Implementation of Trauma-specific Interventions in a Child and Adolescent Mental Health Service

Exposure to childhood adversity is a prevalent and pressing issue due to its established links with poor physical and mental health outcomes across the lifespan. High rates of adversity have been found within young people attending mental health services, and the provision of trauma-informed care and trauma-specific therapies are therefore essential within these services. Trauma-focused cognitive behaviour therapy (TF-CBT) is an evidence-based intervention with established efficacy amongst traumatised children and youth. The primary objective of this study was to evaluate the effectiveness of trauma-specific therapy (TF-CBT) when implemented in a Child and Adolescent Mental Health Service (CAMHS). Given that most youth receive care in community settings, it is important to determine whether treatments tested under controlled conditions can apply in the real world of clinical practice. Specifically, this study aimed to quantify the number and type of adversities reported by clients receiving the intervention; investigate whether the intervention is associated with improvements in participants’ posttraumatic stress disorder (PTSD) symptoms, general mental health symptoms and functioning; and report contact with CAMHS at 3 months and 2 years post-treatment. Results showed that CAMHS clients were more likely to have experienced multiple traumatic events than single events, with 100% of clients experiencing suicidal thoughts. Treatment was associated with significant improvements in PTSD symptom severity and improvements in general mental health symptoms. Many clients received ongoing mental health support at the 3-month and 2-year follow-ups, reflecting the complex nature of their issues and the need for support to maintain gains achieved during treatment. Result from the dissemination and implementation of TF-CBT in a real-world clinical setting. CAMHS clients had experienced multiple traumas, including sexual, emotional and physical abuse. Treatment with trauma-specific interventions was associated with significant improvements in PTSD, general mental health symptoms and greater functioning. A significant number of CAMHS clients required ongoing mental health support in the short and long term. Result from the dissemination and implementation of TF-CBT in a real-world clinical setting. CAMHS clients had experienced multiple traumas, including sexual, emotional and physical abuse. Treatment with trauma-specific interventions was associated with significant improvements in PTSD, general mental health symptoms and greater functioning. A significant number of CAMHS clients required ongoing mental health support in the short and long term.

• CAMHS clients had experienced multiple traumas, including sexual, emotional and physical abuse.
• Treatment with trauma-specific interventions was associated with significant improvements in PTSD, general mental health symptoms and greater functioning.
• A significant number of CAMHS clients required ongoing mental health support in the short and long term.

