Abstract
Exposure to adversity or maltreatment is known to correlate with high-risk behaviors that can increase the risk of contact with the criminal justice system; however, few studies have focused on the role of trauma symptoms and other behaviors that may develop in response to exposure to adverse life events. We also know far less about the role of these experiences in young men compared to young women who have been in contact with the youth justice system. In this study, we examine the associations between different patterns of trauma symptomatology and adverse childhood experienced (ACEs), substance use, behavioral difficulties, and re-offending in a sample of 141 young men under the supervision of a Youth Justice (YJ) agency in Australia. Trauma symptoms were reported by over ninety percent of participants, with Latent Class Analysis used to identify four subgroups of young men based on their clustering of trauma symptomatology: internalizing, externalizing, high, and low trauma symptoms groups. The characteristics of young men across these groups were, however, remarkably similar. Over four-fifths scored in the clinical range for substance use and externalizing behavior problems, while recidivism was reported in over three-quarters of young men across all groups. These findings have implications for the development of trauma-informed responses by specialist adolescent mental health services as well as for adult forensic services that seek to understand the developmental origins of psychopathology and offending behavior. There appears to be a strong rationale for more compassionate and trauma-informed justice system practices.
There is now consistent evidence that the experience of adversity and maltreatment in childhood is not only common (Racine et al., 2020), but also associated with a wide range of emotional and behavioral problems, including depression, post-traumatic stress disorder (PTSD), school and social difficulties, and substance dependency (e.g., Darnell et al., 2019; Felitti et al., 1998; Finkelhor et al., 2015). Even though only a small proportion of children who have these experiences will go on to engage in offending behavior (see Braga et al., 2017), it is also the case that many justice-involved young people will have had a number of life experiences that can be associated with ongoing trauma symptomatology (Malvaso et al., 2021) and elevate the likelihood of them engaging in behaviors that increase contact with the justice system (Baidawi & Sheehan, 2019). Of course, adverse childhood experiences (ACEs) are also particularly prevalent in the backgrounds of many forensic mental health service users, with Stinson et al. (2021) recently showing that they predict both aggression and criminality. Thus, from a treatment and rehabilitation perspective, it becomes important to understand the different developmental pathways through which childhood experiences result in trauma and increase the risk of ongoing offending. The aim of this study is to establish the extent to which different patterns of trauma symptomatology can be identified in a sample of justice-involved young people.
Childhood adversity and trauma
Research into adverse childhood experiences (ACEs) shows that various forms of abuse (whether physical, sexual, emotional abuse or neglect) and household adversity (e.g., living with a parent with mental illness, substance use problems or who has been incarcerated) tend to cluster (Dube et al., 2006), and can disrupt development in areas such as executive functioning, decision-making and self-regulation (Levenson & Willis, 2019). As a result, children who have been exposed to ACEs are more likely to report trauma symptomatology, including increased anxiety, depression, post-traumatic stress (PTS) and anger (Darnell et al., 2019; Kretschmar et al., 2018). Trauma reactions and symptoms can also manifest in risky behaviors, including problematic substance use and conduct problems. Research using data collected by the National Child Traumatic Stress Network, for example, found childhood adversity to be associated with both internalizing (mood/emotion based, i.e., anxiousness/depression) and externalizing (i.e., aggressive or delinquent) behavior problems (Greeson et al., 2014). While problematic alcohol and substance use and conduct problems are commonly observed among justice-involved adolescents, these behaviors have been conceptualized as coping mechanisms developed in response to previous, or ongoing, experiences of adversity (Kerig et al., 2009).
It is important from the outset to make the distinction between the terms ‘trauma’ and ‘adversity’, with the term trauma used in this study to refer to particular symptoms that develop in response to exposure to a negative life event (i.e., a reaction to the event), as opposed to simply being exposed to a potentially traumatic event (e.g., one of the aforementioned ACEs). The term 'trauma’ has also been defined in numerous ways (Briere, 1996; Levenson & Willis, 2019); for example, even though the American Psychiatric Association (APA, 2013) defines trauma as “an experienced or witnessed event that threatens an individual’s or other person’s physical or psychological wellbeing, and which engenders a sense of fear, vulnerability, helplessness, or shock”, it is now widely recognized that traumatic experiences can stem from more than a single event (i.e., natural disasters; motor vehicle/work accidents; or violence/conflict resulting from war or terrorism; Briere & Scott, 2006). Trauma can also result from prolonged exposure to child maltreatment, witnessing family violence, experiencing the death of a loved one, and exposure to parental difficulties such as substance use, mental illness, separation or incarceration. Herman (1992), for example, introduced the term complex PTSD (cPTSD), otherwise referred to as complex trauma to describe exposure to potentially traumatic stressors at an age (such as childhood), or otherwise in a context (i.e., from prolonged suffering), that affects their emotional self-regulation (Ford et al., 2012). cPTSD is multifaceted (i.e., results from numerous stressors), such as physical, sexual, or emotional abuse; neglect; ‘chronic childhood victimization’ (such as household and neighborhood violence); and bullying (Finkelhor et al., 2009). It also tends to be cumulative (i.e., involves repeated victimization), which is a factor known to increase severe emotional and behavioral difficulties (Ford et al., 2012). We also recognize that an individual’s immediate and long-term response to adversity and maltreatment will be influenced by a variety of biopsychosocial and cultural factors and that not all individuals who experience trauma symptoms will meet the criteria for a DSM-5 mental diagnosis (Bottalico & Bruni, 2012). Therefore, even if an individual does not fulfill diagnostic criteria for trauma-related disorders, it is critical to acknowledge that trauma symptoms can substantially affect an individual’s life (McLachlan, 2022; Substance Abuse & Mental Health Services Administration, 2014).
