & Wieland, 2006). The temporal sequence of the core PTSD symptoms and anger facets was unclear until Orth et al. (2008) examined whether PTSD symptoms predicted anger or whether anger predicted PTSD symptoms. Their re-analysis of comprehensive US and German data sets showed that PTSD symptoms predicted the subsequent extent of anger, whereas anger did not predict the subsequent symptoms of PTSD. In emotion research, anger is seen as one of the social affects that are called the hostility triad (Izard, 1971). The other two social affects of the hostility triad are disgust and contempt. All three affects involve not only negative evaluations but also hostile tendencies towards another person. Hostility, mostly in the context of anger, has been researched in the PTSD literature and has been found to show pronounced associations with post-traumatic symptoms (Orth & Wieland, 2006). Whereas post-traumatic shame, guilt and anger have been investigated prolifically and thoroughly, other social affects, such as revenge or vengefulness, have received almost no attention. This is despite their commonly manifesting in trauma survivors. Other possibly relevant social emotions, such as contempt and feelings of dignity and honour, have also remained uninvestigated. Revenge, or vengeance, is often phenomenologically associated with anger but has been researched much lesser than anger. Gäbler and Maercker (2010) proposed that the development of revenge in the context of PTSD should be studied more intensely. The authors called for the examination of revenge as an interpersonal process that can be expressed in revenge fantasies and action-related intentions. Revenge fantasies are general ruminations about the idea of seeing the revenge target harmed. In contrast, action-related revenge intentions are defined by involving the planning of specific actions of revenge. In this perspective, revenge encompasses affective, cognitive and motivational components that take effect in a particular order. A recent study found that the effects of feelings of revenge moderated the time course of posttraumatic symptoms (Orth et al., 2006). PTSD symptoms remained at a consistent level over the years if strong

7 Socio-interpersonal Perspective on PTSD feelings of revenge were present, whereas PTSD symptoms tended to decline over the years if feelings of revenge were not present. More recently, Gäbler and Maercker (2011) studied individuals who had been traumatized approximately 30 years ago. They found that revenge fantasies, which were present in 30% of the sample, contributed to the maintenance of PTSD. However, revenge actionrelated intentions did not contribute to the maintenance of PTSD. In summary, social affective reactions are common phenomena that are concomitant to, or even constitutive of, PTSD. Research has shown that social affective reactions can predict the time course of PTSD symptoms. The current socio-interpersonal model of PTSD builds on these socio-affective changes in trauma survivors that were described in the PTSD models above. The Second Level: Close Social Relationships Interpersonal or social phenomena, such as disclosure, social support, distrust, feeling ostracized and compassion fatigue, have long been identified as important in PTSD. Accordingly, one important improvement to the current PTSD models stems from the need to integrate these additional theoretical frameworks. As indicated earlier, PTSD research has recently made substantial progress in integrating a dyadic view of trauma by including the particular processes that occur in couples after one partner has been traumatized (e.g., Monson et al., 2010). This inclusion provides a substantial contribution to the field as romantic relationships are characterized as the closest relationships in adult life. As such, these relationships produce the strongest effects on coping and adaptation processes over time. The following section comprises a discussion of the broader conceptual and empirical considerations regarding the interpersonal processes in close relationships. The most frequently discussed phenomena in the trauma literature include social support and disclosure as protective factors, and ostracization and blaming the victim as risk factors. These are found to affect the trauma victim s close relationship context. Disclosure The Need to Share Emotional Experiences In their classic study, Pennebaker and Harber (1993) studied inhabitants of San Francisco to investigate their communicative patterns in reaction to the Loma Prieta earthquake of The authors found very high rates of discussion about the event in the first weeks, referred to as the emergency phase. However, within the next month, the tendency to talk to others about the experience declined sharply. The authors termed this phase the social inhibition phase in their Social Stages of Coping model. In this phase, participants continued to think about the 471 events frequently but tried not to discuss them with others. This was not necessarily because the need to talk had decreased but out of fear that others might be tired or emotionally challenged listening to the emotionally loaded discussions. Numerous studies have replicated related findings, demonstrating the need for disclosure. One significant study was Rimé s work on social sharing after emotional upheavals. This impressively illustrates that the tendency to share emotional experiences is highly prevalent (Rimé, 2009). Rimé framed this tendency as being part of a collective coping process that reflects an increase in socio-affective needs for social reintegration that occurs after a traumatic experience and its possibly shattering impacts on world views and self-images. This framework supports the general hypothesis that any kind of disclosure is considered the starting point of an interactional process that occurs during the inception of psychological intimacy (Reis & Shaver, 1988). In other words, opening oneself emotionally is a required condition for the establishment of close relationships. The needs to be both understood and to be close to others are challenged by traumatic and intense emotional experiences. A more recent study, involving survivors of and witnesses to the 11 September 2001 terrorist attacks, applied an ecologically valid assessment method of social interactions in everyday life (Mehl & Pennebaker, 2003). This study showed that, after these events, individuals tended to engage in more dyadic interactions in their everyday lives, possibly to allow intimate disclosure exchanges. Interestingly, the amount of group conversations decreased after the