Background
Childhood trauma is not only prevalent in society it also has long-lasting and detrimental effects on the health and wellbeing of individuals throughout their lives (Felitti et al., 1998;Hughes et al., 2017;Kalmakis & Chandler, 2015). Childhood adversity may include physical, emotional and sexual abuse, as well as neglect and experiencing or witnessing a traumatic event. The Australian Bureau of Statistics (2016) estimated that an overwhelming 2.5 million Australians (13% of the population) have experienced physical or sexual abuse in childhood. In 2004, the PATH (Personality and Total Health Through Life) study in the Australian Capital Territory (ACT) determined that 59.5% of randomly selected residents had experienced some form of an adverse childhood event, with 37% having experienced more than one event (Rosenman & Rodgers, 2004).
In 1998 in the United States, the Adverse Childhood Experiences (ACE) study was conducted in order to examine the connection between psychological, physical and sexual abuse during childhood and measures of adult risk behaviour, health status and disease. Individuals having experienced higher numbers of ACEs were found to have worse physical and mental health outcomes in adulthood (Anda et al., 2006;Felitti et al., 1998). Furthermore, individuals having experienced more than three adverse events in childhood had 4-to 12-fold risk of substance abuse and mental health disorders compared to individuals with no ACEs (Felitti et al., 1998). Experiencing any adverse childhood event has been shown to increase the risk of attempted suicide by two to five times (Dube et al., 2001). Posttraumatic stress disorder (PTSD) in children and adolescents is highly correlated with other mental health disorders such as anxiety, depression and severe behavioural problems, and tends to become chronic (Bolton et al., 2000). The 1998 ACE Study had some limitations worth noting, including most participants (80%) being white and a reliance on self-report data, which can lead to over-or under-representation of adverse events, health conditions and health risk behaviours.
The type of adversity experienced in childhood may also influence adult health outcomes. Emotional or psychological abuse has been most strongly correlated with negative mental health outcomes (Cecil et al., 2017;Taillieu et al., 2016), as well as behavioural problems and risk behaviours (Spinazzola et al., 2014). Sexual abuse has been associated with poorer functioning (Archer et al., 2017), while childhood physical abuse is associated with lower IQ (Cecil et al., 2017). However, these conclusions are difficult to interpret as studies have shown that children were more likely to experience multiple ACEs rather than just one. Sugaya et al. (2012) reported that of the participants experiencing physical abuse in childhood, 79% also experienced sexual abuse and 76% experienced parental psychopathology. Subsequently, numerous studies have found that there is a cumulative effect of maltreatment, in that the number of adverse events in childhood strongly correlates with greater symptom severity of poor health outcomes in adulthood (Cecil et al., 2017;Felitti et al., 1998;Mock & Arai, 2010;Spinazzola et al., 2014;Sugaya et al., 2012).
Prior research by the authors, referred to as 'Phase 1', explored the prevalence of trauma adversity in the ACT Child and Adolescent Mental Health Service (CAMHS), by surveying 162 young people (and their parents/guardians) attending the service. Results showed that 69% of clients reported having experienced physical, emotional or sexual abuse, neglect or another traumatic event, and it was more common for the children to have experienced three or more adversities than a single event (Reay et al., 2015). It was also evident that a history of trauma was transgenerational, in that the children who had experienced trauma were likely to have a parent who had also experienced a potentially traumatic event (Reay et al., 2015).
The strong association between childhood trauma and adulthood health raises the question as to how well mental health services recognise, respond to, and treat clients who have been exposed to trauma. A systematic review determined that abuse (emotional and sexual) and neglect (emotional and physical) were up to nine times more likely to be discovered by researchers than the initial mental health service (Read et al., 2018). Another study revealed that only 37% of mental health patients were asked about childhood abuse on assessment, due to the clinicians reporting they did not have the confidence to address and manage the past trauma (Mansfield et al., 2017). In light of the finding that mental health clinicians feel ill-equipped to deal with the trauma histories that underpin their clients' symptoms and presentations, there has been a growing interest in traumainformed care (TIC) and trauma-specific interventions. Trauma-informed organisations recognise the high prevalence of trauma exposure amongst their clients, understand the association between trauma exposure and lifelong functioning and actively work to reduce the possibility of retraumatising clients (Conners-Burrow et al., 2013;Hodas, 2006). It is essential that therapeutic services that work within a trauma-informed network have access to traumaspecific interventions or services (Cohen & Mannarino, 2008;CATS Consortium, 2007;Harper et al., 2008). According to Hanson and Lang (2016, p 97), TIC involves three primary domains: 'workforce development (training, awareness, secondary traumatic stress), trauma-focused services (use of standardised screening measures and evidence-based practices); and organisational environment and practices (collaboration, service coordination, safe physical environment, written policies, defined leadership).' In order to follow best practice, in 2014 the authors implemented a programme of training, research and supervision within ACT CAMHS, referred to as 'Phase 2'. The first stage of the programme was to train all staff in traumainformed care (see Palfrey et al., 2019). Following this training, staff were trained in the trauma-specific intervention Trauma-Focused Cognitive Behaviour Therapy (TF-CBT; Cohen et al., 2004).

