Tilburg University
The burden of interpersonal violence
Kunst, M.J.J.
Publication date:
2010
Document Version
Publisher's PDF, also known as Version of record
Link to publication in Tilburg University Research Portal
Citation for published version (APA):
Kunst, M. J. J. (2010). The burden of interpersonal violence: Examining the psychological aftermath of victimisation. [s.n.].
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal Take down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
The Burden of Interpersonal Violence:
Examining the psychosocial aftermath of victimisation
Proefschrift
ter verkrijging van de graad van doctor aan de Universiteit van Tilburg, op gezag van de rector magnicus, prof.dr. Ph. Eijlander, in het openbaar te
verdedigen ten overstaan van een door het college voor promoties aangewezen commissie in de aula van de Universiteit op woensdag 12 mei 2010 om 14.15 uur door Maarten Jacob Johannes Kunst, geboren op 23 mei 1978
2
Promotores: Prof. dr. S. Bogaerts
Prof. dr. F.W. Winkel
Promotiecommissie: Prof. dr. B.P.R. Gersons
Prof. dr. J.E. Hovens Prof. dr. F. Hutsebaut Prof. dr. D.G. Kilpatrick Dr. P.G. van der Velden Prof. dr. A. Vrij
Vormgeving en druk: Artoos Communicatiegroep B.V.
Copyright
3
Voorwoord1‚What is this thing called science?‛ Zo luidt de titel van het boek van Alan Chalmers dat in 1976 voor het eerst werd uitgegeven en daarna twee maal is herzien. De laatste herziene versie uit 1999 is herhaaldelijk in herdruk verschenen en verplichte kost voor vrijwel elke beginnende student in de Sociale Wetenschappen in het kader van een cursus Wetenschapsfilosofie. Chalmers beschrijft in dit boek onder meer hoe onderzoekers op basis van wetenschappelijke theorieën komen tot het formuleren van falsifieerbare hypothesen en die vervolgens trachten te toetsen in empirisch onderzoek. De gedane bevindingen kunnen steun bieden voor de theorie die als uitgangspunt werd genomen of aanleiding geven die te verwerpen of op punten aan te passen. Het gaat hierbij om een oneindige keten waarvan niemand weet wanneer die is begonnen en wanneer die zal eindigen. Elke theorie is namelijk niets anders dan een alternatieve of aangepaste en op waarnemingen gebaseerde zienswijze over een verschijnsel dat een eerdere, reeds bestaande theorie niet of slechts gedeeltelijk in staat was te verklaren. Het zal u niet verbazen dat Chalmers, zoals hij ook zelf toegeeft, de lezer een (eenduidig) antwoord op de door hem zelf opgeworpen vraag schuldig moet blijven.
Na slechts drieënhalf jaar in de wetenschap als onderzoeker vertoefd te hebben heb ik niet de illusie het vraagstuk van Chalmers wel te kunnen oplossen. Hooguit kan ik op basis van mijn eigen ervaringen aangeven hoe ik als jonge wetenschapper te werk ben gegaan om in ieder geval een bijdrage, hoe minimaal die vaak ook was, te leveren aan de verdere uitbouw van de keten van theorie en empirie (en vice versa) die Chalmers heeft beschreven.
Voordat ik in september 2006 begon aan mijn promotietraject binnen
het INTERNATIONAL VICTIMIMOLOGY INSTITUTE TILBURG
(INTERVICT), heb ik bijna vier jaar gewerkt als jurist bij het Uitvoeringsinstituut werknemersverzekeringen (UWV). Ik deed dat in combinatie met een avondstudie Psychologie. De werkdruk bij het UWV was weliswaar bescheiden in vergelijking met bijvoorbeeld de advocatuur of de rechterlijke macht, maar door de combinatie van werken en studeren had ik een behoorlijk druk bestaan. Ik werd hierdoor regelmatig gedwongen prioriteiten te stellen. Uitgangspunt daarbij was altijd wat haalbaar was en niet zozeer wat wenselijk was. Zo kan ik me nog herinneren dat ik tijdens de
1 I apologise to those who cannot read Dutch. I decided to write the preface to this thesis in
4
bachelorfase een schriftelijk tentamen met open vragen binnen een uur moest afronden om tijdig op de rechtbank ’s-Hertogenbosch te kunnen verschijnen voor de mondelinge behandeling van enkele werkloosheids- en arbeidsongeschiktheidszaken. Ik wist dit uiteraard van tevoren en had me daarom voorgenomen alleen die vragen te beantwoorden waarvan ik zeker het antwoord wist. Deze strategie leverde een bescheiden zes op, die voor mij echter voelde als een acht of hoger. Tijdens mijn promotietraject heb ik getracht deze praktische instelling vast te houden.
Bij aanvang van mijn aioschap werd ik voor de vraag gesteld waar ik mijn onderzoeksgegevens tijdig vandaan zou kunnen halen. Mijn onderzoek had aanvankelijk de bedoeling de lange termijngevolgen (in het bijzonder carrièregerelateerde gevolgen) van slachtofferschap van geweld in kaart te brengen. Idealiter vergt dit een onderzoeksopzet waarbij een grote groep mensen, bij voorkeur aan de hand van een steekproef die is getrokken uit de algemene bevolking, gedurende enkele jaren worden gevolgd. Zo’n opzet maakt het namelijk mogelijk om slachtoffers en niet-slachtoffers met elkaar te vergelijken op uitkomstmaten en daarbij te controleren voor verschillen die al bestonden voordat het geweldsmisdrijf gepleegd werd. Gelet op de termijn waarbinnen je als aio een promotietraject dient af te ronden (in mijn geval vijf jaar) zou een dergelijke aanpak zeer moeilijk te realiseren zijn en moest ik nadenken over een mogelijke andere opzet. Een serieus alternatief werd me al snel na mijn indiensttreding aangereikt; Tijdens het congres Victim Empowerment, dat in oktober 2006 door INTERVICT werd georganiseerd, liep ik adjunct-directeur Jan Bierhoff van het secretariaat van het Schadefonds Geweldsmisdrijven tegen het lijf. Dit is een instantie die onderdeel uitmaakt van het Ministerie van Justitie (Directie
Sanctie- en Preventiebeleid) en die slachtoffers van ernstige
5
Toen ik eenmaal toegang had tot een geschikte onderzoekspopulatie, kon ik aan de slag met het opzetten van mijn onderzoek. Ik koos ervoor sets met vragenlijsten af te nemen onder alle slachtoffers die in 2006 een aanvraag bij het Schadefonds hadden ingediend. Na het samenstellen van de vragenlijst heeft Koen Buddelmeijer, die destijds werkzaam was op de afdeling ICT van het fonds, voor mij een lijst van ruim 4500 slachtoffers gegenereerd uit het geautomatiseerde persoonsregistratiebestand van het fonds. Dit was geen eenvoudige klus, omdat de geregistreerde ontvangstdatum van de aanvraag niet betekende dat ook daadwerkelijk een aanvraag was ingediend. Om dat betrouwbaar te kunnen bepalen moest hij tevens gebruik maken van andere parameters. Alle slachtoffers die hij voor me geselecteerd had heb ik in het laatste kwartaal van 2007 schriftelijk benaderd met het verzoek een vragenlijst via het internet in te vullen en met de vraag of zij wilden meedoen aan één of meer vervolgstudies. Koen en alle andere medewerkers van het Schadefonds zal ik eeuwig dankbaar blijven voor hun ondersteuning bij de uitvoering van mijn onderzoek.
