European Psychiatry 40 (2017) 88–95

Contents lists available at ScienceDirect

European Psychiatry journal homepage: http://www.europsy-journal.com

Original article

Posttraumatic stress and youth violence perpetration: A population-based cross-sectional study M. Aebi a,b,c,*, M. Mohler-Kuo d, S. Barra a, U. Schnyder e, T. Maier f, M.A. Landolt g,h a

Department of Forensic Psychiatry, University Hospital of Psychiatry, Neptunstrasse 60, 8032 Zurich, Switzerland Department of Child and Adolescent Psychiatry and Psychotherapy, University Hospital of Psychiatry, Neumu¨nsterallee 3, 8032 Zurich, Switzerland c Department of Clinical Psychology with Children/Adolescents & Families/Couples, Institute of Psychology, University of Zurich, Attenhoferstrasse 9, Zurich, Switzerland d Epidemiology, Biostatistics and Prevention Institute, University of Zurich, Hirschengraben 84, 8001 Zurich, Switzerland e Department of Psychiatry and Psychotherapy, University Hospital Zurich, University of Zurich, Culmannstrasse 8, 8091 Zurich, Switzerland f Psychiatric Services of the County of St. Gallen North, Zurcherstrasse 30, 9501 Wil, Switzerland g Division of Child and Adolescent Health Psychology, Department of Psychology, University of Zurich, Binzmuehlestrasse 14/8, 8051 Zurich, Switzerland h Department of Psychosomatics and Psychiatry, University Children’s Hospital Zurich, Steinwiesstrasse 75, 8032 Zurich, Switzerland b

A R T I C L E I N F O

A B S T R A C T

Article history: Received 13 July 2016 Received in revised form 23 August 2016 Accepted 23 August 2016 Available online 16 December 2016

Background: Exposure to trauma was found to increase later violent behaviours in youth but the underlying psychopathological mechanisms are unclear. This study aimed to test whether posttraumatic stress disorder (PTSD) is related to violent behaviours and whether PTSD symptoms mediate the relationship between the number of trauma experiences and violent behaviours in adolescents. Method: The present study is based on a nationally representative sample of 9th grade students with 3434 boys (mean age = 15.5 years) and 3194 girls (mean age = 15.5 years) in Switzerland. Lifetime exposure to traumatic events and current PTSD were assessed by the use of the University of California at Los Angeles Posttraumatic Stress Disorder Reaction Index (UCLA-RI). Logistic regression was used to assess associations between PTSD and violent behaviours, and structural equation modelling (SEM) was used to examine the meditation effects of PTSD. Results: PTSD (boys: OR = 7.9; girls: OR = 5.5) was strongly related to violent behaviours. PTSD symptoms partially mediated the association between trauma exposure and violent behaviours in boys but not in girls. PTSD symptoms of dysphoric arousal were positively related to violent behaviours in both genders. Anxious arousal symptoms were negatively related to violent behaviours in boys but not in girls. Conclusions: In addition to trauma, posttraumatic stress is related to violent outcomes. However, specific symptom clusters of PTSD seem differently related to violent behaviours and they do not fully explain a trauma-violence link. Specific interventions to improve emotion regulation skills may be useful particularly in boys with elevated PTSD dysphoric arousal in order to break up the cycle of violence.

C 2016 Elsevier Masson SAS. All rights reserved.

Keywords: Trauma Aggression PTSD Dysphoric arousal Anxious arousal Numbing

1. Introduction With prevalence rates between 60–92%, exposure to family or community violence (victimisation and witnessing) is common among juvenile offenders [1–4]. Persistent and repeated adverse childhood experiences were found to have a serious impact on brain functioning [5] and to negatively affect a child’s mental health [6], but also to increase the risk for later violent behaviours

* Corresponding author. University Hospital of Psychiatry, Neptunstrasse 60, 8032 Zurich, Switzerland. Tel.: +41 43 556 40 13; fax: +41 43 556 40 41. E-mail address: [email protected] (M. Aebi).

[7,8]. Not only interpersonal forms of violence but also other forms of trauma (e.g. natural disasters) seem to increase aggressive behaviours in youth [9,10] and more diverse forms of adverse experiences may have a cumulative effect on violent perpetration [11]. Little research has addressed the developmental mechanisms of trauma exposure and subsequent violent perpetration by addressing mental health disorders [9]. In adult veterans, posttraumatic stress disorder (PTSD) was consistently found to predict later violent behaviours [12,13]. In youth, PTSD has not been specifically investigated as a risk factor for violent behaviours. However, some recent findings suggest that PTSD is of importance when

http://dx.doi.org/10.1016/j.eurpsy.2016.08.007 C 2016 Elsevier Masson SAS. All rights reserved. 0924-9338/

Downloaded from ClinicalKey.com at Drexel University February 18, 2017. For personal use only. No other uses without permission. Copyright ©2017. Elsevier Inc. All rights reserved.