Trauma-Specific Intervention
Trauma-Focused Cognitive Behavioural Therapy (TF-CBT) The efficacy of TF-CBT has been demonstrated in 20 randomised controlled trials and more than 50 published studies (see Cohen et al. (2017) for a review of studies). These studies have involved children and young people (aged 3-18) who experienced single or multiple incidences of trauma or complex trauma. The studies demonstrate the effectiveness of TF-CBT for a range of traumatic experiences including physical abuse, sexual abuse, neglect, family and community violence, natural and man-made disasters and traumatic grief (Cohen et al., 2004O'Callaghan et al., 2013). According to Cohen and Mannarino (2015, p 3), TF-CBT incorporates gradual exposure therapy in a phase-based treatment, of which the core phases include 'stabilisation, trauma narration and processing, and integration and consolidation', while including parents or caregivers to encourage family resilience. Children with dangerous behaviours and active substance abuse (in the child or their caregivers) are usually not suitable for TF-CBT. These children and their families may require an alternative intervention until their problems are treated or stabilised (Pollio & Deblinger, 2017). TF-CBT treatment is associated with significantly improved mental health symptoms, specifically in the treatment for PTSD, depression, anxiety, behavioural and cognitive issues, and relationships (Cohen & Mannarino, 2008de Arellano et al., 2014;Konanur et al., 2015;Zhou et al., 2015). The compelling evidence for the efficacy of trauma-specific interventions prompted the authors to investigate the implementation of TF-CBT to treat children and young people attending a community mental health service.
Knowledge translation (the practice of incorporating research findings into clinical practice) is a highly desirable goal of health services. However, it is often difficult for clinicians due to lack of time, expertise, funding and resources (Forman-Hoffman et al., 2017). The usual practice is to conduct research in academic settings by highly trained and supervised therapists with low caseloads. Yet it is well recognised that findings from efficacy studies under carefully controlled conditions with highly selected patients may not work in the real world of clinical practice. When mental health clinicians participate in research, they are more likely to adopt evidence-based practices, resulting in better health outcomes for their clients (Jennings & Walsh, 2013). According to Orygen (2016), knowledge translation is an ongoing process that involves synthesising the evidence and converting and disseminating insights from research into effective real-world practices. Thus, Phase 2 involved the authors working collaboratively with the clinicians and management of ACT CAMHS to design and implement a feasible study that would capitalise on the (1) clinicians' skills and expertise, (2) existing resources and (3) policies and practices of the service.
According to Cohen et al. (2018), there is a pressing need for more evidence demonstrating the long-term mental health outcomes following TF-CBT. Mannarino et al. (2012) showed that the positive improvements in 14 measures following treatment with TF-CBT for childhood sexual abuse were sustained at 6 and 12 months. However, there is very limited evidence beyond 12 months. It would be pertinent to note whether, following completion of TF-CBT, patients fully disengage from mental health services, require regular maintenance therapy sessions, or experience relapses in their mental health status (such as mental healthrelated hospitalisations). Furthermore, this study will provide unique information on the implementation of TF-CBT in settings where clinicians combine their experience and clinical judgement with the therapy framework to meet the needs of the clients they serve. This paper aims to address this gap in the literature by reporting on these outcomes, including at 3-month and 2-year follow-ups.

Aims and Hypothesis
This study aims to test the effectiveness of a traumafocussed cognitive behavioural therapy (TF-CBT) for children and adolescents exposed to trauma, when implemented in a community setting. Specifically, the study aims to: 1. Quantify the number and type of traumas and adversities reported by clients receiving the intervention. 2. Investigate whether the intervention is associated with improvements in participants' PTSD symptoms, general mental health symptoms and functioning. 3. Report study participants' contact with CAMHS at 3 months and 2 years post-treatment.
We hypothesised that CAMHS clients receiving the trauma-specific intervention would experience a significant improvement in PTSD symptoms, general mental health symptoms and functioning following treatment.