Nadat ik de dataverzameling had afgerond, moest ik de verzamelde gegevens analyseren en hierover op een zinvolle wijze rapporteren. ‚Zinvol‛ betekent binnen de Psychologie, zo begreep ik van mijn promotoren prof. Frans Willem Winkel en prof. Stefan Bogaerts: Schrijf een aantal papers, bied die aan ter publicatie bij een zogenaamd peer-reviewed tijdschrift, wacht totdat enkele daarvan zijn geaccepteerd voor publicatie en bundel het geheel uiteindelijk, vergezeld van een in- en uitleiding, tot een proefschrift. Zij waarschuwden mij dat deze opgave een stuk moeilijker is dan die op het eerste gehoor lijkt te zijn. Er zijn namelijk boekenkasten vol geschreven over de gevolgen van geweld; Bij het opgeven van de zoektermen ‚violence‛, ‚victim‛ en ‚consequences‛ in Google Scholar vind je alleen al om en nabij de 388.000 treffers. Daar kwam nog bij dat aan de methode die ik had gehanteerd (een surveyonderzoek onder een specifieke groep slachtoffers) de nodige methodologische bezwaren kleefden. (Ik zal u hier op deze plaats niet mee lastigvallen). Om iets geaccepteerd te krijgen voor publicatie moest ik de redactieleden en reviewers van de tijdschriften die ik had geselecteerd er dus van zien te overtuigen dat mijn onderzoek desondanks nog van toegevoegde waarde was voor de eerder genoemde keten van theorie en empirie.
6
proberen een ‚niche‛ te detecteren in de bestaande literatuur en die proberen op te vullen met de resultaten uit mijn onderzoek. Sommige collega-wetenschappers zullen dit een verwerpelijke strategie vinden, omdat dit in strijd is met de ethische regel dat onderzoeksvragen vóór aanvang van een studie dienen te worden geformuleerd. Vanuit ideologisch oogpunt ben ik het volledig met hen eens, maar zoals Bertolt Brecht al in de Driestuiversopera uit 1928 opmerkte: ‚Erst das Fressen, dann die Moral.‛ Anders gezegd: Er is een hoop geoorloofd in de bikkelharde publicatiestrijd waar onderzoekers binnen de Sociale Wetenschappen gedurende hun carrière aan moeten deelnemen en dat geldt wat mij betreft zeker voor beginnende wetenschappers die pas kunnen promoveren bij een minimum aantal publicaties. De suggestie van Frans Willem leek me bij deze strijd zeker niet tot het arsenaal van ongeoorloofde gevechtstactieken te behoren en heb ik dan ook zonder aarzeling als praktische tip ter harte genomen bij het schrijven van de artikelen die als afzonderlijke hoofdstukken in dit proefschrift zijn opgenomen. Welke niches ik de afgelopen jaren heb proberen op te vullen kunt u in het vervolg van dit manuscript zelf lezen.
7
Table of contents
Chapter 1: General introduction ... 9
Part I:Prevalence and DVCF (file) predictors of posttraumatic stress disorder (PTSD) ...23
Chapter 2: Prevalence and predictors of posttraumatic stress disorder (PTSD) among victims of violence applying for state compensation ...25
Part II:Personality dispositions, posttraumatic stress disorder (PTSD), and posttraumatic growth (PTG) ...43
Chapter 3: Posttraumatic anger, recalled peritraumatic emotions, and
violence-related posttraumatic stress disorder (PTSD) symptoms ...45 Chapter 4: Type D personality and posttraumatic stress disorder (PTSD) in victims of violence...61 Chapter 5: Affective personality type, posttraumatic stress disorder (PTSD) symptoms, and posttraumatic growth (PTG) in victims of violence ...75
Part III:The relationship between posttraumatic stress disorder (PTSD) and posttraumatic growth (PTG) ...87
Chapter 6: Peritraumatic distress (PD), posttraumatic stress disorder (PTSD), and posttraumatic growth (PTG) in victims of violence ...89 Chapter 7: Posttraumatic growth (PTG) moderates the association between violent revictimisation and persisting posttraumatic stress disorder (PTSD) symptoms in victims of interpersonal violence ...101
Part IV:Violent victimisation, posttraumatic stress disorder (PTSD), employment, and income attainment ...115
8
Part V: Peritraumatic reactions, posttraumatic stress disorder(PTSD)
symptoms, and malingering ...141
Chapter 10: Recalled peritraumatic reactions, self-reported posttraumatic stress disorder (PTSD) symptoms, and the impact of malingering and fantasy proneness in victims of interpersonal violence who have applied for state compensation ...145
Chapter 11a:Summary, discussion, and conclusions ...161
Hoofdstuk 11b:Samenvatting, discussie, en conclusies ...171
9
10
1.1 BackgroundThe burden of violent victimisation has received much attention in scientific literature. Most previous studies seem to have focussed on adverse psychological or medical consequences of victimisation (e.g., Denkers, 1996). A few others have attempted to uncover its negative socioeconomic impact (e.g., Dolan, Loomes, Peasgood, & Tsuchiya, 2005). And finally, several studies have considered the positive side of violence (e.g., Cobb, Tedeschi, Calhoun, & Cann, 2006).
Despite the abundance of available studies on the aftermath of violent victimisation, many issues still remain to be uncovered. Relying on a sample of victims of violence who had claimed compensation from the Dutch Victim Compensation Fund (DVCF), the purpose of this PhD project was to further unravel the psychosocial aftermath of violent victimisation in this specific subgroup of interpersonal violence. A number of studies have investigated mental health outcomes of violence in victims with a history of application for compensation from the state. These studies primarily focussed on victims of mass casualties, such terrorist bombings (e.g., Verger et al., 2004). However, on the other hand, victims of individual casualties,
such as civilian violence, seem to have been neglected in previous research.2
Before introducing the main topics addressed by this thesis and the
study’s3 design, the following paragraphs will briefly elaborate on the
current state of the art. First, a three-partite model of adjustment following violent victimisation is shortly discussed. This model should not be considered as a general theory underlying the topics addressed by this thesis, but aims to provide the reader a broad framework that relates the separate chapters to one another. Next, the literature on maladjustment to violent victimisation is summarised. Finally, the concept of posttraumatic growth (PTG) in the context of violence is presented.
1.1.1 A conceptual model of adjustment to violent victimisation
Sales, Baum, and Shore (1984) proposed a general three-partite model of adjustment following violent victimisation. They argued that three categories of variables determine the aftermath of violent victimisation: pre-,
2 It should be acknowledged that the text of the Dutch act on compensation for violent crime
does not limit eligibility to victims of individual violence (Wet Schadefonds Geweldsmisdrijven van 1975). However, given the low number of terrorist attacks and terrorism-related victims in the Netherlands during the last decades, it was deemed highly unlikely that the fund had compensated victims of terrorism.