M. Aebi et al. / European Psychiatry 40 (2017) 88–95

understanding youth violent perpetration. For example, PTSD was reported by 11–32% of adolescent detainees [1,14] but only by 4% of community adolescents [15]. Furthermore, the number of PTSD symptoms was related to adolescents’ antisocial behaviours [16,17]. Finally, some studies tested the role of PTSD as a mediator between trauma exposure and violent behaviours in youth, but revealed rather inconsistent findings [10,18–22]. A study among 1358 urban US youth found that PTSD partially mediated the relationship between violence exposure and violence perpetration in boys but not in girls [20]. Except for the study by Ruchkin et al. [20], most of the previous studies on PTSD and its relation to youth violence: (a) were based on small and non-representative samples with limited generalizability of the findings, (b) did not consider multiple forms of trauma experiences, (c) did not use comprehensive statistics to test mediation, and (d) did not control for other trauma-related covariates such as other mental health problems and substance use. Therefore, additional studies using large community samples that include assessments of multiple potentially traumatic events are necessary to overcome some of the limitations mentioned above. The present study had two aims: First, to test whether PTSD is associated with an increased presence of violent behaviours in a nationally representative sample of adolescent boys and girls. Secondly, to test whether PTSD symptoms fully or partially mediate the relationship between a cumulative trauma score and committed violent behaviours. Given the previous findings on a cumulative effect of different adverse childhood experiences [11,23], we assumed a dose-response relationship between the number of trauma experiences and violence perpetration. In addition, the specific associations of PTSD symptoms of reexperiencing, avoidance, numbing, dysphoric arousal and anxious arousal were analysed in regard to the presence of violent behaviours while controlling for mental health problems and substance use (i.e., alcohol, cannabis, and hard drugs). 2. Method 2.1. Participants and procedures The present study was based on a nationally representative sample of 9th grade students attending public schools in Switzerland. From September 2009 to May 2010, data were anonymously collected using computer-assisted self-report questionnaires. Study design and procedures were published previously in detail [15,24,25]. The final sample consisted of 177 schools with 449 classes. Due to the absence of some students because of illness (n = 537) and students’ individual refusals (n = 63), 6841 students participated in the survey (response rate = 91.9%). Due to computer-related problems, 15 questionnaires subsequently were lost and 39 questionnaires excluded because of invalid data. A further 159 participants were excluded because of missing data on mental health problem scales (n = 122, see below) and/or violent behaviour items (n = 37). Therefore, the final sample consisted of 6628 participants (total response rate 89.1%; 3434 boys, 3194 girls). The study was approved by ethics committees and departments of education from 25 cantons and 177 schools that were included in the present study (lead ethic committee of the Canton Zurich, ref. 54/08). All students who were included in the study provided informed consent. 2.2. Measures 2.2.1. Demographic variables Parental education was assessed using two items measuring paternal and maternal education on 5-point Likert scales ranging from 1 (= no vocational training) to 5 (= a university degree).

89

Paternal and maternal scores were added up to create a single summation score. If one parent’s education level was unknown, the other parent’s score was doubled. If education level was reported for neither parent, the score was coded as ‘missing’. Parental education was categorised as ‘low’ if the summation score had a value of two or three. Living area was coded as ‘urban’ if the participant was living in a community with more than 50,000 residents. 2.2.2. Trauma and PTSD Trauma exposure and PTSD symptoms were assessed by a computerised form of the Adolescent Version of the University of California Los Angeles PTSD Reaction Index (UCLA-RI), a widely used self-report tool to assess PTSD in adolescence [26]. Trauma exposure according to DSM-IV criteria was assessed through 13 items with a yes/no answer format (more details on the frequency and type of trauma events is given in a previous publication based on the same sample [15]). The presence of PTSD was coded according to DSM-IV criteria including the DSM-IV requirement of functional impairment (see [15] for further description). A cumulative trauma exposure score was built based on the presence of all 13 items reflecting the number of potentially traumatic events (PTE). PTSD symptoms from DSM-IV clusters reexperiencing (criterion B), avoidance (criterion C) and hyperarousal (criterion D) were rated on a 5-point Likert scale ranging from 0 (= none of the time) to 4 (= most of the time). Based on our interest in dysphoric arousal and numbing, we further considered PTSD symptom clusters that have recently been found in factor analyses of the UCLA-RI (re-experiencing, avoidance, emotional numbing, dysphoric arousal and anxious arousal) [27]. PTSD symptom scores (frequency plus severity) were calculated (total score and subscale scores). The UCLA-RI has demonstrated excellent internal consistency and good convergent validity ([26] r = .82 compared to the Child and Adolescent Version of the Clinician-Administered PTSD Scale [28] and r = .70 compared to the PTSD module of the Schedule for Affective Disorders and Schizophrenia for School-Age Children [29]). The internal consistency of the UCLA-RI in the present study was similar to the figures reported in the original manual (Cronbach’s a = .87). 2.2.3. Violent behaviour Violent behaviour was assessed using self-reports on four dichotomous items of physical and three dichotomous items of sexual violence. Participants were asked if they had ever (a) physically assaulted/attacked another person, (b) committed a robbery, (c) threatened someone with a weapon, (d) used a weapon to hurt someone, (e) forced someone else to undress himself/herself, (f) touched someone else’s private parts against their will and (g) forced someone else into sexual activities. Internal consistency of the violent behaviour scale was acceptable (Cronbach’s a = .63). As expected, the resulting scale was strongly skewed (skewness = 1.79, SE = 0.04) with the majority of youth reporting none or one violent behaviour (n = 5493, 82.9%). For logistic regression analyses, the violent behaviour scale was dichotomised with scores of two and higher considered as present. 2.2.4. Mental health problems Mental health problems were assessed by the self-report version of the Strengths and Difficulties Questionnaire (SDQ). This measure of mental health problems in children aged 11–17 contains 20 items relating to emotional problems, conduct problems, hyperactivity, and peer problems [30]. Reliability and validity of the instrument were found to be sufficient [31]. In the present analyses, scores falling within the clinical range defined in the manual were considered clinically relevant.

Downloaded from ClinicalKey.com at Drexel University February 18, 2017. For personal use only. No other uses without permission. Copyright ©2017. Elsevier Inc. All rights reserved.