Associations between adversity and trauma in justice-involved young people
There is now substantial evidence that justice-involved young people are more likely to have experienced both exposure to adverse experiences and trauma symptomatology than young people in the general population (Malvaso et al., 2021). For example, several studies have now been published that show how ACEs and exposure to potentially traumatic events often coincide with difficulties with emotional regulation and processing (including numbing and dissociation; Modrowski & Kerig, 2017), depression, anger, somatic complaints (Kerig et al., 2009), and the development of ‘callous-unemotional’ (CU) traits (i.e., lack of empathy and responsiveness to punishment, impaired emotional processing; Mozley et al., 2018). These traits may emerge as a reaction to maltreatment, by emotionally detaching oneself to help “cope with the distress associated with the trauma exposure” (Mozley et al., 2018 p. 745) and may exacerbate if left untreated, thereby increasing the risk of further poor outcomes, including escalating psychological complications, substance abuse, disengagement with schools and help services, and recidivism (Kerig et al., 2009).
Patterns of adversity exposure and trauma symptomatology among justice-involved young people are likely to differ according to sex. Child maltreatment has been identified as a significant predictor of juvenile offending in males (Asscher et al., 2015). Males are known to be more likely to develop externalizing disorders (e.g., substance use; behavioral difficulties, anger; Chong et al., 2020) because of trauma, which can lead to delinquency and violent offending; while females are more likely to develop internalizing disorders (e.g., depression, anxiety; SAMHSA, 2014). Drapalski et al. (2009), for example, found that females prisoners experienced higher rates of posttraumatic stress, with the prevalence of alcohol related problems twice as high than for male prisoners. However, Malvaso et al. (2022) found young men to have slightly higher internalized symptoms of trauma than females, specifically in relation to dissociation. Thus, even though much of the previous research has focused on understanding trauma symptom experiences in young women, it is important to also explore trauma symptom experiences in young men. This is especially important in light of research that has found stronger associations between exposure to childhood maltreatment and subsequent violent offending among young men compared to young women (DeGue & Widom, 2009; Rivera & Widom, 1990), as well as the practice implications of the potential need to develop gender-specific trauma interventions (e.g., see Covington & Rodriguez, 2016).
Current study
This study examined associations between trauma symptoms, ACEs, substance use, and offending behavior in a representative sample of young men who were under YJ supervision in an Australian jurisdiction. There were three primary research questions: Are there subgroups of young men under YJ supervision who report experiencing different patterns of trauma symptoms? Do the groups differ in terms of experiences of adversity, substance use, and social and emotional behavioral problems? Are there any differences between the groups in terms of recidivism and subsequent contact with YJ within 12-months of initial assessment?
Method
This study involved a subgroup re-analysis of data from a larger research project. The original study included data collected from 184 young people aged 14 and 21 (males: n
Participants
A total of 173 young men under YJ supervision were approached to participate in this study, with the majority (n = 155; 90%) consented to participate. Of the 155 young men who participated in this study, 53 (34.2%) identified as Aboriginal and/or Torres Strait Islander. The minimum age of male participants was 14 and the maximum 19 years, with a median age of 16 years. Based on age and cultural background, the subsample of young men utilized in this study remained broadly representative of the male YJ population. Just over half of the young people participated in the study while under supervision in the community (n = 89; 57%) and the other half participated while they were in custody (n = 66; 43%).
Measures and materials
Demographic information, including age (in years), Aboriginal identification (yes/no) was collected from all participations, and the following self-report assessments were completed: the Trauma Symptom Checklist for Children (TSCC; Briere, 1996); an adapted version of the Adverse Childhood Experiences (ACEs) Questionnaire (Felitti et al., 1998); the Adolescent Alcohol and Drug Involvement Scale (AADIS; Moberg, 2005); and Child Behavior Checklist-R Youth Self-Report (CBCL-YSR; Achenbach, 2001).
Trauma symptoms
The TSCC is a 54-item tool that measures trauma symptoms across six clinical scales: Anxiety, Depression, Anger, PTS, Dissociation, and Sexual Concerns. The dissociation scale includes two subscales: overt and fantasy dissociation. Similarly, the sexual concerns scale includes two subscales: sexual preoccupation and sexual distress. Participants were asked how often they had experienced the symptom described in each item in the past month, with options on a four-point scale ranging from zero (not at all) to three (very often). For the six clinical scales, T-scores of over 65 are indicative of clinically significant symptomatology, and T-scores between 60 and 65 indicate subclinical (but significant) symptomatology. Dichotomized variables (yes/no) for each clinical scale were created to indicate which young men had significant trauma symptomatology (i.e., young men with T-scores above 60 were coded as experiencing significant trauma symptomatology according to each of the six clinical scales). The TSCC includes two validity checks to identify hyper-responding (i.e., over-reporting experiences) and under-responding (i.e., under-reporting experiences; Briere, 1996). Of the 155 young men who participated, seven (4.5%) were identified as hyper-responders, but were included in the study sample following professional consultation which determined their scores were likely to reflect true symptoms experienced by young men in the justice system. However, 14 young men had invalid TSCC scores due to under-responding and so were not included in the trauma symptom results; this left a total of 141 young men whose data were analyzed in this study. Eight critical items are also assessed through affirmative responses to the following items: (1) wanting to hurt myself, (2) wanting to hurt other people, (3) feeling scared of men, (4) feeling scared of women, (5) not trusting people because they might want to have sex, (6) getting into fights, (7) feeling afraid someone will kill me, and (8) wanting to kill myself. These items are used to screen for problems or issues that may require more immediate clinical attention. The TSCC has good psychometric properties (test-retest reliability of r=.81), and good construct validity (Lanktree et al., 2008).