3 Note that the term study in this chapter at one time refers to the overall study and at other
11
peri-, and postvictimisation factors. Previctimisation factors comprise a wide array of biological, social, demographic, and contextual factors, personality traits, and previctimisation psychological states. Perivictimisation factors involve characteristics of the violent act, such as its nature, severity and length, and the victim’s behaviour during the event. Postvictimisation factors include, among other things, individual coping and appraisal strategies, tangible damage compensation, perceived social support, and recognition of one’s victim status by public authorities. Separate factors may cumulate or interact with one another to produce the outcome of the violent event. Sales et al.’s model was previously used in Dutch Victimological research by Denkers (1996).
1.1.2 Psychosocial maladjustment to violent victimisation: a brief summary of the literature
Violence-related posttraumatic stress disorder (PTSD)
As mentioned above, adverse psychological outcomes have been the most extensively studied psychosocial consequences of violent victimisation. The majority of studies into this area has focussed on (symptoms of) posttraumatic stress disorder (PTSD) (e.g., Andrews, Brewin, Rose, & Kirk, 2000; Johansen, Wahl, Eilertsen, & Weisaeth, 2007; Kilpatrick et al., 2003; Lawyer, Ruggiero, Resnick, Kilpatrick, & Saunders, 2006; Orth, Cahill, Foa, & Maercker, 2008). PTSD is a highly impairing disorder, both to the individual involved and to society (e.g., Kessler, 2000). It was first introduced in the third edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM) of the American Psychological Association (APA) in 1980. According to the revised fourth version of the DSM (DSM-IV-TR, APA, 2000), exposure to a traumatic event is a prerequisite for diagnosis of PTSD. To qualify as a traumatic event, the person in question must have ‚experienced, witnessed, or been confronted with an event or events that involved ‚actual or threatened death or serious injury, or a threat to the physical integrity of self or others‛, while the initial reaction to the event must have involved ‚intense fear, helplessness, or horror‛ (APA, 2000, p. 463). These quotations reflect DSM criteria A1 and A2 for PTSD, respectively.4 Violent personal assault (i.e., sexual assault, physical attack,
4 The wording of PTSD criterion A is subject of much controversy and debate (see Brewin,
12
robbery, and mugging) is explicitly listed in DSM-IV-TR as an example of an event that fulfills criterion A1 (APA, 2000). Criterion A2 requires assessment by validated instruments or clinical investigation. A diagnosis of PTSD is only warranted if the response to the traumatic event is characterised by at least one symptom of reexperiencing/intrusion (criterion B), three symptoms of avoidance/numbing (criterion C), and two symptoms of hyperarousal (criterion D). In addition, the full symptom picture must be present for more than 1 month (Criterion E) and the disturbance must cause clinically significant distress or impairment in social, occupational, or other important
areas of functioning (Criterion F; APA, 2000).56
Relying on prospective evidence, Winkel (2007) estimated that the prevalence of PTSD in crime victims varies between 10% and 15%. However, he also emphasised that individual studies found percentages as high as 25% (Wohlfarth, Winkel, & Van den Brink, 2002) and 32% (Birmes et al., 2003). Adverse outcomes other than PTSD
Other possible outcomes of violent victimisation include depression (e.g., Bargai, Ben-Shakhar, & Shalev, 2007; Sorenson & Golding, 1990), avoidance behaviour, fear of crime, (phobic) anxiety, hostility, somatisation (Norris & Kaniasty, 1994), panic disorder (Winkel, 2009), eating disorder (Brady, 2008), substance abuse (e.g., Vermeiren, Schwab-Stone, Deboutte, Leckman, & Ruchkin, 2003), insomnia and sleep-disordered breathing (Krakow et al., 2001), sexual dysfunctioning (Letourneau, Resnick, Kilpatrick, Saunders, & Best, 1996), and secondary victimisation (e.g., Campbell, 2008). Finally, perceived impairment in physical (e.g., Campbell, 2002; Kawsar, Anfield, Walters, McCabe, & Forster, 2004; Koss, Koss, & Woodruff, 1991; Mouton, 2003) and social functioning (e.g., Boney-McCoy & Finkelhohr, 1995a; Browne, Salomon, & Bassuk, 1999) have been observed in victims of violence. However, more hard socioeconomic outcomes of violent victimisation seem to have been largely neglected in previous studies. One of the few exceptions was performed by Macmillan (1998, 2000) and Macmillan and Hagan (2004), who found a relationship between adolescent victimisation and later income attainment.
5 For a historical overview of PTSD DSM criteria, see McNally, 2004.
6 Several theoretical explanations have been proposed for the development and maintenance of
13
Risk factors of psychosocial maladjustment postvictimisation
A wide array of risk and protective factors for maladjustment following violent victimisation has been identified over the past decades. All of the may be categorised as pre-, peri-, or postvictimisation factors (cf. Sales et al., 1984). Examples are the nature and severity of the act of violence (e.g., Beaton, Cook, Kavanagh, & Herrington, 2000); coping strategies (e.g., Gutner, Rizvi, Monson, & Resick, 2006; Krause, Kaltman, Goodman, & Dutton, 2008; Valentiner, Foa, Riggs, & Gershuny, 1996); social support (e.g., Yap & Devilly, 2004); emotional reactions, such as feelings of revenge (Orth et al., 2006), anger (Orth et al., 2008), shame (Andrews et al., 2000), and self-blame (Kraaij, Arensman, Garnefski, & Kremers, 2007; Ullman, Townsend, Starzynski, & Long, 2006); peritraumatic dissociation (e.g., Hetzel & McCanne, 2005), acute stress symptoms (Birmes et al., 2003), and peritraumatic tonic immobility (Fiszman et al., 2008); information processing style (e.g., Elwood, Williams, Olatunji, &, Lohr, 2007); attachment style (Bogaerts, Daalder, Van der Knaap, Kunst, & Buschman, 2008; Elwood & Williams, 2007); loneliness (Kunst, Bogaerts, & Winkel, in press-a); relationship to the perpetrator (Gutner et al., 2006); and history of alcohol abuse (e.g., Kaysen, Simpson, Dillworth, Larimer, Gutner, & Resick, 2006). Furthermore, consistent evidence seems to exist regarding the additional deterrent effect of history of youth victimisation on adjustment after victimisation in adulthood (e.g, Andrews et al., 2000; Boney-McCoy & Finkelhor, 1995b, 1996; Campbell, Greeson, Bybee, & Raja, 2008; Hembree, Street, Riggs, & Foa, 2004; Nishith, Mechanic, & Resick, 2000). Finally, a number of studies have found that revictimisation renders victims at an increased risk of PTSD (e.g., Krause et al., 2008; Winkel, Blaauw, Sheridan, & Baldry, 2003). Unfortunately, however, due to methodological limitations, most previous studies have failed to address the specific mechanism
through which these factors determine the outcome of violent victimisation.7
1.1.3 Violent victimisation and posttraumatic growth (PTG)
In addition to psychological distress and its determinants, several times experiences of posttraumatic growth (PTG) have been investigated in various types of violence, such as rape (Burt & Katz, 1987), sexual assault (e.g., Frazier, Conlon, & Glaser, 2001), intimate partner violence (Cobb et al.,
7 Several studies have attempted to establish mediation or moderation models in victims of
14
2006), and child abuse (e.g., McMillen, Zuravin, & Rideout, 1995). Zoellner and Maercker (2006) defined PTG as ‚the subjective experience of positive psychological change reported by an individual as result of the struggle with trauma‛ (p. 628). Individuals who have experienced a highly traumatic event need to accommodate pretrauma schema’s about the world, self, and others (Janoff-Bulman, 1992) and thereby tend to find meaning in the event itself (Taylor, 1983). The outcome of this mean-making process may result in PTG (Joseph & Linley, 2005, 2006; Tedeschi & Calhoun, 2004). PTG refers to the development of additional beneficial outcomes and thus involves more than recovery (i.e., returning back to pretrauma levels of functioning).8 According to Tedeschi and Calhoun (1995), a distinction must be made between positive changes in the self, relatedness to others, and a changed philosophy of life (i.e., setting other priorities, a greater appreciation for life, and spiritual or religious transformations).