90

M. Aebi et al. / European Psychiatry 40 (2017) 88–95

2.2.5. Substance use Frequency of alcohol consumption was reported on a 6-point Likert-type scale from ‘every day’ to ‘less than once per month’; two or more times a week was chosen to indicate ‘regular alcohol use’ [32]. Participants were also asked if they had ever used cannabis or any hard drugs (e.g., cocaine, amphetamines, ecstasy, heroine, hallucinogens). 2.3. Statistical analysis Multiple logistic regressions were used to analyse the presence of PTSD and the presence of trauma exposure without PTSD as a predictor of violence (the statistical term predictor will be used for logistic regression although the present study is cross-sectional). Structural equation modelling (SEM) procedures with robust maximum likelihood estimation (that takes into account nonnormal distributions of PTE and violence) were used to analyse the mediating role of PTSD. Comparable to other studies [33], the cumulative trauma score was included as a directly observed variable in the SEM, whereas PTSD and violent behaviours were defined as latent constructs. Separate analyses were conducted for boys and girls. Three different goodness of fit indicators (GFI) were assessed, i.e., the Standardised Root Mean Square Residual (SRMR) as an indicator of the unexplained co-variances of the model, the Root Mean Square Error of Approximation (RMSEA) which includes a parsimony correction, and the Comparative Fit Index (CFI) for evaluating the hypothesised model compared to a null model. Acceptance of any model was based on the following cut-offs: SRMR < .05, RMSEA < .05 and CFI > .95. Furthermore, the SatorraBentler scaled –x2difference test and the log likelihood test for nested models were used when comparing a direct effect model between trauma-violence behaviour (assuming PTSD is not related to violent behaviours) with a PTSD mediated model (where PTSD was additionally allowed to predict violent behaviours). The mediating role of PTSD was additionally re-examined using logistic regression and bootstrapping (using PROCESS model 4; see [34]). Multivariate logistic regression analyses were performed to evaluate the contribution of specific PTSD symptom clusters (subscores of re-experience, avoidance, numbing, dysphoric arousal and anxious arousal; see [27]) to violent behaviours while controlling for demographics, trauma, mental health problems and substance use. None of the predictors/covariates showed multicollinearity. All analyses were conducted using Mplus 7.31 [35] and SPSS 23 for Mac OS X, were two-sided, and considered a P = .05 as significant.

trauma, PTSD and violent behaviours, and Appendix B for frequencies of violent behaviours in girls and boys). 3.2. PTSD and exposure to PTE without PTSD as predictors of violent outcomes Multiple logistic regressions revealed that PTSD and to a smaller degree also exposure to PTE without PTSD were significantly associated with committed violent behaviours in boys and girls while controlling for demographics (Table 1). Comparable odds ratios (OR) were found for boys and even higher OR were found for girls (see Appendix C) in more severe forms of violence (presence of item c, d or g). 3.3. PTSD symptoms as mediators between trauma exposure and violent behaviours In order to test a mediation effect of PTSD we first established a direct effect model with the cumulative trauma score that predicted PTSD and violent behaviours. In this model, the path from PTSD to violent behaviours was constrained to zero assuming that PTSD was not directly related to violent behaviours. Secondly, we established a PTSD mediated model that additionally allowed PTSD to predict violent behaviours (Fig. 1). Changes in model fit were analysed to test the mediation hypothesis. All models in the male and the female sample fit the data well. However, some relevant differences were found between the examined models: For boys, the mediated model showed increased fit to the data compared to the direct effect model according to all goodness of fit indicators, the Satorra-Bentler test and the log likelihood test (Table 2). In contrast, the PTSD mediated model in the girls sample did not fit better than the direct effect model according to the BIC, sample size adjusted BIC, the RMSEA and the Log likelihood test (Table 2). The PTSD mediated models with standardised path coefficients are presented in Fig. 1 for boys and girls (in italic). In boys and girls the cumulative trauma scores were significantly associated with PTSD as well as with violent behaviours. However, only in boys but not in girls PTSD was significantly related to violent behaviours (boys: path coefficient = .13, P < .01 vs. girls: path coefficient = .07, P = n.s.). The findings of the SEM were confirmed by the use of logistic regression and bootstrapping [34]. PTSD significantly mediated the relationship between trauma exposure and the presence of violence in boys (effect = 0.029, 95%CI = 0.013–0.045, P < .01) but not in girls (effect = 0.030, 95%CI = –0.068 to 0.069, P = n.s.). Statistical power was found sufficient (see Appendix D).

3. Results 3.1. Descriptive findings The sample consisted of 6628 youth with a mean age of 15.50 years (SD = 0.66 years, range 13.1 to 20.3 years) including 3434 boys (mean age = 15.52 years, SD = 0.67 years) and 3194 girls (mean age = 15.47 years, SD = 0.66 years). About a quarter (1697, 25.6%) were of non-Swiss nationality (boys 24.4%; girls 26.9%), 4922 (74.3%) were living in urban areas (boys 73.1%; girls 75.5%) and 1713 (25.8%) were from families with low parental education (boys 24.8%; girls 26.9%). Of the total sample, 56% (boys 55.7%; girls 56.6%) reported having had experienced at least one PTE in their lifetime and 4.4% currently met DSM-IV criteria for PTSD (boys 2.6%; girls 6.4%). For more details on PTE and PTSD, see Landolt et al. [15]. The presence of self-reported committed violent behaviours differed between genders (boys 26.1% vs. girls 7.5%, x2(1) = 404.79, P < .001). A subsample of 3742 youth who reported at least one PTE (1815 girls and 1927 boys) were included in the mediation analyses (see Appendix A for descriptive findings on

Table 1 Multivariate logistic regression analyses of trauma exposure without PTSD and trauma exposure with PTSD as predictors of committed violence in boys and girls. Variables

Covariates Age Urban vs. rural living area Low parental education Non-Swiss nationality Trauma/PTSD (reference group neither trauma nor PTSD) Presence of PTE without PTSD Presence of PTE with PTSD

Boys (n = 3434)

Girls (n = 3194)

OR (95% CI)

OR (95% CI)

1.23 0.91 1.06 1.26

(1.09–1.38)** (0.76–1.09) (0.88–1.28) (1.04–1.52)*

3.37 (2.82–4.03)*** 7.89 (5.03–12.37)***

1.05 0.98 0.99 1.48

(0.86–1.29) (0.71–1.35) (0.73–1.34) (1.10–1.98)**

2.70 (1.94–3.77)*** 5.53 (3.49–8.79)***

PTSD: Posttraumatic stress disorder; PTE: potentially traumatic event; OR: odds ratio; CI: confidence interval. * Significance (two-tailed), P < .05. ** Significance (two-tailed), P < .01. *** Significance (two-tailed), P < .001.

Downloaded from ClinicalKey.com at Drexel University February 18, 2017. For personal use only. No other uses without permission. Copyright ©2017. Elsevier Inc. All rights reserved.

M. Aebi et al. / European Psychiatry 40 (2017) 88–95

.53*** .46***

91

.46*** .34***

Sexual Violence

.65*** .66***

Physical Violence

Violence

Cumulative Trauma Score

.28*** .37***

.13** .07 n.s.