Adverse childhood experiences (ACEs)
The ACEs questionnaire was adapted from that developed by Felitti et al. (1998) and included the original 28 items relating to 10 adverse experiences: (1) emotional abuse, (2) physical abuse, (3) sexual abuse, (4) emotional neglect, (5) physical neglect, (6) family violence, (7) household substance abuse, (8) household mental illness, (9) parental separation or divorce, and (10) incarcerated household member. The Felitti questionnaire was constructed using questions from the original Conflicts Tactics Scale (Straus & Gelles, 1990) to define psychological and physical abuse during childhood and violence against the respondent’s mother. However, we adapted the question of witnessing violence against the mother to encompass all forms of family violence, in recognition that violence toward women is just one form of family violence. This question thus included physical and psychological abuse between parents and between siblings, and adolescent violence toward parents. Questions from Wyatt (1985) were used to define sexual abuse, and questions about exposure to alcohol or drug abuse were adapted from the National Health Interview Survey (Blackwell & Tonthat, 2002). We also included three additional adverse childhood experiences: death of a close relative or friend, peer violence and bullying, and witnessing neighborhood violence. Therefore, 13 individual ACEs were assessed in this study.
ACEs exposure was measured in several ways. First, affirmative responses to any individual item under each ACEs category were used to define those who had experienced a particular ACE. For the maltreatment ACEs, we also used administrative child protection (CP) substantiations to create combined self-reported and officially recorded exposure to physical, sexual, emotional abuse and neglect. It was not possible to differentiate physical neglect from emotional neglect using the administrative data; therefore, a combined category of ‘any neglect’ was created and included information from both the self-report questions and administrative records. A total ACEs score, the sum of all affirmative responses in each of the 13 categories, was then calculated. We used the Centers for Disease Control and Prevention’s (2015) cutoff of four or more ACEs to better understand how exposure to ACEs in the current sample compared with other international studies. We only used the original ACEs items for the purposes of these comparisons.
We also measured the frequency of these experiences by using the response categories defined by Straus and Gelles (1990), including ‘never’, ‘once or twice’, ‘sometimes’, ‘often’ or ‘very often’. We defined frequency using endorsements of the ‘often’ and ‘very often’ categories for particular ACEs (physical abuse, sexual abuse, emotional abuse, neglect, family violence, bullying and neighborhood violence), supplemented by two or more substantiations for the four maltreatment types.
Substance use
Use of alcohol and other drugs was measured using a screening tool for adolescents adapted from the well-established Adolescent Drug Involvement Scale (Moberg, 1991). This is a face-valid and reliable measure (α
Social and emotional behavioral problems
Emotional and behavioral problems and social competencies were assessed using a standardized screening questionnaire, the Child Behavior Checklist-R Youth Self-Report Form (CBCL-YSR; Achenbach, 2001). The CBCL-YSR is normed for use with young people aged 11 to 18 years. Only the syndrome profiles of the CBCL-YSR were used in this study, which comprises 118 items scored from zero to two. There are eight syndrome scales: anxious/depressed, withdrawn/depressed, somatic complaints, social problems, thought problems, attention problems, rule-breaking behavior, and aggressive behavior. High scores on these scales indicate problems in these areas. Problems are broken down into clinical (scores above the 97th percentile of the normative sample) and borderline (scores between the 93rd and 97th percentiles) concerns. The syndrome scales can be scored in terms of two broad groupings of syndromes: internalizing (comprising problems that are mainly within the self) and externalizing (comprising problems that mainly involve conflict with other people). A total problems score can also be computed by summing the scores of the eight syndrome scales, with clinical ranges at or above the 84th percentile. The CBCL-YSR has been reported to have good test–retest reliability (with the mean score ranging from 0.79 to 0.88), and our study demonstrated strong internal consistency (α
Administrative records
Youth justice details
Supervision order details (community-and custodial-based), age and first supervision type, supervision length and number of orders, were all extracted from the Connected Client Case Management System. C3MS is an Australian state-based CP database program used to record and store all client details, including correspondence, actions, and decisions. This system replaced the previously used Client Information System in 2009–2010. All offense details (agreed to, proven, and/or convicted) were extracted from the Justice Information System. ‘Proven’ or ‘agreed offenses’ refers to those that have been legally sustained, but when the young person did not receive a formal conviction. This occurs in accordance with South Australia’s emphasis on diversion as stipulated by the Young Offenders Act 1993 (SA) and the Youth Justice Administration Act 2016 (SA). Matters that had been dismissed, not proceeded with or committed to trial in which the defendant was subsequently found not guilty were not counted. Offense type was recorded, as well as age at time of the individual’s first agreed to, proven, or convicted offense, all of which were classified according to the third edition of the Australian and New Zealand Offense Classification. Follow-up at the 12-month period involved a second record extraction of YJ system information to assess recidivism and any details of new proven or agreed to offenses, convictions and YJ supervision orders in the 12 months post-interview. In this study we use two dichotomized measures of recidivism: any new conviction in the 12 months post interview (yes/no) and any new YJ supervision order in the 12 months post interview (yes/no).