Research into the correlates of PTG among victims of violence is scare, but seems to suggest that high dispositional optimism (Updegraff & Marshall, 2005), old age (Grubaugh & Resick, 2007), and having a role model (Cobb et al., 2006) are positively related to PTG. However, whether and how PTSD and PTG relate to each other is still unclear.
1.2 Outline of the thesis
The studies included in this thesis cover a wide array of topics and are therefore not rooted in one particular theoretical perspective. However, broadly stated, they cover five general themes that require further empirical exploration. All regard the pre-, peri-, and postvictimisation eras.
Part I: Prevalence and (file) predictors of PTSD
Part I is introductionary in nature and involves a single study that attempts to estimate prevalence and predictors of PTSD in victims applying for
8 Whether these additional beneficial outcomes are real is subject of much debate. Several
15
compensation with the DCVF. Despite the abundance of studies on violence-related PTSD, these issues have never been addressed in previous research. Victims of violence can apply for compensation if (1) the perpetrator has committed the act of violence intentionally, (2) the act of violence has been committed in the Netherlands, (3) the victim has suffered severe physical or psychological loss, (4) the victim does not bear any guilt for the act of violence, and (5) the damages suffered cannot be compensated through other means (Wet Schadefonds Geweldsmisdrijven van 1975). Due to the latter requirement, many victims do not seek state compensation until other potential sources of compensation have been explored and exhausted. Application forms should be submitted to the DVCF within three years after victimisation though. Both material damage and pain and suffering are
amenable to compensation by the DCVF (Wet Schadefonds
Geweldsmisdrijven van 1975). Yearly, approximately 7000 victims of violence apply for compensation with the DVCF (Schadefonds Geweldsmisdrijven, 2008). Previous studies (Hoogeveen & Van Burik, 2008; Spapens & Hoogeveen, 2001) have estimated that DVCF applicants cover around 20% of all victims entitled to compensation in the Netherlands. Hoogeveen and Van Burik (2008) found that particularly women and the elderly fail to claim compensation from the state. To broaden the DCVF’s reach, they suggested to take up policy measures that increase the number of applications in these subgroups of potential applicants (Hoogeveen & Van Burik, 2008; see also Hoogeveen & Van Burik, 2009). However, from a mental health perspective, enlarging the DVCF’s target population seems less important than referral of those in need for psychosocial assistance. Therefore, study 1 investigates to what extent DVCF applicants suffer from PTSD and which file data provide an indication of PTSD at time of application. The (file) variables used in this study comprise pre- (e.g., acquaintenance between victim and perpetrator), peri- (e.g., type of violence), and postvictimisation (e.g., level of compensation for pain and suffering) factors.
Part II: Personality dispositions, PTSD, and PTG
16
and thinking about the environment and oneself that are exhibited in a wide range of social and personal context‛ (APA, 2000, p. 686) and are often assumed to have existed already prior to victimisation (Sales et al., 1984). An important feature of each study is the exploration of third variables that may explain these associations. Study 2 investigates the contribution of
posttraumatic anger disposition9 to PTSD (cf. Chemtob, Novaco, Hamada,
Gross, & Smith, 1997). It argues that posttraumatic anger disposition is characterised by a threat bias that triggers a survival mode of functioning when this is inappropriate. Consequently, individuals scoring high on posttraumatic anger disposition are likely to recall peritraumatic emotions and to trigger posttraumatic intrusions - a factor associated with
maintenance of PTSD. Studies 3 and 4 both examine configurations10 of
personality traits in relation to adjustment following victimisation. Study 3 describes the association between the distressed personality type (type D personality) and PTSD. The concept of type D personality (Denollet et al., 1996; Denollet, Sys, & Brutsaert, 1995) was developed within Medical Psychology and has recently been introduced in Victimology (Kunst, Bogaerts, & Winkel, 2009). Its main premise is that high negative affectivity (i.e., the tendency to experience negative emotions) is particularly malicious if it is accompanied by high social inhibition (i.e., the tendency to inhibit self-expression). Study 4 builds on previous work by Norlander, Von Schedvin and, Archer (2005) and forms an overture for part III by examining four combinations of positive and negative affectivity in relation to PTSD and PTG. It postulates that PTSD and PTG scores are particularly intercorrelated in high affective individuals (i.e., those with a high score both on negative and positive affectivity).
Part III: The relationship between PTSD and PTG
Part III of the thesis consists of two studies exploring the association between PTSD and PTG. To date, empirical research has failed to find consistent support for a relationship between PTSD and PTG. Most previous studies have documented either insignificant or small correlations between
9 Given DSM’s definition of personality, the combination of the terms posttraumatic and
disposition may appear to be contradictory. In this thesis the terms posttraumatic anger disposition
are used to indicate a person’s tendency to experience anger across situations during the period between victimisation and assessment (cf. R.W. Novaco, personal communication, August 17, 2007).
10 The term configurations refers to discrete personality types involving a particular combination of
17
the two (Zoellner & Maercker, 2006). Several scholars have tried to explain these findings by arguing that their interrelation is not linear in nature (e.g., Helgeson, Reynolds, & Tomich, 2006). Studies 5 and 6 address this issue and propose two alternative conceptualisations of the association between PTSD and PTG. Based on the speculation that those with intermediate levels of PTSD are most likely to experience PTG (cf. Kleim & Ehlers, 2009), study 5 explores whether the two are curvilinearily related to each other. In addition, curvilinearity between peritraumatic distress and PTG is explored (cf. McCaslin et al., 2009). Study 6 attempts to establish a moderator relation between PTG and violent revictimisation. More specifically, it investigates whether PTG protects against increases in PTSD symptom severity after violent revictimisation. Both study 5 and study 6 mainly reflect on the postvictimisation era.