trauma. Moreover, compared with the control days before11september,therelativeincreaseindyadicinteractions after 11 September was marginally associated with better mental health 2 weeks after the disaster. Using this same method, Mehl and colleagues showed that less small-talk and more substantive conversation were related to measures of happiness in the general population. These results also indicate the protective power of the processes related to the establishment of psychological intimacy (Mehl et al., 2010). It has recently been suggested that disclosure can be considered an interpersonal emotion regulation strategy (Horn, 2011). Daily momentary assessment of mood and disclosure in couples supports this assumption and reveals pronounced effects on both the disclosing and receiving partner (Horn et al., 2011). As mentioned earlier, many studies indicated that solitary written disclosure has beneficial effects on mental health (Pennebaker & Chung, 2007). However, there is an ongoing debate about how to use trauma disclosure in a clinical context. This is because individuals have shown no positive effects, or even negative effects, when forced to disclose immediately after a trauma (e.g., Bonanno, 2004). The potentially beneficial effects of disclosure depend critically on finding the individually appropriate

8 472 A. Maercker and A. B. Horn context, administration and time for disclosure, particularly the close social context. For example, in a recent diary study using non-traumatized couples naturally occurring daily disclosure of negative experience, the disclosures were associated with positive affective outcomes in the partner (Horn et al., 2011). This finding supports the role of positive processes in the relationship that are triggered by disclosure. Moreover, shortly after the daily disclosure of negative contents, both the disclosing individual s moodandthe partner s mood were found to be more positive, when controlled for rumination tendencies. This moderation illustrates the need to account for further individual differences when investigating interactional processes such as disclosure. In our own lab, Müller and colleagues investigated individuals subjective ability or inability to disclose after experiencing a trauma, (Maercker & Müller, 2004; Mueller et al., 2008). Firstly, we found that individuals perceived inability to disclose ( I can t talk about it ) and the need to disclose ( I just have to tell somebody about it, otherwise it ll drive me mad ) are independent factors. Moreover, disclosure is marked by a third independent factor, this being the emotional reaction that occurs while talking ( I feel highly agitated inside every time I talk about it ). In a recent study of crime victims, all three individual disclosure factors predicted later symptoms with medium-sized effects over and above conventional PTSD-related cognitive factors (Mueller et al., 2009). These self-referential perceptions of disclosure abilities should influence the actual sharing behaviour of the individuals. In turn, this sharing behaviour should impact the individuals social realities. As the interactional dimension of the socio-interpersonal context model proposes, interactions over time between intrapersonal and interpersonal processes may lead to negative or positive outcomes. It is necessary to further research disclosure behaviour during interaction and its interplay with individual dispositions, such as perceived disclosure abilities or avoidant emotion regulation tendencies. Pielmaier and Maercker (2011) already investigated such interactive pattern by using the sample case of traumatic brain injury patients and their partners. They found that significant portions of the patients PTSD levels were explained by partners dysfunctional disclosure attitudes (e.g., reluctance to talk) and the interactions between the patients own disclosure styles and their partners disclosure styles. A high rate of dysfunctional disclosure in both partners dramatically increased post-traumatic symptom distress. Social Support The Quality, Extent and Direction of Effects The term social support is used in very different contexts and is thus a type of omnibus concept. As mentioned in the outline of the article, the literature has consistently noted social support as being an important resource available after the experience of a trauma (Kaniasti & Norris, 2008; King et al., 1998; Norris et al., 2008; Solomon, 1995). Perceived social support has been consistently associated with positive mental and physical health outcomes (for a review, see Schwarzer & Knoll, 2007). However, frequently there is a discrepancy between subjectively perceived support and that which is actually provided (Schwarzer & Knoll, 2007). Furthermore, studies that include both actual and perceived social support show a more complex picture of the processes that underlie social support s effects. This complexity is especially true of diary studies, which analyse momentary or minimally retrospective effects. In these studies, disparities between the provided social support, as reported by a close person and as perceived by the recipient, often indicate detrimental effects on the patient (Bolger et al., 2000). This is especially pertinent in cases where the support provider is perceived as being unresponsive (Maisel & Gable, 2009). It is meaningful to keep these conceptual and methodological differences in mind. This complex result illustrates how the use of more sophisticated research methods, in this case dyadic momentary reports, yields distinct conclusions that have direct clinical implications. Therefore, we distinguish between perceived and actually received social support in our model to better foster more holistic research in this field. Moreover, in the PTSD field, negative or detrimental social support has repeatedly shown the strongest effect in positively predicting PTSD (Brewin et al., 2000; see also the next paragraph). In general, the effect of negative support was found exclusively through measures of selfperceived support and not through provided support or observational interaction measurements. Still, this effect supports reports from other fields about the possibly detrimental impact on patients (see below). Few of the studies that establish the direction of associations between social support and post-traumatic stress responses address time as an important factor. Koenen et al. (2003) found in a Vietnam veteran follow-up survey that both perceived positive and perceived detrimental social support decreased over time and that detrimental support was a longitudinal predictor of PTSD. Kaniasty and Norris (2008) allude to social causation and selection effects after a trauma. A cross-lagged analysis of longitudinal data revealed that social support is the most prominent cause of a decrease in symptoms during the earlier phase after a trauma. However, the causal relationship appears to switch directions 18 months after the trauma, when an increase in trauma-related symptoms leads to less reported social support. Again, this result demonstrates the importance of longitudinal studies and investigation of causal directionality and multiple mechanisms over time. Social support is referred to in so many measures that the need to innovate its analysis is undoubted (see Bolger