Child protection
In South Australia (SA), child protection (CP) contact ranges from notifications (or reports) of suspected maltreatment, screened-in notifications (notifications assessed to meet a threshold of concern), investigations (notifications resulted in a CP investigation) and substantiations (verification maltreatment occurred or was at risk of occurring). The CP agency in this jurisdiction records a primary substantiated maltreatment type, which reflects the type of maltreatment considered to have, or most likely to have harmed or placed the child at risk of harm and includes physical, sexual and emotional abuse, and neglect. In situations where children are assessed as unable to remain safely in the care of their families, the CP agency applies to the Youth Court for orders that enable the removal of children to out-of-home (OOHC). There are different types of orders that can be made, but the two most common types include shorter-term guardianship orders to place children in OOHC arrangements for 12 months or longer-term guardianship orders until the child is aged 18. Types of OOHC placements include: foster care where children are placed with foster parents, kinship care where children are placed with the child’s extended family or kin network and residential care where children are in houses staffed by carers on a rotational basis employed through the CP agency or contracted agencies. In this study, all notification details (i.e., alleged maltreatment, risk-of-harm, and adolescent-at-risk reports) made to the Department for Child Protection and ensuing outcomes (i.e., investigations, substantiations, guardianship orders, and out-of-home care placement details) were extracted from the Client Information System and the Connected Client Case Management System. This information also included the age of each contact and/or placement, and number of contacts and/or placements. From these data, markers of ‘ever being on a long-term guardianship order until age 18′, ‘ever placed in foster care’, and ‘ever placed in residential care’ were created. Administrative CP substantiations were used to create a combined self-reported and officially recorded exposure to the four ACEs relating to maltreatment (i.e., physical, sexual, emotional abuse and neglect). That is, if a young person self-reported or had a substantiation for one of the four types of maltreatment an affirmative response was coded. As four young men did not consent to their CP records being accessed, they are not included in the total (N) for CP history.
Procedures
Data collection took place over a 12-month period, from March 2019 to February 2020. Recruitment involved the lead researcher [author blinded] liaising with community YJ case managers and custodial center staff to identify young people who were eligible to participate and to facilitate interviews. The lead researcher then met with eligible young people and invited them to participate in the study. Participation was completely voluntary and no incentives to take part were offered. Participants were given the option to complete the assessment themselves or have the researcher read out the questions and fill in the responses. If the young person chose to complete the assessment themselves, the researcher first assessed potential literacy issues by asking the young person to read the first set of instructions out loud. If any literacy issues were identified, the researcher then assisted the young person to read the questions and explained their meaning. Most young people (>85%) asked the researcher to verbally administer the assessment but provided verbal responses. The interviews varied in duration from approximately 40 to 60 min. As part of our risk management strategy, if a young person endorsed any of the eight critical items during the interview, the Youth Justice Principal Psychologist was alerted to undertake a welfare check.
Statistical analysis
The analysis proceeded in three stages that corresponded to each of the research questions. First, Latent Class Analysis (LCA) was used to determine whether there were groups of young men with distinguishable clusters of trauma symptoms. Dichotomized indicators of significant trauma symptoms based on the six subscales of the TSCC (anxiety, depression, PTS, anger, dissociation and sexual concerns) were used to identify different patterns of trauma symptoms. Two-, three-, four- and five-class LCA models were tested, where the posterior probability of membership in each class computed for each individual. Cases were then assigned to the class for which they had the highest probability. Each solution was then assessed according to fit and diagnostic statistics as well as face validity checks. While there is no consensus about the best statistical criteria for comparing classes, we considered and reported multiple fit and diagnostic statistics as suggested by Weller et al. (2020). This included likelihood ratio squared (G2), the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and Entropy (a measure of separation between latent classes). While lower AIC and BIC scores indicate a better fit, higher entropy denotes better class separation. Given our small sample size, we also considered the adjusted BIC (aBIC) and consistent AIC (CAIC). Statistical criteria were evaluated in conjunction with face validity checks regarding the most suitable distinction between classes. This is in accordance with recommendations made by Weller et al. (2020) who emphasize the importance of theoretical interpretability and validity alongside the evaluation statistical criteria.
Second, we examined whether the experiences of adversity, substance use, and social and emotional behavioral problems differed between the groups identified in stage one. Last, we identified whether there were any differences between the groups regarding new offenses and subsequent contact with YJ within 12-months from the initial assessment. We report Pearson’s Chi-square tests of independence for comparison between categorical variables and class, and Cramer’s V as a measure of the magnitude of the association. Whilst p values are provided in the results, we advise that these not be used as a sole basis for interpretation, as recommended by the American Statistical Association (Wasserstein & Lazar, 2016). All analyses were conducted in Stata version 17, with LCA performed using the Stata Plugin (version 1.2; Lanza et al., 2018; StataCorp, 2021).