Part IV: Violent victimisation, PTSD, employment, and income attainment
18
Part V: Recalled peritraumatic reactions, PTSD, and malingering
The thesis is concluded with an exploration of malingered symptoms in victims of violence who have applied for state compensation with the DVCF. DSM-IV-TR (APA, 2000) has defined malingering as ‚the intentional production of false or grossly exaggerated physical or psychological
symptoms, motivated by external incentives‛ (p. 739).11 Malingering may
take either of three forms: (1) pure malingering (i.e., complete fabrication of symptoms), (2) partial malingering (i.e., overreporting of existing symptoms) and (3) false imputation (i.e., misattribution of symptoms to a traumatic event) (Resnick, 1995, 1997). Malingering should be ruled out in scientific research (Rosen, 2003; Rosen & Taylor, 2007) and during the diagnostic process of PTSD when ‚financial renumeration, benefit eligibility,
and forensic determinations play a role‛ (DSM-IV-TR, APA, 2000, p. 467).12
Since PTSD is an important factor in determining level of compensation for pain and suffering from the Dutch state, self-reported PTSD symptoms in victims who have applied for compensation with the DVCF may be associated to malingering. However, to date, this has never been investigated.
Furthermore, and in line with Candel and Merkelbach (2004), malingering may be argued to account for the relationship between peritraumatic dissociation and PTSD often observed in trauma research (e.g., Marshall & Schell, 2002). A similar hypothesis may explain associations between other types of peritraumatic reactions and self-reported PTSD. Therefore, study 9 explores the confounding role of malingering in the relationships between PTSD symptoms and three dimensions of peritraumatic responses: peritraumatic dissociation, peritraumatic distress,
and peritraumatic tonic immobility.13 Additionally, the associations between
11 Malingered PTSD differs from factitious PTSD in that symptoms are not overreported to
acquire the sick role (Fear, 1996).
12 Unlike many other psychiatric disorders, PTSD is linked to a specific event and is therefore
more vulnerable to disability claims (Slovenko, 1994; see Rosen & Taylor, 2007 for a contrasting view). In the United States, disability allowance for mental health problems due to army service even depends on diagnosis of PTSD (Code of Federal Regulations of 2008). Not surprisingly, many studies suggest that malingered PTSD is a common phenomenon in American war veterans (e.g., Fontana & Rosenheck, 1998; Frueh et al., 2003; Frueh, Gold, & De Arellano, 1997; Frueh, Smith, & Barker, 1996; Gold & Frueh, 1999).
13 It should be noted that malingering can only be assessed by proximity measurement.
19
peritraumatic dissociation and the two other peritraumatic phenomena are investigated to determine which risk factor will emerge as the most dominant one in predicting PTSD symptomatology. This topic has never been addressed in previous research either.
1.3 Study design
Approval of data collection by the DVCF Committee was obtained in March 2007. Psychological measures were selected and prepared for administration during spring and summer 2007. Data collection comprised three phases. The first phase covered the last quarter of 2007 (T1). Victims who had applied for compensation during 200614 (T0); were ≥ 16 years old; and did not have missing electronic file data on age, gender, and date of crime15, were contacted with a letter that informed them on the purpose and content of the study. It also indicated that they had the right to refuse participation and were allowed to withdraw from the study at any time. Anyone
interested was instructed to fill out an internet survey.16 Those who did not
have access to the World Wide Web could contact the DVCF for a paper-and-pencil version of the questionnaire. In line with DVCF policy to bother applicants as little as possible, it was agreed upon that nonrespondents were not sent reminders. Respondents who had consented to participate in the first follow-up study, were recontacted six months after initial assessment (T2). A paper-and-pencil version of the T2 questionnaire and cover letter was sent to those who had requested a paper-and-pencil questionnaire at T1.
experts agree that a golden standard to assess malingering does not exist (e.g., Guriel & Fremouw, 2003; Hall & Hall, 2006; Institute of Medicine, 2007; Resnick, 1997; Rogers, 1997; Wilson & Moran, 2004). It follows that the term malingering always refers to possible or at best
probable malingering (for a more elaborate discussion of this issue, see chapter 10).
14 The period of application chosen to select potential participants was prompted by the
possibility that those who had applied at a later moment had not received a decision on their claim yet or had objected to it. Unfortunately, the electronic database used by the fund was not able to produce a reliable list of recently closed files.
15 Workers of the DVCF ICT department notified the author of this thesis that missing data on
these variables indicated that the victim involved had failed to return the application form after an initial request for sending a copy through the mail.
16 Applicants were divided in four equal groups at T0: those who had applied for compensation
20
Others only received the cover letter and could again choose between online and paper-and-pencil participation (by request). In February 2009 (T3), a paper-and-pencil questionnaire and cover letter was sent to T2 respondents who had consented to participate in the second follow-up study. Response rates were approximately 20% at T1, 35% at T2, and 65% at T3. Figure 1.1 presents a flowchart of study participants.17
Figure 1.1 Flowchart of study participants
Simultaneously with T2 assessment, a convenience sample of nonvictims was composed by dispersion of 20.000 unaddressed leaflets in 15 postal code areas randomly selected in each province of the Netherlands. Leaflets contained information comparable to that provided to the DVCF group and the URL of the website hosting the questionnaire for the comparison group. This strategy resulted in a comparison group of 283 respondents who had either completely or partly filled out the questionnaire (online or by paper-and-pencil).
17 Note that sample sizes reported in the separate studies may slightly differ from those
included in the figure due to diverging inclusion criteria.
4513 victims selected from electronic files at T0
3832 invited to participate
813 respondents at T1
251 respondents at T2 681 not eligible for participation due to missing file data or age
128 respondents at T3 122 refused further participation
21
1.4 Guide to the readerIn the following chapters, the nine studies that comprise this thesis will be described in more detail. As each chapter has been submitted to an international journal for peer review and publication, chapters are primarily to be read as stand-alone documents. Consequently and inevitably, the thesis contains several repetitions (particularly of methods descriptions, footnotes, and references) and contradictory statements. To simplify understanding, all chapters were slightly adapted to be presented in a largely similar format. Content was not changed though. All chapters largely follow APA (2001) format for (in-text) references, tables, and figures.18 The last two chapters provide a brief summary and discussion of the thesis’ main results and conclusions in English and Dutch, respectively.
18 Contrary to APA style, paragraph titles are numbered. British English spelling was preferred
23
25
Chapter 2
Prevalence and (file) predictors of posttraumatic stress
disorder (PTSD) in victims of interpersonal violence who
have applied for compensation from the state
Adapted from: Kunst, M.J.J., Winkel, F.W., & Bogaerts, S. (in press-a). Prevalence and predictors of posttraumatic stress disorder among victims of violence applying for state compensation. Journal of Interpersonal Violence.
Abstract
26
2.1 IntroductionVictims of violence need to cope with the psychological burden that the act of violence has laid upon them. Although most of them appear to adapt well to the stress experienced shortly after victimisation and return to their previctimisation levels of functioning rather soon (e.g., Norris & Kaniasty, 1994), some still suffer from distress after several months have passed and may even experience symptom levels that qualify them for chronic posttraumatic stress disorder (PTSD).
To be diagnosed with chronic PTSD, the individual involved must fulfill DSM-IV-TR criteria for PTSD three months after the traumatic event has occurred (APA, 2000). Previous research suggests that the prevalence of violence-related chronic PTSD recorded by studies using follow-up assessments after initial case identification one month postvictimisation approximates 10% to 15% (Winkel, 2007). However, individual studies have recorded much higher prevalence rates, with percentages as high as 25% among victims recruited through police stations (Wohlfarth, Winkel, & Van den Brink, 2002) and 32% in emergency department samples (Birmes et al., 2003). In addition, many studies have, either prospectively (e.g., Kilpatrick et al., 2003; Lawyer, Ruggiero, Resnick, Kilpatrick, & Saunders, 2006) or retrospectively (Orth, Cahill, Foa, & Maercker, 2008), investigated PTSD symptom severity in victims of rather diverse categories of interpersonal violence with varying intervals between time of victimisation and PTSD assessment.