PTSD

.69*** .76***

Intrusion

.89*** .74***

Irritability

.78*** .80***

FIT Indices: = .99, .99 CFI = .01, .01 SRMR RMSEA = .03, .04

Avoidance

.

Fig. 1. Structural equation models testing PTSD as mediator between cumulative trauma and violent behaviours in boys and girls (in italic). SRMR: Standardised Root Mean Square Residual; RMSEA: Root Mean Square Error of Approximation; CFI: Comparative Fit Index.

Table 2 Fit statistics for structural equation models. Model Boys (n = 1927) Direct effect model PTSD mediated model Girls (n = 1815) Direct effect model PTSD mediated model

AIC

BIC

aBIC

SRMR

RMSEA

CFI

SB-test (direct vs. mediated)

LL-test (direct vs. mediated)

45905.62 45894.13

46011.33 46005.40

45950.97 45941.86

0.026 0.014

0.040 0.034

0.985 0.991

– 75.13***

– 10.04**

40236.25 40235.49

40340.82 40345.56

40280.46 40282.03

0.017 0.014

0.034 0.036

0.988 0.988

– 19.76***

– 1.62 n.s.

PTSD: posttraumatic stress disorder; AIC: Akaike information criterion; BIC: Bayesian information criterion; aBIC: sample size adjusted BIC; SRMR: standardised root mean square residual; RMSEA: Root Mean Square Error of Approximation; CFI: comparative fit index; SB-test: Satorra-Bentler test; LL-test: log likelihood test. ** Significance (two-tailed), P < .01. *** Significance (two-tailed), P < .001.

3.4. PTSD symptom clusters as predictors of violent behaviours in boys and girls We also tested if symptom clusters of PTSD were related to violent behaviours while controlling for demographics and the cumulative trauma score (model 1) as well as for substance use and mental health problems (model 2). In both genders, PTSD symptoms of dysphoric arousal were positively associated with the presence of violent behaviours. In boys but not in girls, symptoms of anxious arousal were negatively associated with the presence of violence behaviours while controlling for covariates (Table 3). Besides PTSD symptom clusters, the cumulative trauma score and the presence of conduct problems as well as the use of alcohol, cannabis and hard drugs were related to the presence of violence. 4. Discussion 4.1. PTSD and PTE without PTSD as correlates of self-reported violence perpetration Recently, the World Health Organization (WHO) noted in their 2015 report that youth violence is the 4th leading cause of death in

young people worldwide [36]. The present study aimed to improve our understanding of potential psychopathological mechanisms for youth violent perpetration. Youth reporting PTSD symptoms were significantly more likely to report engaging in violent behaviours compared to youth not exposed to PTE. These findings suggest that PTSD may be an equally important factor to understand violent behaviours in adolescents as previously shown in adult veterans [13]. 4.2. PTSD as a mediator of cumulative trauma and violent behaviour Independent of the statistical approach (SEM, regression analyses) PTSD symptoms were found to partially mediate the relationship between trauma exposure and violent behaviours in boys but not in girls. These findings suggest that boys who develop PTSD symptoms after exposure to PTE tend to commit more violent behaviours than those who do not. This finding mirrors and expands previous findings in a selective sample of urban community youth with a large proportion of minority adolescents [20]. As most previous studies lack rigorous statistical tests to test mediation, our finding is of particular interest. According to Steiner et al. [14], PTSD creates possible pathways for aggression by arousal driven interference of social activities, inappropriate

Downloaded from ClinicalKey.com at Drexel University February 18, 2017. For personal use only. No other uses without permission. Copyright ©2017. Elsevier Inc. All rights reserved.

M. Aebi et al. / European Psychiatry 40 (2017) 88–95

92

Table 3 Logistic regression analyses of PTSD symptom clusters as predictors of committed violence in a subsample of boys and girls who reported exposure to at least one potentially traumatic event. Boys (n = 1927)

Variables

Demographics and trauma exposure Age Urban vs. rural living area Low parental education Non-Swiss nationality Cumulative trauma exposure score PTSD symptom clusters PTSD Re-experiencing score PTSD Avoidance score PTSD Numbing score PTSD Dysphoric Arousal score PTSD Anxious Arousal score Substance use/mental health problems Alcohol (> 2+ times per week) Ever used cannabis Ever used hard drugs Clinical range of SDQ Conduct problems Clinical range of SDQ Emotional problems Clinical range of SDQ Hyperactivity Clinical range of SDQ Peer problems

Girls (n = 1815)

Multivariate modela

Multivariate modelb

Multivariate modela

Multivariate modelb

OR (95% CI)

OR (95% CI)

OR (95% CI)

OR (95% CI)

1.13 0.94 1.14 1.13 1.41

(0.97–1.32) (0.75–1.19) (0.90–1.44) (0.89–1.43) (1.32–1.51)***

1.08 0.90 1.17 1.16 1.32

(0.92–1.27) (0.70–1.15) (0.91–1.50) (0.90–1.50) (1.23–1.42)***

0.92 1.04 0.89 1.38 1.39

(0.73–1.17) (0.71–1.52) (0.62–1.27) (0.98–1.95) (1.28–1.52)***

1.02 0.96 0.85 1.45 1.30

(0.79–1.32) (0.65–1.43) (0.59–1.24) (1.00–2.08)* (1.18–1.42)***

1.01 0.92 1.00 1.25 0.84

(0.97–1.05) (0.86–0.99)* (0.97–1.04) (1.18–1.31)*** (0.79–0.90)***

1.03 0.95 0.99 1.16 0.90

(0.99–1.08) (0.88–1.03) (0.96–1.03) (1.10–1.23)*** (0.84–0.96)**

0.96 0.99 1.02 1.17 0.86

(0.91–1.00) (0.90–1.08) (0.98–1.07) (1.10–1.26)*** (0.79–1.94)**

0.96 0.99 1.03 1.10 0.93

(0.91–1.01) (0.90–1.09) (0.98–1.08) (1.02–1.18)* (0.84–1.02)

1.45 1.74 1.66 4.61 0.90 1.01 0.37

(1.10–1.91)** (1.36–2.23)*** (1.26–2.19)*** (3.37–6.31)*** (0.46–1.80) (0.69–1.49) (0.19–0.69)**