Results
Descriptive characteristics
Of the 155 young men, over four-fifths of young men (n = 133; 85.8%) had a previous proven offense or conviction prior to the interview, with over half (n = 87; 56.1%) having committed a violent offense. The age of a first proven offense consisted of 60 (45.1%) young men aged 10–13 years, and 73 (54.9%) aged 14–17 years. More than three-quarters of the young men had spent at least one night in custody (n = 123; 79.4%). Almost all were the subject of CP contact (n = 142; 94%), with 86 (57%) the subject of an investigation for alleged maltreatment, 69 (45.7%) the subject of a substantiation for maltreatment, and 41 (27.2%) having experienced at least one placement in out-of-home care (OOHC). Descriptive statistics for all interview measures utilized in this study are included in Table 1.
Sample characteristics.
ACEs – adverse childhood experiences.
A total of 127 out of the 141 (90.1%) young men with valid scores on the TSCC had scores that were indicative of clinical level difficulties on at least one trauma symptoms scale. This included 56.7% who reported scores indicative of clinical levels of anxiety, 58.8% for depression, 63.8% for anger, 68.1% for posttraumatic stress, and 73.1% for dissociation. One in ten young men reported scores indicating sexual concerns (12.1%). When considering the critical items in the scale, more than a quarter of young men (27.0%) reported wanting to kill themselves, or feeling afraid that someone will kill them (25.5%).
Nearly all (99.4%) young men reported that they had experienced at least one ACE, with more than four-fifths (83.9%) experiencing four or more ACEs. The most common ACE reported was emotional abuse (79.4%), followed by experiencing the death of a close friend or family member (78.1%), neighborhood violence (71.6%) and physical abuse (68.4%). A total of 143 (92.3%) young men reported having experienced any one ACE frequently (i.e., ‘often’ or ‘very often’), specifically: physical abuse (45.1%), sexual abuse (5.2%), emotional abuse (61.3%), neglect (60.0%), family violence (44.5%), bullying (43.2%), and neighborhood violence exposure (39.3%). A total of 133 (85.8%) young men had scores indicative of problematic drug and/or alcohol use, with over half (52.9%) reporting smoking marijuana daily. Finally, nearly two-thirds of young men (n = 101; 65.2%) provided responses that reflected the presence of internalizing behavioral problems, and over four-fifths experienced externalizing behavioral problems (133; 85.8%).
Latent class analysis
To answer the first question, if there are subgroups of young men under YJ supervision who report experiencing different patterns of trauma symptoms, a LCA was conducted to identify potential underlying subgroups of young men who shared similar trauma symptom characteristics. The six subscales of the TSCC were used as indicators of different patterns of trauma symptoms (anxiety, depression, PTS, anger, dissociation and sexual concerns). Two-, three-, four- and five-class LCA models were tested, where the posterior probability of membership in each class computed for each individual. Cases were then assigned to the class for which they had the highest probability. The four-class model was chosen as best distinguishing between groups of young men with varying trauma symptoms based on its fit statistics and face validity checks (Table 2). Although the four-class model had a slightly lower entropy score, it also had the lowest aBIC and AIC, and the subgroups across the six trauma symptoms were more clearly interpretable, whereas the three-class model did not sufficiently separate the trauma symptom groups, despite having the highest entropy. The five-class model had insufficient (< 5) young men in one group, thus it was not considered for selection as advised in Shanahan et al. (2013).
Fit statistics for two-, three- and four class trauma symptoms latent class analysis.
Four subgroups of young men under YJ supervision were identified from the LCA, based on different patterns of trauma symptoms (Figure 1; see also Table S1 for a summary of class membership and item response probabilities):
Four-class model of trauma symptoms (N = 141).
an
an
a
a
Comparative analysis of inter-group differences
A comparative analysis of the characteristics of the groups identified through the LCA was undertaken to identify where groups differed in terms of experiences of adversity, substance use, and social and emotional behavioral problems. Overall the ‘high trauma symptoms’ group had the highest proportion of young men reporting any ACE (see Table 3). Experiences of physical and emotional abuse were highest amongst this group (80.7% and 91.2%, respectively); however, emotional abuse was also prevalent in just over three-quarters of the internalizing and externalizing trauma symptom groups. Sexual abuse was relatively low across all groups, with just under one-quarter of the ‘high on all trauma symptoms’ group reporting this. Experiences of any neglect were relatively high across all groups (>74%); however, physical, and emotional neglect were highest among the ‘high trauma symptoms’ group. Under half of the ‘internalizing and externalizing trauma symptom groups’ experienced family violence, with over four-fifths reporting it in the ‘high trauma symptoms’ group.
Comparison of subgroups of young men according to patterns of trauma symptoms, by age, child protection history, adverse childhood experiences, substance use, and social and emotional behavioral problems (%).
Note. N = 141; CP = child protection; OOHC = out of home care; ACE = adverse childhood experiences.