27
equipped with the diagnostic skills necessary to adequately act as gatekeepers in the referral process (e.g., Winkel, Wohlfarth, & Blaauw, 2003). Recently, several short screening instruments, such as the RISK10 (Winkel, 2000), the RISK11 (Winkel, Wohlfarth, et al., 2003; Winkel, Wohlfarth, & Blaauw, 2004; Wohlfarth et al., 2002), and the Trauma Screening Questionnaire (TSQ; Brewin et al., 2002), have been developed. Each of these screeners can be quickly and easily administered among victims of crime to estimate vulnerability for PTSD-related symptomatology shortly after victimisation.
However, despite the development of adequate screening tools, many victims highly in need of psychological help remain unidentified during the early coping stages. Several reasons have been suggested for the problem of underidentification. First, screening for psychological problems and referral to victim support organisations or other social service agencies by the police (and primary care services available to victims) appears to function rather haphazardly (Brienen, Groenhuijsen, & Hoegen, 2000) and is only incidentally based on screening examinations. Presumably, many police departments do not dispose of sufficient financial and organisational resources to incorporate psychology-based screening instruments into their existing working practices. Second, many victims do not report their cases to the police (e.g., Van Dijk, 2008) and do not seek or need medical care for physical injuries (e.g., Logan, Evans, Stevenson, & Jordan, 2005). Third, some victims do not develop psychological health problems until several months have passed since the traumatic victimisation experience. This phenomenon is often referred to as delayed distress (e.g., Yoshihama & Horrocks, 2003). Victims with late onset of trauma-related psychopathology may fail to report symptom levels that qualify them for referral during the initial recovery phase.
Given the variety of reasons that underlie the problem of underidentification, policy measures focussing on the early stages of mental illness development cannot be expected to prevent all cases of chronic PTSD or other forms of persistent psychopathology, even if implemented properly. Therefore, in addition to the uptake of primary and secondary prevention strategies (i.e., activities initiated to prevent the onset or to interrupt the progression of disease), information and referral services should also be provided to victims who have been overlooked at first, but are (still) highly in need of professional help and assistance (i.e., tertiary prevention strategies) many years postvictimisation.
28
role in referring victims with current psychological problems to appropriate treatment services, since they are likely to maintain relations with the victim involved until all compensation possibilities have been exhausted. Support services automatically provided to victims who contacted the police or emergency services may have ended by that time, while those who failed to do so may also hesitate to seek help afterwards (e.g., McFarlane, Soeken, Reel, Parker, & Silva, 1997). Previous research suggests that victims of violence compensated for damages may still suffer from PTSD several years after victimisation, although prevalence rates in this population have not been estimated yet. Recently, for example, Orth, Montada, and Maercker (2006) investigated the association between feelings of revenge and current PTSD symptom levels among victims who had received financial support from a German victim assistance association for legal costs made within the past five years. PTSD symptom levels were measured with the 22-item Impact of Event Scale (Weiss & Marmar, 1997). The reported mean symptom scores (M ≥ 20 for each symptom cluster) seem to indicate that many victims still suffered from PTSD symptom levels that lay well above the diagnostic cut-off (IES total score ≥ 35) used by several prior studies (e.g., Knipscheer & Kleber, 2006; Mooren, De Jong, Kleber, & Ruvic, 2003). In addition, the study’s results suggest that rather objectively assessable criteria, such as time elapse since victimisation and physical harm resulting from the act of violence, may predict persistent PTSD symptomatology. To further explore this topic, the present paper addressed the potential utility of file characteristics as predictors of chronic PTSD among victims applying for state compensation with the Dutch Victim Compensation Fund (DVCF).
29
clinical psychologist or psychiatrist. To date, this possibility seems to have been overlooked in the recent literature on victims of violence. A large advantage of using file characteristics to screen for PTSD is that existing working practices do not require substantial adaptation.
Following from the aforementioned, our study had two objectives. First, we intended to assess the prevalence of PTSD among a sample of victims who had filed a claim for state compensation with the DVCF. Although we did not have prior expectations with regard to the exact prevalence to be found, we expected that the rate of PTSD cases would be rather high. Several studies on accidental trauma suggest that victims initiating litigation suffer from more severe PTSD symptoms than nonlitigants (e.g., Blanchard et al., 1998; Fujita & Nishida, 2008; Harris, Young, Rae, Jalaludin, & Solomon, 2008). As the fund serves as an ‚ultimum remedium (...)‛ (Groenhuijsen, 2001, p. 931), many victims who apply for compensation are likely to be on the verge of despair. Consequently, PTSD rates may be assumed to be higher than among victims of violence not seeking compensation from the fund. Second, we wanted to explore whether current PTSD could be predicted from data available in the fund’s electronic archives. Since we did not have prior knowledge of which data are included in the fund’s files, we did not speculate on the predictive value of individual file characteristics.
2.2 Methods
2.2.1 Procedure and participants
Identification of file information suitable for analysis
Prior to participant recruitment, the fund’s electronic database was consulted to determine which file data are systematically recorded and are appropriate for quantative evaluation. The selected data were broadly categorised into four domains of information. Each domain comprised a set of predictors considered to be relevant for the identification of PTSD. Relevance was based on existing literature.
30
Kilpatrick, Dansky, Saunders, & Best, 1993), while only a few have explicitly tested differences in effects on health and thereby failed to observe that health problems particularly occur in a subgroup of victims (e.g., Basile, Arias, Desai, & Thompson, 2004; Coker, Smith, McKeown, & King, 2000). Time since traumatisation may either be negatively (e.g., Ulvik, Kvale, Wentzel-Larsen, & Flaaten, 2008) or positively (e.g., Whiting & Bryant, 2007) correlated with symptomatology, but has also been found to be irrelevant in the prediction of symptom levels, particularly when several years since trauma have passed (e.g., Thapa, Van Ommeren, Sharma, De Jong, & Hauff, 2003). Furthermore, positive associations between time since trauma and negative outcomes may depend on the number of prior trauma (e.g., Steel, Silove, Phan, & Bauman, 2002) or, in the case of violent victimisation, on concurrent feelings of revenge (e.g., Orth et al., 2006).
The second domain comprised information related to the perpetrator: acquaintance with the perpetrator and perpetrator conviction. Both may be assumed to be associated with victimisation-related PTSD. Some empirical evidence with regard to the predictive value of the first factor was provided by Ullman (2007), who found that adult survivors of child sexual abuse by relatives had more PTSD symptoms than stranger victims. On the other hand, conviction of the perpetrator marks the public recognition of the victim as the victim of a criminal offence (cf. Orth, 2002, 2003) and therefore may modify posttraumatic stress reactions. Preliminary evidence for this line of reasoning was reported by Orth (2004). He observed a negative association between satisfaction with perpetrator punishment severity (with no conviction as the least severe form of punishment) and victims’ feelings of revenge (e.g., Orth, 2004) - a potential correlate of PTSD in victims of crime (e.g., Orth et al., 2006).