– – – – – – –

1.60 1.80 1.66 3.46 0.77 1.25 0.93

(1.01–2.52)* (1.23–2.63)** (1.11–2.48)* (2.24–5.32)*** (0.47–1.27) (0.77–2.03) (0.42–2.06)

– – – – – – –

PTSD: posttraumatic stress disorder; SDQ: strength and difficulties questionnaire; OR: odds ratios; CI: confidence interval. a Entering variables PTSD symptom clusters, demographic covariates, and cumulative trauma exposure. b Entering variables PTSD symptom clusters, demographic covariates, cumulative trauma exposure, and substance use/mental health problems. * Significance (two-tailed), P < .05. ** Significance (two-tailed), P < .01. *** Significance (two-tailed), P < .001.

hostile reactions to stressors and intrusion driven continuous reenactment. Additionally, social learning mechanisms may explain why boys with PTSD have a higher risk to act violently in social situations: Exposure to PTE leads to the development of symptoms that are found to sensitise the brain to similar situational cues [37]. Thus, youth with PTSD may become more prone to aggressive and violent behaviours in situations that resemble the initial trauma experience. 4.3. Specific PTSD symptom clusters and associations with violent behaviours Additional analyses of specific PTSD symptom clusters shed more light on the dysfunctional emotional processes of traumatised violent offenders. Dysphoric arousal was positively associated and anxious arousal was negatively associated with violence perpetration. The antagonistic role of dysphoric and anxious arousal symptoms within PTSD is probably one reason why some previous studies did not find that PTSD predicted aggression [18,19]. High levels of dysphoric arousal were previously reported in youth who show reactive forms of aggression [21,38]. These youth showed atypically elevated threat-circuitry responsiveness including elevated amygdala responses to threat [39]. Low levels of anxiety were related to violent youth with callousunemotional traits who showed more intended forms of aggression [39,40]. Interestingly, some researchers have suggested that callous-unemotional traits do not necessarily reflect an inherent personal disposition but also may arise through the result of repeated exposure to PTE and persistent maltreatment [41]. Detachment, unemotionality or callousness were often observed to accompany violent behaviours and were suggested to reflect trauma-related emotional numbing [42,43]. However, contrary to our hypothesis we did not find that numbing symptoms were specifically associated with violent behaviours in traumatised youth. Most probably, only specific symptoms of fear numbing are

associated with committed violence [42]. In sum, the present findings suggest that not PTSD as an entity but rather specific PTSD symptom profiles put an adolescent at risk for violent perpetration. 4.4. Cumulative trauma and substance use Our findings showed a cumulative effect of different intra- and extra-familial trauma on the commitment of violent behaviours and expanded previous studies that found adverse childhood experiences to affect the physical, psychological and social development throughout the course of the child’s development into adulthood [23]. Children and adolescents may successfully adapt to one trauma type, but adaptation to several types becomes more difficult. Although PTSD symptoms and particularly emotional arousal are important factors for understanding violent behaviours, these symptoms do not fully explain the link between trauma exposure and violent perpetration. The present study focused on mental health problems but other mechanisms such as deficient information processing (e.g. hostility bias) and the acceptance of violent cognitions [18,44] may further explain the violent exposure-perpetration link in youth. Furthermore and as supported by the present findings, the consumption of alcohol and drugs was strongly related to criminal and violent behaviours in adolescent boys and girls [45] and may serve as an inadequate coping strategy to handle PTSD symptoms [12]. 4.5. Gender differences This study found that PTSD symptoms seem to play a less specific role in the development of violent behaviours in girls compared to boys. In line with previous studies [46], we found a lower number of girls compared to boys showing serious violent behaviours with particularly lower rates of sexual violence. Furthermore, traumatised girls were more often found to show

Downloaded from ClinicalKey.com at Drexel University February 18, 2017. For personal use only. No other uses without permission. Copyright ©2017. Elsevier Inc. All rights reserved.

M. Aebi et al. / European Psychiatry 40 (2017) 88–95

93

behaviours. The current findings become an important challenge for mental health and child welfare institutions to serve in a preventive manner. Clinicians should screen youth who report multiple types of potentially traumatic events for symptoms of arousal. In addition, they should assess their committed violent acts in community and family environment youth. Specific interventions to improve emotion regulation skills (e.g. Trauma Affect Regulation: Guide for Education and Therapy; TARGET [49]) may be useful particularly in boys with elevated PTSD dysphoric arousal and reduced anxious arousal symptoms in order to break the cycle of violence.

self-directed and more covert forms of aggression [47] than physical violence. The existence of distinct pathways to mental health problems of trauma-exposed girls was supported by previous studies which found higher levels of multiple traumatisation and less specific psychopathology in female than in male adolescent offenders [48]. More specific research seems necessary to disentangle the relationship between trauma experience and specific forms of reactive and proactive aggression in male and female youth and adult samples. 4.6. Study strengths and limitations

Ethical standards

The present study was based on an anonymous survey of a large, nationally representative sample of youth attending schools in a Western country. The large number of participants allowed for extensive multivariate testing of multiple risk factors with sufficient statistical power. However, our survey was crosssectional, thus causal relationships between trauma exposure, PTSD and subsequent violent behaviours cannot be established. Furthermore, the present findings were based solely upon anonymous self-reports and are subject to possible recall bias and potentially confounding variables related to the relative willingness of participants to report trauma experiences, UCLA-RI based criteria of PTSD and violent behaviours. Heritable and familial factors that link trauma exposure and violent perpetration [46] were not considered in the present study.

The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. Financial Support The data presented in this paper were assessed in the context of the Optimus Study. The Optimus Study was initiated and funded by the UBS OPTIMUS FOUNDATION (http://www.optimusstudy.org). Disclosure of interest

5. Conclusions

The authors declare that they have no competing interest. Acknowledgements

Our findings suggest that there is a relationship between PTSD and self-reported youth violence perpetration. The association was stronger in boys than in girls. Furthermore, it was specific to PTSD symptoms of dysphoric arousal in both genders and of anxious arousal in boys. Nevertheless, most youth who reported extensive trauma exposure and PTSD symptoms did not report violent

The authors are grateful to all the students who participated in the study. We also thank Ursula Meidert from the Epidemiology, Biostatistics and Prevention Institute, University of Zurich, who helped with data collection.