There were some differences in household dysfunction across groups. Parental separation was relatively lower in the ‘internalizing trauma symptoms’ group (56.5%), with over three-quarters of young men in the ‘externalizing’ and ‘high trauma symptom’ groups having experienced parental separation. While just over half of the ‘internalizing’, ‘externalizing’, and ‘low trauma symptom’ groups had a household member who abused substances, two-thirds in the ‘high trauma symptoms’ group reported this ACE. Similarly, over half of the ‘high trauma symptoms’ group reported living with a household member with mental illness, whereas it was relatively lower across the three other groups (<38%). The proportion who reported having a household member who had been incarcerated was relatively similar across the ‘externalizing’ (62.5%) and ‘high trauma symptom’ (66.1%) groups. Over three-quarters of the ‘low trauma symptoms’ group reported experiencing the death of a close friend or family member, but nearly all (94.6%) of the ‘high trauma symptoms’ group reported this experience. The highest proportion (82.6%) of young men to experience bullying was reported in the ‘internalizing trauma symptoms’ group. Prevalence of neighborhood violence was relatively similar across all groups, with over three-quarters reporting this experience in the ‘externalizing’ and ‘high trauma symptom’ groups, and over two-thirds in the ‘internalizing’ and ‘low trauma symptom’ groups. Cramer’s V indicated that the association between class and each ACE was small to moderate, ranging from 0.16 for neighborhood violence to 0.4 for family violence.
Overall, the proportion of young men who had scores indicating problematic substance use was high across all groups (>82.2%), with all young men in the ‘externalizing trauma symptoms’ group presenting with such issues. Daily marijuana use was highest among the ‘externalizing’ and ‘high trauma symptom’ groups, with over two-thirds of young men in the ‘internalizing’, ‘high’, and ‘low’ trauma symptom groups reporting weekly marijuana use. Weekly alcohol use was most common among the ‘externalizing’ and ‘high trauma symptom’ groups, with the ‘internalizing’ and ‘low trauma symptom’ groups demonstrating relatively lower levels of regular alcohol consumption. Similarly, Cramer’s V indicated that the association between class and substance use was small to moderate, ranging from 0.12 for weekly marijuana use and 0.33 for daily tobacco use.
Similar to the patterns observed for ACEs, the ‘high trauma symptoms’ group had the highest proportion of young people with each of the social and emotional behavioral problems measured (with the exception of externalizing symptoms which was second highest). As expected, the ‘internalizing trauma symptoms’ group had a high proportion of young men (95.7%) with internalizing social and emotional behavioral problems, but the prevalence of these symptoms was equally as prevalent in the ‘high trauma symptoms’ group (96.5%). Similarly, the ‘externalizing trauma symptoms’ group had the highest proportion of young men reporting externalizing and aggressive behavior (100% of young men), but the prevalence of these symptoms were also close to 100% in the ‘high trauma symptoms’ group. The largest magnitude of effect was observed here between class and behavioral problems, with moderate effects (above 0.6) for internalizing and aggressive behaviors.
Finally, an analysis of subsequent contact with the YJ system 12 months post initial interview showed that over three-quarters of young men across all groups had received a new conviction within the 12-months from their initial interview (Table 4). All of the young men in the ‘externalizing trauma symptoms’ group had a new conviction within 12-months. Young men in the ‘internalizing trauma symptoms’ group were the least likely to return to YJ on a new supervision order in the 12-months post-interview (40.9%), with almost two thirds to three quarters of young men in the other three groups returning to YJ supervision. The effect of class on recidivism was small in magnitude (<0.2 for both measures).
Comparison of subgroups of young men according to patterns of trauma symptoms, by indicators of recidivism 12-months post interview (%).
Note. N = 138 for recidivism 12-months post interview as three young men did not consent access to their administrative records, thus they could not be included in the analysis for any new conviction(s) or YJ order.
Discussion
This study examined associations between trauma symptoms, ACEs, substance use, and behavior in a representative sample of young men under YJ supervision in an Australian jurisdiction. It shows distinct subgroups of young men under Youth Justice (YJ) supervision can be identified, based on different clustering of trauma symptoms. Through a Latent Class Analysis (LCA), we identified two groups who experienced trauma symptoms predominantly relating to either internalizing or externalizing problems, respectively. A third group was distinguished by their experience of all trauma symptoms (except for sexual concerns, which had a lower prevalence across all groups). The final group included young men with a relatively lower prevalence of all trauma symptoms, though it is noted that almost one third of this group reported posttraumatic stress (PTS) symptoms.
While distinct subgroups of young men based on patterns of trauma symptomatology were identified, a comparison of characteristics between the four groups revealed more similarities than differences. The majority of young people across all groups reported adverse childhood experiences (ACEs), substance use problems, and social and emotional behavioral problems. Similarly, indicators of recidivism were high across all groups, with the majority in each group having a conviction or proven offense during the 12-month follow-up period. These results are perhaps somewhat unsurprising given that the prevalence of ACEs, substance use problems, social and emotional behavioral problems and indicators of recidivism were all exceptionally high in this study. As has been noted by Lee and Taxman (2020), LCA results continue to strengthen the argument that approaches to the assessment and treatment of justice-involved adolescents need to be more holistically focused. While we cannot draw conclusions about the causal associations using these cross-sectional data, these findings clearly demonstrate that the YJ population of young men in this Australian jurisdiction is characterized by a high prevalence of trauma and adversity that have likely played a role in their adolescent development, which includes problems relating to substance use and social and emotional behavior.