The third domain regarded medical care costs made by the victim: costs spent on victimisation-related hospitalisation, other physical health care services, and psychotherapy. Several studies have found positive associations between health care costs and PTSD symptoms (e.g., Walker et al., 2003; New & Berliner, 2000).
31
Participant recruitment
Participants were recruited through the DVCF. Victims of violent offences can apply to the fund for a single payment if they have suffered physical or immaterial damage (i.e., pain and suffering) and cannot be compensated through other means. In principle, victims need to file their claim within three years of victimisation or at a later moment, if this is justified by circumstances not reasonably attributable to the victim (Wet Schadefonds Geweldsmisdrijven van 1975). As the DVCF primarily serves as a last resort for compensation after all other possibilities have been exhausted, many victims do not apply for compensation until considerable time since victimisation has passed.
Data collection consisted of two phases. The first phase took place from October to December 2007 (Time 1; T1) and consisted of administration of a set of questionnaires measuring current PTSD symptom levels and recalled peritraumatic distress. Recollections of distress during and shortly after victimisation were assessed to ensure the traumatic nature of the act of violence. According to the DSM diagnostic criteria, to qualify as a traumatic event, the person in question must have experienced, witnessed, or been confronted with an event or events that involved actual or threatened death or serious injury, or a threat to the physical integrity of self or others (criterion A1), while the response to the event must have involved intense fear, helplessness, or horror (criterion A2). Since acts of severe violence fall within the DSM definition of a traumatic event, all participants were considered to fulfill criterion A1. By contrast, fulfillment of the A2 criterion can only be determined through explicit assessment of peritraumatic distress.
A letter that explained the purpose of the study was sent to applicants who had filed a claim between January 1st and December 31st 2006 and who were eligible for participation. Inclusion criteria were: age ≥ 18 and no missing file data on age, gender, and date of crime. Missing file data on these variables indicated that the victim involved had failed to return the application form after an initial request for sending a copy through the mail. The period of application chosen to select potential participants was prompted by the possibility that those who had applied at a later moment
had not received a decision on their claim yet or had objected against it.19 In
such cases self-reports of PTSD symptoms may reflect compensation neurosis.
Kennedy (1946, p. 20) cynically described compensation neurosisas ‚a state
of mind, born out of fear, kept alive by avarice,stimulated by lawyers, and
19 The electronic database used by the fund was not able to produce a reliable list of recently
32
cured by a verdict‛. Previous research, particularly in US war veterans, suggests that assessing PTSD during a compensation application procedure may induce symptom overreporting (e.g., Frueh et al., 2003; Frueh, Gold, De Arellano, 1997; Frueh, Hamner, Cahill, Gold, & Hamlin, 2000).
Potential participants were invited to fill out a survey through the internet or to request for a hardcopy version if they did not have access to the World Wide Web or preferred to fill out the questionnaire on paper. To test the stability of PTSD, respondents who had agreed to participate in a follow-up study were asked to fill out the PTSD measure again six months after initial assessment (Time 2; T2).
In total, 4513 victims had filed a claim with the DVCF during the reference period. Seven hundred and ninety-two of them did not fulfill inclusion criteria. Seven hundred and forty-four (20.0%) of those eligible for participation responded. Six hundred and forty participants filled out the initial questionnaire through the internet and 104 on paper. Six hundred and forty participants (551, 86.1% of internet participants; 89, 85.6% of paper participants) agreed to participate in the follow-up study. Two hundred and thirty-five (187, 33.9% of internet participants; 48, 46.2% of paper participants) of them responded. One participant who had requested for a paper version of the initial questionnaire filled out the follow-up questionnaire through the internet.
Only respondents without missing values were included in statistical analyses. Missing data on study variables were not estimated using statistical imputation procedures, since the program running the electronic questionnaires did not allow participants to skip a particular question and to continue with the next one. Consequently, the pattern of missingness for successive questions could not be investigated. Six hundred and eighty-six (92.2%) respondents did not have missing values on any of the study variables at T1. Two hundred and eight (88.5%) respondents neither had missing data at T1 nor at T2. Approval for the study was obtained from DVCF Committee.
33
The rate of females among T2 respondents (131/208, 63.0%) was higher than in T1 respondents (246/478, 51.5%) and nonrespondents (1457/3035, 48.0%). The difference in rates of females between T1 respondents and nonrespondents was not significant. No differences between the three groups were observed on time since victimisation. Mean time since victimisation was 4.7 years (SD = 3.5) for nonrespondents, 4.9 years (SD = 3.7)
for T1 respondents, and 5.3 years (SD = 4.8) for T2 respondents.20
2.2.2 Measures
File characteristics
Domain 1. Type of violence was operationalised by reducing the number of legal classifications of violent acts encountered in the files from 30 to 5: sexual violence, physical assault (severe), physical assault (moderate), theft
with violence, and other (reference category).21 Time since victimisation was
measured by computing the number of years since date of victimisation. Domain 2. Acquaintance with the perpetrator and conviction of the perpetrator indicated the existence of any kind of relationship between the victim and the perpetrator (yes/no) and the imposition of a sentence (yes/no), respectively.22
Domain 3. Requested compensation for costs of hospitalisation, other physical health care services, and psychotherapy were used as indicators of medical costs.
Domain 4. Compensation denial (yes/no) and compensation for pain and suffering regarded the initial outcome of the application procedure. A small number of victims may have received additional payments after a successful
20 Note that the number of background variables on which respondents and nonrespondents are
compared slightly differs for some studies included in this thesis. As mentioned in the introduction, chapters were only slightly adapted to retain their original content.
21 The DCVF categorises type of violence according to their legal classification used in the Dutch
Penal Code (DPC). To enable statistical testing, the number of different categories was reduced from 30 to 5. Severe and moderate physical assault and theft with violence corresponded to the original file categorisation. Sexual violence included all individuals that had experienced an offence falling under Book 2, Title XIV of the DPC. The remainder of the sample is a mixture of offences that were too low in number to form a category of their own.
34
complaint against compensation denial. Five dummy variables were created to represent compensation for pain and suffering. The dummies represented the different scale categories of financial compensation that can be granted for pain and suffering by the DVCF. Dummy 1 included participants who had at least been granted the lowest level of compensation (level 1), while dummies 2, 3, 4, and 5 included participants who had at least been granted level 2, 3, 4, or 5, respectively. No compensation for pain and suffering was the reference category.
Peritraumatic Distress Inventory
The 13-item Peritraumatic Distress Inventory (PDI, Brunet et al., 2001) was used to retrospectively check whether participants fulfilled criterion A2 with regard to the act of violence they had experienced. The psychometric properties of the PDI have been investigated in a sample of police officers exposed to critical incidents, which included both victims of physical and sexual assault. It is has been found to be internally consistent, with good test-retest reliability and good convergent and divergent validity (Brunet et al., 2001). As a Dutch version of the PDI was not available, its items were translated by the first author. The PDI was assessed using a 5-point Likert scale (0 = not at all, 1 = slightly, 2 = somewhat, 3 = very true, 4 = extremely true). Participants were considered to fulfill criterion A2 if they had reported a score of ≥1 on the PDI. In our sample, a good internal consistency reliability for the PDI was observed (α = .87).