Appendix A. Trauma, PTSD symptoms and violent behaviours (means, standard deviations) in male and female youth who reported at least one potentially traumatic event Variables

Trauma Cumulative trauma score PTSD symptoms PTSD symptoms score PTSD Re-experiencing score PTSD Avoidance score PTSD Numbing score PTSD Dysphoric Arousal score PTSD Anxious Arousal score Violent behaviour Violent behaviour scale

Boys (n = 1927)

Girls (n = 1815)

M (SD)

M (SD)

2.44 (1.74) 13.89 7.94 3.50 10.70 7.17 4.58

19.23 10.19 4.25 11.42 7.91 5.46

1.31 (1.32)

PTSD: Posttraumatic Stress Disorder. = significance (two sided), P < .05, ** = significance (two sided), P < .01,

t-test

2.36 (1.61)

(10.33) (3.35) (1.81) (3.92) (2.74) (1.79)

*

Total sample (n = 3742)

1.44 n.s. –14.74*** –17.46*** –11.5*** –5.56*** –7.92*** –14.37***

(11.77) (4.42) (2.16) (4.05) (2.96) (1.96)

22.95***

0.49 (0.82) ***

M (SD) 2.40 (1.68) 16.48 9.03 3.87 11.05 7.53 5.01

(11.37) (4.07) (2.02) (4.00) (2.87) (1.92)

0.92 (1.18)

= significance (two sided), P < .001.

Appendix B. Frequencies of violent behaviours in male and female youth Items

Non-sexual violence a) Physically assaulted/attacked another person b) Ever committed a robbery c) Ever threatened someone with a weapon d) Ever used a weapon to hurt someone Sexual Coercion e) Forced someone else to undress f) Touched private parts of someone else against his/her will

Boys (n = 3434)

Girls (n = 3194)

Frequencies (%)

Frequencies (%)

Chi2 test

Frequencies (%)

1529 1032 299 198

535 477 50 34

595.36*** 215.10*** 169.19*** 108.29***

2064 1509 349 232

(44.5%) (30.1%) (8.7%) (5.8%)

71 (2.1%) 202 (5.9%)

(16.8%) (14.9%) (1.6%) (1.1%)

17 (0.5%) 25 (0.8%)

Total sample (n = 6628)

29.78*** 130.11***

Downloaded from ClinicalKey.com at Drexel University February 18, 2017. For personal use only. No other uses without permission. Copyright ©2017. Elsevier Inc. All rights reserved.

(31.1%) (22.8%) (5.3%) (3.5%)

88 (1.3%) 227 (3.4%)

94

M. Aebi et al. / European Psychiatry 40 (2017) 88–95

Appendix B (Continued ) Boys (n = 3434)

Items

Girls (n = 3194)

Frequencies (%) g) Ever forced someone else into sexual activities *

= significance (two sided), P < .05,

**

= significance (two sided), P < .01,

Frequencies (%)

46 (1.3%) ***

15 (0.5%)

Total sample (n = 6628) 2

Chi test 13.73***

Frequencies (%) 61 (0.9%)

= significance (two sided), P < .001.

Appendix C. Multivariate logistic regression analyses of trauma exposure without PTSD and trauma exposure with PTSD as predictors of committed serious violence (presence of item c, d or g) in boys and girls Variables

Covariates Age Urban vs. rural living area Low parental education Non Swiss nationality Trauma/PTSD (reference group neither trauma nor PTSD) Presence of PTE without PTSD Presence of PTE with PTSD

Boys (n = 3434)

Girls (n = 3194)

OR (95% CI)

OR (95% CI)

1.29 1.14 0.90 1.68

(1.10–1.50)** (0.88–1.48) (0.70–1.16) (1.33–2.14)***

3.39 (2.60–4.43)*** 8.02 (4.79–13.44)***

1.03 1.29 0.99 1.14

(0.72–1.47) (0.71–2.36) (0.58–1.68) (0.69–1.93)

4.09 (1.98–8.49)*** 16.00 (7.10–36.07)***

PTSD: Posttraumatic stress disorder; PTE: potentially traumatic event; OR: odds ratio; CI: confidence interval. * = significance (two sided), P < .05, ** = significance (two sided), P < .01, *** = significance (two sided), P < .001.

Appendix D. Supplement description of power analysis to test mediation in the boy and girl samples Statistical power for PTSD mediation between cumulative trauma and the presence of violence was analysed with the package ‘PowerMediation’ [50] in R statistical software [51]. The resulting statistical power was .979 in the boy sample based on beta = .019 in the multivariate logistic regression (with n = 1927, standard deviation of PTSD = 10.32, prevalence of the outcome = 0.36, correlation between PTSD and cumulative trauma = .26, significance level alpha = .05). Sufficient statistical power was found in the girl sample (power = .801) to detect a smiliar effect (beta = .019 with n = 1815, standard deviation of PTSD = 11.77, prevalence of the outcome = 0.11, correlation between PTSD and cumulative trauma = .34, significance level alpha = .05).