We also observed a high proportion of young people receiving a new conviction or supervision order in the 12 months post interview. This finding needs to be interpreted within the context of the jurisdiction in which this study took place. South Australia has a very strong focus on diverting children and young people from the justice system (e.g., through police cautions and family conferencing), with only a small proportion of young people charged with offenses subsequently placed on a supervision order. We also know that the YJ population in South Australia (SA) has changed over time. For example, we have shown that while the absolute number of young people under YJ supervision has decreased, the population is defined by repeat involvement and increased complexity, with a growing proportion of young people who experience child protection contact and acute mental health issues (Malvaso et al., 2023a,2023b). We also considered breaches in our calculation as these are legislated as new offenses in the jurisdiction of SA, and also contributes to the high proportion of young people who were identified as recidivating in 12 months. It is also possible that the high proportion of young people who were identified as recidivating is an unintended consequence of often relatively short mandates, which may limit the opportunity for more intensive, therapeutic work to take place especially within the context of limited child and adolescent mental health services for justice-involved young people.
Treating trauma and responding to adversity among justice-involved young people
There is now a well-established body of international literature documenting the high prevalence of exposure to ACEs and symptoms of trauma among justice-involved young people. However, the majority of this involves incarcerated populations in the United States and the prevalence of ACEs and trauma symptoms among young men in this study was two-to-three times higher than the pooled prevalence estimates reported in Malvaso et al. (2021) systematic review. While these differences may simply highlight jurisdictional differences (Australian YJ systems are known for their strong focus on diversion that results in a population under supervision with more serious, or repeat, offenses), the importance of collecting local data to guide policy and practice decisions is clear. For example, from this study, we know that the YJ population in this Australian jurisdiction is characterized by a high prevalence of trauma symptoms (even 30% of the ‘low trauma symptoms’ group had scores indicative of clinical or subclinical levels of PTS symptoms). And, as we know that the number of young men under YJ supervision in this Australian jurisdiction is small, it may be both feasible and beneficial to offer intensive, therapeutic services to address trauma-related needs. This profile may well be different in countries that have a different YJ system or legislative mandate for working with justice-involved young people.
A number of meta-analyses and reviews have now documented the efficacy of trauma-focused interventions in reducing trauma symptoms and improving mental health in justice-involved populations (e.g., Baetz et al., 2022; Gagnon et al., 2022; Hodgkinson et al., 2021; Kumm et al., 2019; Olaghere et al., 2021; Rhoden et al., 2019). However, while intensive mental health support may reduce trauma symptoms and increase wellbeing among this population of young men, it is less clear whether these interventions will influence the likelihood of re-offending or returning to YJ supervision. Because indicators of recidivism in this study were high across all groups, even among those with a relatively lower prevalence of trauma symptoms, it is difficult to determine the extent to which these experiences are directly linked to offending, or re-offending, behavior. It may be that other factors, such as the high prevalence of substance use problems across all groups, is more strongly associated with re-offending. But, as other researchers have suggested, adverse experiences and trauma may still serve to increase vulnerabilities such as substance use and associating one-self with negative peers, which in turn increases the likelihood of re-offending (Arredondo, 2003; Kerig et al., 2009; Ward, 2020).
Another issue that arises in relation to the provision of intensive therapeutic mental health services in YJ relates to the context and logistical arrangements in which interventions are provided. Identifying which young people might benefit most from these services requires screening and assessment. The challenge here is that the nature and impact of exposure to traumatic events will be unique to each young person, so individualized case formulation is required to identify those for whom these experiences are consequential for engagement in offending. Willmot (2022) suggests some key areas of focus, including: neuropsychological impairment, disrupted attachment, cognitive and attitudinal impact, social learning of abusive behavior, and PTS symptoms. However, it is important to acknowledge that screening and assessment are only beneficial if they can be linked to appropriate treatments and service provision (Batty & Kivimaki, 2021; Finkelhor, 2018) as YJ agencies are limited in their ability to provide intensive specialist mental health treatment (e.g., over an appropriate duration due to short mandates which makes establishing rapport required for engaging in intensive therapeutic treatment regimens challenging). Thus, providing the necessary support will likely extend beyond the remit of YJ agencies and requires close coordination and collaboration with forensic child and adolescent mental health services. In addition, there is a need to focus on the transition period from adolescence to adulthood, when both offending behavior and mental health problems peak but rates of help-seeking, engagement and retention in treatment programs are typically lower (Aalsma & Dir, 2021). And for those who provide forensic mental health services for adults, understanding the associations between exposure to ACEs, the onset and maintenance of trauma symptomatology, substance use, and offending behavior will be key to the provision of effective treatment. This study describes some of these associations for those who are involved in the justice system from an early age.
There is also a need, perhaps, to develop more inclusive eligibility criteria for trauma intervention services for forensic populations. We would argue, as others have (e.g., Ford et al., 2012) that subclinical trauma presentations are common in forensic populations and that a large proportion of young people will not fulfill stringent criteria for clinical disorders. And yet, these are both clinically and forensically important and warrant the attention of mental health service providers. It may also be important to recognize that traditional therapeutic interventions for trauma that rely heavily on cognitive modalities may not be sufficiently responsive to the needs of all justice-involved young people, especially those with cognitive disabilities and neurodevelopmental disorders - which are very common in this population (Baidawi & Piquero, 2021; McVilly et al., 2022). Additionally, intergenerational repercussions of historical trauma—including systemic oppression, poverty, and discrimination (i.e., amongst Indigenous peoples, ethnic minorities, and other marginalized groups)—are also relevant but have generally been disregarded in definitions of trauma (Day & Malvaso, 2021; Levenson & Willis, 2019) and warrant further consideration.