9 Socio-interpersonal Perspective on PTSD et al., 2003). This necessity supports the proposition that daily behaviour should be examined for interactions. In doing so, this assessment should include a dyadic perspective about the interpersonal processes that are inherent in social support and that have direct clinical implications. One conclusion that can already be drawn from recent research in social psychology is that simply telling significant others to be supportive to the traumatized patient in certain cases may have negative effects. Further research is needed to explore this important question. Also within PTSD research, the first attempts involving a dyadic perspective have yielded promising results. For example, female partners of those men with PTSD reported symptoms specific to their partner s identified traumas (e.g., Dekel, Solomon & Bleich, 2005; Renshaw et al., 2008). Particularly interesting are those studies that examine the interaction between the victim and his or her relations. Renshaw et al. (2008) showed that spouses of traumatized soldiers experienced greater symptom severity when their perceptions of their partners symptom levels were higher than the partners own reported symptom levels. Thus, agreement between spouses that they are experiencing severe problems appears to buffer the partners to some degree against psychological distress. These results contradict the assumption that spouses internalize the distress of the partner. Negative Social Exchanges Ostracization and Blaming the Victim Discussion of the possible positive effects of social support must acknowledge that social exchanges are not necessarily characterized by supportive intentions but can also be openly negative and detrimental (Newsom et al., 2008). For example, feelings of being ostracized or blamed are common among trauma survivors. Ostracism, or social exclusion, is a common phenomenon in PTSD. In one study, the ostracism experimental paradigm consisted of a ball tossing game with other people in a consulting waiting room (Williams, 1997). Two confidantes appeared to incidentally begin playing with a ball. Study participants in the experimental condition were included in the tossing for 1.5 min, whereas control participants were included until the end of the game. It was found that even healthy subjects felt severely ostracized after this manipulation. In a study that included both PTSD and healthy control participants, Nietlisbach and Maercker (2009) assumed aggravated effects for PTSD participants in the experimental condition. The results showed interaction effects for main psychopathological assessments (depression, anxiety, psychoticism) and the expected main effects for the majority of outcome measures (psychopathology, well-being, sense of belonging and sense of meaningful existence). An analogous virtual reality version ( cyberball ) of the study revealed similar results 473 with other disorders (borderline personality disorder, social phobia), indicating that the specificity of the findings has yet to be investigated. One prominent aspect of detrimental support is blaming the victim. Frequently, victims report that not only do they no longer find the reaction of the environment to be supportive but also that people blame them for the traumatic event ( Why do you have to hang around in areas where things like that happen? and You brought what happened back upon yourself ). Blaming the victim has been examined primarily in victims of sexual child abuse. A puzzling result reported by Lamb and Edgar-Smith (1994) showed that the most supportive form of reaction from others consisted of the perception of not being blamed by others who referred to the abuse. Ullman (2000) studied a sample of women who had experienced sexual aggression. In addition to two positive forms of reaction (instrumental support and emotional support), the author identified the following five negative forms of reaction in the victims close relationship context: blaming the victim, taking away decisions, stigmatization, distraction and egocentric reaction. Ullman (2007) showed that these negative reactions were associated with higher levels of PTSD symptoms. However, it is worth mentioning that the studies cited here only rely on information provided by the victim. Again, social interactive studies that include the interaction partner s perspectives are needed in this area. Other evidence from self-reported data stems from research on former political prisoners from East Germany (former German Democratic Republic; Maercker & Müller, 2004), who report strong recent perceptions of rejection. As these findings demonstrate, blaming the victim can happen at the level of distant social environments, as well as in close relationships. This is a good example of the nested nature of social environments that imply mutual interdependence. If, e.g., victims of political crimes are blamed, this blame impacts both the victim s everyday life interactions with close relations and the victim s intrapersonal cognitive representation about the trauma. Important aspects of the situation may be neglected when research focuses only implicitly or explicitly on the individual. One explanation for the latter case may be the rejection of the moral rebels effect, which has been described in social psychology (Monin et al., 2008). The rebels are rejected when they threaten the positive self-perceptions of the bystanders who do not oppose what the rebels oppose. Take the example of political activists who openly express generally held discontent about a societal issue (e.g., bad government). Many of these activists who get arrested and are tortured later experience rejection by their fellow citizens (e.g., the punishment they received was deserved and they surely lost control of themselves ). Other social psychological processes may explain the whole range of victim-blaming attitudes in this particular context.