References [1] Abram KM, Teplin LA, Charles DR, Longworth SL, McClelland GM, Dulcan MK. Posttraumatic stress disorder and trauma in youth in juvenile detention. Arch Gen Psychiatry 2004;61:403–10. [2] Ford JD, Chapman J, Connor DF, Cruise KR. Complex trauma and aggression in secure juvenile justice settings. Crim Justice Behav 2012;39:694–724. [3] Krischer MK, Sevecke K. Early traumatization and psychopathy in female and male juvenile offenders. Int J Law Psychiatry 2008;31:253–62. [4] Moore E, Gaskin C, Indig D. Childhood maltreatment and post-traumatic stress disorder among incarcerated young offenders. Child Abuse Negl 2013;37:861–70. [5] Teicher MH, Samson JA. Annual research review: enduring neurobiological effects of childhood abuse and neglect. J Child Psychol Psychiatry 2016;57: 241–66. [6] Alisic E, Zalta AK, van Wesel F, Larsen SE, Hafstad GS, Hassanpour K, et al. Rates of post-traumatic stress disorder in trauma-exposed children and adolescents: meta-analysis. Br J Psychiatry 2014;204:335–40. [7] Maas CD, Fleming CB, Herrenkohl TI, Catalano RF. Childhood predictors of teen dating violence victimization. Violence Vict 2010;25:131–49. [8] Wilson HW, Stover CS, Berkowitz SJ. Research review: the relationship between childhood violence exposure and juvenile antisocial behavior: a metaanalytic review. J Child Psychol Psychiatry 2009;50:769–79. [9] Maas C, Herrenkohl TI, Sousa C. Review of research on child maltreatment and violence in youth. Trauma Violence Abuse 2008;9:56–67. [10] Marsee MA. Reactive aggression and posttraumatic stress in adolescents affected by Hurricane Katrina. J Clin Child Adolesc Psychol 2008;37:519–29. [11] Duke NN, Pettingell SL, McMorris BJ, Borowsky IW. Adolescent violence perpetration: associations with multiple types of adverse childhood experiences. Pediatrics 2010;125:e778–86. [12] Elbogen EB, Johnson SC, Wagner HR, Sullivan C, Taft CT, Beckham JC. Violent behaviour and post-traumatic stress disorder in US Iraq and Afghanistan veterans. Br J Psychiatry 2014;204:368–75. [13] Trevillion K, Williamson E, Thandi G, Borschmann R, Oram S, Howard LM. A systematic review of mental disorders and perpetration of domestic violence among military populations. Soc Psychiatry Psychiatr Epidemiol 2015;50: 1329–46. [14] Steiner H, Garcia IG, Matthews Z. Posttraumatic stress disorder in incarcerated juvenile delinquents. J Am Acad Child Adolesc Psychiatry 1997;36:357–65. [15] Landolt MA, Schnyder U, Maier T, Scho¨nbucher V, Mohler-Kuo M. Trauma exposure and posttraumatic stress disorder in adolescents: a national survey in Switzerland. J Trauma Stress 2013;26:209–16.

[16] Becker SP, Kerig PK. Posttraumatic stress symptoms are associated with the frequency and severity of delinquency among detained boys. J Clin Child Adolesc Psychol 2011;40:765–71. [17] Fehon DC, Grilo CM, Lipschitz DS. A comparison of adolescent inpatients with and without a history of violence perpetration: impulsivity, PTSD, and violence risk. J Nerv Ment Dis 2005;193:405–11. [18] Allwood MA, Bell DJ. A preliminary examination of emotional and cognitive mediators in the relations between violence exposure and violent behaviors in youth. J Community Psychol 2008;36:989–1007. [19] Kimonis ER, Ray JV, Branch JR, Cauffman E. Anger mediates the relation between violence exposure and violence perpetration in incarcerated boys. Child Youth Care Forum 2011;40:381–400. [20] Ruchkin V, Henrich CC, Jones SM, Vermeiren R, Schwab-Stone M. Violence exposure and psychopathology in urban youth: the mediating role of posttraumatic stress. J Abnorm Child Psychol 2007;35:578–93. [21] Silvern L, Griese B. Multiple types of child maltreatment, posttraumatic stress, dissociative symptoms, and reactive aggression among adolescent criminal offenders. J Child Adolesc Trauma 2012;5:88–101. [22] Stimmel MA, Cruise KR, Ford JD, Weiss RA. Trauma exposure, posttraumatic stress disorder symptomatology, and aggression in male juvenile offenders. Psychol Trauma 2014;6:184–91. [23] Felitti VJ, Anda RF, Nordenberg D, Williamson DF, Spitz AM, Edwards V, et al. Relationship of childhood abuse and household dysfunction to many of the leading causes of death in adults. The Adverse Childhood Experiences (ACE) Study. Am J Prev Med 1998;14:245–58. [24] Aebi M, Landolt MA, Mueller-Pfeiffer C, Schnyder U, Maier T, Mohler-Kuo M. Testing the ‘‘sexually abused-abuser hypothesis’’ in adolescents: a populationbased study. Arch Sex Behav 2015;44:2189–99. [25] Mohler-Kuo M, Landolt MA, Maier T, Meidert U, Scho¨nbucher V, Schnyder U. Child sexual abuse revisited: a population-based cross-sectional study among Swiss adolescents. J Adolesc Health 2014;54:304–11. [26] Steinberg AM, Brymer MJ, Decker KB, Pynoos RS. The university of California at Los Angeles Post-traumatic Stress Disorder Reaction Index. Curr Psychiatry Rep 2004;6:96–100. [27] Elhai JD, Layne CM, Steinberg AM, Brymer MJ, Briggs EC, Ostrowski SA, et al. Psychometric properties of the UCLA PTSD reaction index. Part II: investigating factor structure findings in a national clinic-referred youth sample. J Trauma Stress 2013;26:10–8. [28] Newman E, Weathers FW, Nader K, Kaloupek DG, Pynoos RS, Blake DD, et al. Clinician-Administered PTSD Scale for Children and Adolescents (CAPSCA) – Interviewer’s Guide. Los Angeles: Western Psychological Services; 2004.

Downloaded from ClinicalKey.com at Drexel University February 18, 2017. For personal use only. No other uses without permission. Copyright ©2017. Elsevier Inc. All rights reserved.