There may also be opportunities for researchers and practitioners to further consider the benefit of options that extend currently available clinical models of care (McLachlan, 2022), such as by embedding role of trauma-informed care and trauma-informed practice (SAMHSA, 2014) across all levels of a service. While a growing number of researchers and policymakers are endorsing trauma-informed practice approaches in justice and forensic settings (Jackson et al., 2023; Ward, 2020), there is an ongoing need to better understand the activities and outcomes that constitute trauma-informed practice. Embedding trauma-focused forensic treatment programs within service- and system-wide, trauma-informed models of care will require adequate resourcing and organizational support. We will also need evaluations that consider the impact of activities across system-level interventions, and these should consider the views and experiences of both service provides and users (Day et al., 2023).
Further research is also needed to understand how trauma-informed practice may fit within current criminal justice case management models and rehabilitation services. While there may be no inherent tension between working in ways that are trauma-informed within current service models (e.g., see Levenson & Willis, 2019), it is likely that different assumptions underpin trauma-informed youth justice approaches compared with the more dominant risk-needs-responsivity models. For example, Griffin et al. (2012) have argued that trauma-informed approaches conceptualize ‘risk’ in terms of the developmental vulnerabilities that arise in response to childhood maltreatment and social and structural drivers of inequalities. A trauma-informed justice service might therefore conceptualize criminogenic need as arising from developmental vulnerabilities and, as a result, implement interventions aimed at targeting these underlying vulnerabilities. For example, Kerig and colleagues have demonstrated how problematic substance use may develop as a response to previous, or ongoing, experiences of adversity (Kerig et al., 2009), as well as how emotional numbing and symptoms of detachment and dissociation underpin the presentation of callous-unemotional traits in justice-involved young people (Mozely et al., 2018). Trauma-informed youth justice might also place particular emphasis on outcomes that are in addition to reducing recidivism, such as those that relate to the physical, psychological and cultural safety and wellbeing of both young people and staff (e.g., see Day et al., 2023).
Limitations
Interpretation of this study’s findings should be considered in the context of some limitations of, and decisions made regarding the data and analyses. For example, we considered exposure to the four types of maltreatment to be affirmative based on both self-report and official records, whereas the measures of family dysfunction were based solely on self-report. Future research could examine whether there are any differences in associations between ACEs and the measures of interest in this study based on the concordance or discordance between self-reported and officially recorded maltreatment. A further consideration here is that different jurisdictions will have varying thresholds for notifying, screening in, investigating and substantiating child maltreatment allegations, especially given that child protection agencies are often resource constrained. This may limit generalizability of our results to some extent. Another consideration is that while one of the goals of the broader study was to consider the difference between exposure to ACEs and symptoms of trauma, it did not necessarily follow that all young people exposed would display trauma symptoms. One omission in this study was a measure of resilience, which may have provided some insight not only into why a small proportion of young people did not report symptoms of trauma but also into their strengths.
Finally, our choice of statistical methodology, LCA, and the criteria used to decide number of classes is continuously evolving and subject to debate (Weller et al., 2020). One potential concern in our paper is whether our sample size was adequate for performing this particular statistical technique. However, while Nylund-Gibson & Choi, 2018) recommended that the desirable sample size for conducting LCA is greater than 300 cases, they also note that simpler models can be tested on smaller samples with fewer indicators. While there is also no consensus on the number of indicators that can be included, Nylund-Gibson & Choi’s review indicated that studies may have as few as four indicators and still perform well according to fit indices. We limited our analyses to the six clinical scales of the TSCC, and did not run into any issues in terms of model convergence failures or failure to uncover classes with low membership. Entropy scores for all solutions were also acceptable at above 0.8. We followed Weller et al. (2020) recommendations to evaluate statistical criteria in conjunction with interpretability to arrive at our decision that the 4-class solution was the most appropriate; however, we recognize the inherent subjectivity in these decisions. While LCA is becoming a more popular technique utilized in criminology and related disciplines (e.g., see Barra et al., 2018; Charak et al., 2019; Ford et al., 2013; Wolff et al., 2018), there is not always a lot of consistency in patterns identified. As Lee and Taxman (2020) have argued, LCA results may simply add further credence to the need for more holistic approaches to the assessment and treatment of young people under YJ supervision, with trauma symptoms being potentially key targets in intervention efforts.
Conclusion
This study highlights the importance of research that identifies the characteristics of young people in YJ, such that policies and practice can be accordingly adapted and tailored to the needs of justice-involved young people. We identified four distinctive groups of young men who experience different types of trauma symptomatology; however, these groups were also remarkably similar in terms of previous exposure to ACEs, the prevalence of substance use, social and emotional behavioral problems, and importantly, in terms of indicators of recidivism. These data, in our view, strengthen arguments for the provision of more holistic approaches to forensic mental health service delivery and highlight the need to provide clinically-focused trauma interventions and system-wide trauma-informed practices when responding to the needs of justice-involved young people. More broadly, the findings from this study raise important social and political questions about whether we are criminalizing childhood adversity and trauma especially in youth justice systems. They provide a strong rationale not only for more compassionate and trauma-informed justice system practices, but also the need for integrated multi-system supports that are better able to meet the needs of this population of young people and prevent downstream economic and personal costs, such as progression through to the adult criminal justice system.