PTSD Symptom Scale, Self-Report version
35
times or more) in relation to the event under investigation. The psychometric properties of the PSS-SR have been found to be statisfactory in crime victim samples (Wohlfarth, Van den Brink, Winkel, & Ter Smitten, 2003; Foa et al., 1993). In the current study, the PSS-SR demonstrated good internal consistency reliabilities, bot at T1 and at T2, with Cronbach’s α = .95 at T1 and .94 at T2 for the total scale, α = .90 at T1 and .91 at T2 for the intrusion subscale, α = .88 at T1 and .88 at T2 for the avoidance subscale, and α = .83 at T1 and .80 at T2 for the hyperarousal subscale. Comparable with procedures used by previous studies (e.g., Birmes, Brunet, Coppin-Calmes, et al., 2005), we classified participants who met criterion A2 as measured with the PDI and had a PTSD symptom score of ≥15 on the PSS-SR as suffering from probable PTSD. This cut-off score was derived from a study that evaluated the suitability of the PSS-SR as a screener for PTSD and corresponded to the best possible sensitivity (90%) coupled with the best possible specificity (90%) among victims of crime (Wohlfarth et al., 2003).
2.2.3 Statistical analyses
36
testing, as this is not required for exploratory research (Bender & Lange, 2001). All statistical tests were performed using the software package SPSS 15.0 for Windows (SPSS Inc., Chicago, Illinois).
2.3 Results
2.3.1 Self-reported PTSD
Three hundred and thirty-three (48.5%) T1 respondents had PSS-SR and PDI total scores corresponding to probable PTSD (M = 28.6, SD = 9.8). Mean PSS-SR subscale scores for individuals with PTSD were 7.8 (SD = 3.8) on the intrusion subscale, 11.2 (SD = 4.8) on the avoidance subscale, and 9.7 (SD = 3.0) on the hyperarousal subscale. Mean PSS-SR scores for non-PTSD participants were 6.2 (SD = 4.5) on the total scale, 1.6 (SD = 1.8) on the intrusion subscale, 1.8 (SD = 1.9) on the avoidance subscale, and 2.9 (SD = 2.4) on the hyperarousal subscale.
2.3.2 Stability of self-reported PTSD
The kappa coefficient for the reliability of violence-related PTSD at T1 and T2 was 0.56 (p < .001) and could be considered moderate according to the rules of thumb developed by Landis and Koch (1977). Violence-related PTSD (yes versus no) among 46 subjects (22.1% of the entire group and 38% of all violence-related cases of PTSD identified at T1 and T2) were discrepant between T1 and T2. PSS-SR total scores at T1 and T2 were highly correlated, r = .78, p < .001, indicating good test-retest reliability over a six-month period. Correlations between subscale scores were also very high, with r = .88 (p < .001) for the intrusion subscales, r = .92 (p < .001) for the avoidance subscales, and r = .87 (p < .001) for the hyperarousal subscales. Significant decreases in PSS-SR scores were observed between T1 and T2, with t = 3.69 (p < .001) for total symptom scores, t = 9.54 (p < .001) for intrusion scores, t = 11.04 (p < .001) for avoidance scores, and t = 14.79 (p < .001) for hyperarousal scores, indicating that the recovery process had still not been completed several years after victimisation. Prevalence of PTSD at T2 was 48.1%. McNemar’s test revealed that the prevalence rates at both assessments did not differ from each other (χ² = 1.07, df = 1, p = .302).
2.3.3 Relationships between claim characteristics and self-reported PTSD
37
were related to PTSD. Hierarchical analysis showed that predictors within domains 1, 2, and 3 contributed to a better fit of the model. Adding domain 4 variables to the equation did not improve the prediction of PTSD. When demographic and file characteristics were considered simultaneously, yet without taking into account the effect of recalled peritraumatic distress severity, age (Wald = 22.21, OR = 1.03, CI 95% 1.02 - 1.04, p < .001), female sex (Wald = 5.10, OR = 1.52, CI 95% 1.06 - 2.18, p = .024), time since victimisation (Wald = 5.69, OR = 1.08, CI 95% 1.01 - 1.15, p = .017), acquaintance with the perpetrator (Wald = 15.56, OR = 2.33, CI 95% 1.53 - 3.54, p < .001), violence-related hospitalisation costs (Wald = 7.93, OR = 2.28, CI 95% 1.28 - 4.04, p = .005), and compensation for pain and suffering ≥ 5 (Wald = 4.43, OR = 1.98, CI 95% 1.05 - 3.73, p = .035) appeared to be the only independent predictors of PTSD. This means that the odds for having PTSD increased by 3% with each age year and by 8% with each year time elapse since victimisation. The odds for female sex indicate that females were 1.5 times more likely to have
PTSD than males.23 The odds for acquaintance with the perpetrator suggest
that victims who had been acquainted to the perpetrator were 2.33 times more likely to suffer from PTSD than stranger victims. The odds for violence-related hospitalisation costs appear to imply that victims who had claimed costs for hospitalisation were 2.28 times more likely to have PTSD than those who had not claimed such costs. Finally, the odds for compensation for pain and suffering ≥ 5 seem to entail that victims who had been granted a compensation level of at least 5 were almost twice as likely to qualify for PTSD compared to those granted with lower compensation levels.
Prior to entry of recalled peritraumatic distress, the model explained between 14% and 19% of the pseudo variance in PTSD (Cox & Snell R² = 0.140; Nagelkerke R² = 0.187), while the Hosmer & Lemeshow goodness-of-fit test indicated that the model goodness-of-fitted the data well (χ² = 3.638, df = 8, p = 0.888). The classification table showed that 65.2% of the cases were classified correctly by the model (60.7% for participants with PTSD and 69.4% for participants without PTSD), which is more than a 50% random chance classification.
A substantial correlation was found between recalled peritraumatic distress and PTSD (r = 0.61, p < .001), implying that both outcomes were closely related in the present study, but not identical. Since the value for the bivariate correlation did not exceed the upper boundary of .7 as proposed by Tabachnick and Fidell (1996), entry of recalled peritraumatic stress into the model was deemed appropriate. The final step yielded an enormous
38
improvement of the fit of the model over the model without recalled peritraumatic distress and included five independent predictors of current PTSD: age, time since victimisation, acquaintance with the perpeptrator, violence-related hospitalisation costs, and recalled peritraumautic distress severity. Female sex and compensation for pain and suffering ≥ 5 failed to reach significance after adjustment for recalled peritraumatic distress. The pseudo variance in PTSD after adjusting for recalled peritraumatic distress reactions lied between 36 and 48% (Cox & Snell R² = 0.358; Nagelkerke R² = 0.477). The value for the Hosmer and Lemeshow goodness-of-fit test still indicated an acceptable fit of the data χ² = 10.989, df = 8, p = 0.202). The total model classified 77.4% of participants correctly (75.4% for participants with PTSD and 79.3% for participants without PTSD).
2.4 Discussion