M. Aebi et al. / European Psychiatry 40 (2017) 88–95 [29] Kaufman J, Birmaher B, Brent D, Rao U, Ryan N. Kiddie-SADS – Present and Lifetime Version (K-SADS-PL). University of Pittsburgh: Department of Psychiatry; 1996. [30] Goodman R. The Strengths and Difficulties Questionnaire: a research note. J Child Psychol Psychiatry 1997;38:581–6. [31] Goodman R. Psychometric properties of the strengths and difficulties questionnaire. J Am Acad Child Adolesc Psychiatry 2001;40:1337–45. [32] Seto MC, Kjellgren C, Priebe G, Mossige S, Svedin CG, Langstrom N. Sexual coercion experience and sexually coercive behavior: a population study of Swedish and Norwegian male youth. Child Maltreat 2010;15:219–28. [33] Kerig PK, Ward RM, Vanderzee KL, Arnzen Moeddel M. Posttraumatic stress as a mediator of the relationship between trauma and mental health problems among juvenile delinquents. J Youth Adolesc 2009;38:1214–25. [34] Hayes AF. Introduction to mediation, moderation, and conditional process analysis: a regression-based approach. New York: Guilford Press; 2013. [35] Muthen B, Muthen L. Mplus user’s guide, 7th ed., Los Angeles, CA: Muthe´n & Muthe´n; 2012. [36] World Health Organization. Preventing youth violence: an overview of the evidence; 2015. [37] Pechtel P, Lyons-Ruth K, Anderson CM, Teicher MH. Sensitive periods of amygdala development: the role of maltreatment in preadolescence. Neuroimage 2014;97:236–44. [38] Herts KL, McLaughlin KA, Hatzenbuehler ML. Emotion dysregulation as a mechanism linking stress exposure to adolescent aggressive behavior. J Abnorm Child Psychol 2012;40:1111–22. [39] Blair RJ, Leibenluft E, Pine DS. Conduct disorder and callous-unemotional traits in youth. N Engl J Med 2015;372:784. [40] Frick PJ, White SF. Research review: the importance of callous-unemotional traits for developmental models of aggressive and antisocial behavior. J Child Psychol Psychiatry 2008;49:359–75. [41] Bennett DC, Kerig PK. Investigating the construct of trauma-related acquired callousness among delinquent youth: differences in emotion processing. J Trauma Stress 2014;27:415–22. [42] Allwood MA, Bell DJ, Horan J. Posttrauma numbing of fear, detachment, and arousal predict delinquent behaviors in early adolescence. J Clin Child Adolesc Psychol 2011;40:659–67. [43] Kerig PK, Bennett DC, Thompson M, Becker SP. ‘‘Nothing really matters’’: emotional numbing as a link between trauma exposure and callousness in delinquent youth. J Trauma Stress 2012;25:272–9. [44] Dodge KA, Bates JE, Pettit GS. Mechanisms in the cycle of violence. Science 1990;250:1678–83. [45] Aebi M, Giger J, Plattner B, Metzke CW, Steinhausen HC. Problem coping skills, psychosocial adversities and mental health problems in children and adolescents as predictors of criminal outcomes in young adulthood. Eur Child Adolesc Psychiatry 2014;23:283–93. [46] Forsman M, Langstrom N. Child maltreatment and adult violent offending: population-based twin study addressing the ‘cycle of violence’ hypothesis. Psychol Med 2012;42:1977–83. [47] Odgers CL, Moffitt TE, Broadbent JM, Dickson N, Hancox RJ, Harrington H, et al. Female and male antisocial trajectories: from childhood origins to adult outcomes. Dev Psychopathol 2008;20:673–716. [48] Wood J, Foy DW, Goguen CA, Pynoos RS, Boyd James C. Violence exposure and PTSD among delinquent girls. J Aggress Maltreat Trauma 2011;6:109–26. [49] Ford JD, Elhai JD, Connor DF, Frueh BC. Poly-victimization and risk of posttraumatic, depressive, and substance use disorders and involvement in delinquency in a national sample of adolescents. J Adolesc Health 2010;46:545–52. [50] Qiu W. Power/sample size calculation for mediation analysis; 2015 [from https://cran.r-project.org/web/packages/powerMediation/powerMediation. pdf]. [51] R Development Core Team. R: a language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2011.

Marcel Aebi, Ph.D., born 1971, is a senior researcher at the Department of Forensic Psychiatry and the Department of Child and Adolescent Psychiatry and Psychotherapy, University Hospital of Psychiatry, Zurich, Switzerland. In addition, he is a behavioural therapist and forensic expert at the Institute of Clinical Child and Adolescent Psychology, University of Zurich. His current major research interests include forensic psychology, developmental psychopathology and various neuroscientific, genetic and clinical issues in child and adolescent psychopathology.

95

Meichun Mohler-Kuo, Sc.D., born 1968, is a trained psychiatric epidemiologist, title Professor at Epidemiology, Biostatistics and Prevention Institute, University of Zurich. She is also the head of competence center for mental health at the University of Zurich. Her current major research interests include mental health and service use, substance use, and child and adolescent health and development.

Steffen Barra, M. Sc., born 1989, is a research assistant at the Department of Forensic Psychiatry, University Hospital of Psychiatry, Zurich, Switzerland, and a Ph.D. candidate at the University of Zurich. His research interests include short- and long-term effects of childhood trauma and current life adversities on psychological and behavioral outcomes in children and adolescents. His current research focuses on forensic child and adolescent psychology/psychiatry, e.g. on risk factors for juvenile delinquency, such as adverse childhood experiences.

Ulrich Schnyder, M.D., born 1952, psychiatrist and licensed psychotherapist. Professor of psychiatry and psychotherapy. Head, Department of Psychiatry and Psychotherapy, University Hospital Zurich, Switzerland. Research activities are currently focused on various aspects of traumatic stress research, including epidemiology, neurobiology, psychotherapy and pharmacotherapy for PTSD, and resilience to stress. Past President, European Society for Traumatic Stress Studies. Past President, International Federation for Psychotherapy. Past President, International Society for Traumatic Stress Studies. Thomas Maier, M.D., born 1967, is board psychiatrist and currently head of the Psychiatric Services of St. Gallen North/Switzerland. He is a lecturer at Zurich University for psychiatry and psychotherapy. Formerly he headed the Outpatient Clinic for victims of war and torture at Zurich University Hospital. His research interests span from psychotraumatology to psychiatric epidemiology, transcultural psychiatry, and child sexual abuse.

Markus A. Landolt, Ph.D., born 1962, is Professor of Child and Adolescent Health Psychology at the Department of Psychology, University of Zurich and Head of Pediatric Psychology at the University Children’s Hospital Zurich, Switzerland. His research activities focus on different topics of posttraumatic stress in children and adolescents as well as their parents.

Downloaded from ClinicalKey.com at Drexel University February 18, 2017. For personal use only. No other uses without permission. Copyright ©2017. Elsevier Inc. All rights reserved.