Participants
Participants were recruited from the Child and Adolescent Mental Health Service (CAMHS), Canberra, ACT, Australia. Twenty-six children and adolescents were recruited to the study by their treating professionals, including psychologists, social workers, nurses and occupational therapists. Children and adolescents were eligible to participate if they were aged 8-17 years, had been exposed to a potentially traumatic event and were assessed as suitable for a course of TF-CBT by their clinician. Written and informed assent was obtained from all participants (and consent obtained from their parent/guardians) by their treating health professional, in accordance with the ACT Human Health Ethics committee which approved all protocols. Mature minors (aged 16 or more) estranged from their caregiver could provide their written assent in order to participate. Exclusion criteria included children with an intellectual disability or psychotic symptoms. Note: participants were not required to complete a certain number of pre-treatment sessions to be eligible for the study.

Study Setting
The CAMHS service is a tertiary referral setting in a major metropolitan centre. It provides a community mental health service for children and adolescents aged 0-17 with moderate to severe mental health problems. A recent audit revealed approximately 350 new referrals annually in the eligible age range (9-17 years) with slightly more females (55%) than males. The audit found that approximately 50% of new clients engage in any type of therapy (n = 175) whilst the remainder attend for a few sessions, disengage, or are referred elsewhere. In terms of family demographics, recent surveys of our population found two-thirds of young people lived with a parent with a mental illness (66%), while less than half lived in a two-parent family (43%) and approximately 6% lived out of home. Whilst we did not collect family income data, 22% were not in paid employment and 18% reported experiencing financial hardship.

Procedures
The study team consisted of clinician-researchers with extensive experience working in community mental health settings. The team worked in a collaborative way with the administration, clinical staff and management of the ACT CAMHS to design and implement the study. Information flyers, posters and in-service presentations were developed to promote the study aims, targets, methods and procedures. Research packs were prepared and placed at the two participating CAMHS sites. These contained the study information sheet, information flyer and appropriate consent forms. Clients and their caregiver/s were invited to participate in the research by their allocated Clinical Manager at their initial appointment. A prepost study design was used to examine changes in clients' PTSD symptoms, general mental health symptoms and functioning following the intervention. The clinician, young person and their caregiver completed assessment measures at baseline (pre-treatment), after 6-8 sessions of therapy (mid-treatment) and at completion of intervention/discharge from the service (post-treatment). Consistent with prior research methods used in community settings, if a closing assessment was not completed, then the last follow-up assessment was carried forward as post-treatment data (see Lang et al., 2015). These routine measures were administered by the treating clinician, reflecting current clinical practice. Client files were retrospectively accessed by the authors, from patient electronic medical record. (MAJICeR), and data was collected by thoroughly examining all documentation provided by the treating psychiatrists, psychologists and social workers. Information extracted for each client included the following: medications, diagnoses, types of traumas experienced, therapy type, number of therapy sessions, reason for file closure, number of hospital presentations (with or without admission), record of self-harm, suicidal ideation and suicide attempts. We also recorded any further contact or therapy pursued and any hospital presentations at the 3-month and 2-year points following completion of treatment. This data was inputted into an Excel spreadsheet and coded appropriately for analysis.

Trauma-Focused Cognitive Behavioural Therapy (TF-CBT) protocol
The psychosocial treatment model, Trauma-Focused Cognitive Behavioural Therapy (TF-CBT) (Cohen et al., 2004), is designed to treat PTSD symptoms and related emotional and behavioural problems in children and adolescents. The model aims to address psychological trauma in children who have experienced a variety of traumatic experiences. Trained therapists provide individual sessions with children, and parallel sessions with their parent/s or guardian/s, with additional parent-child sessions progressively integrated over the course of treatment. The components of the TF-CBT model are reflected in the acronym PRACTICE: Psychoeducation and parenting skills, Relaxation skills, Affect expression and regulation skills, Cognitive coping skills and processing, Trauma narrative, In vivo exposure (when needed), Conjoint parent-child sessions, and Enhancing safety and future development (Cohen & Mannarino, 2015). TF-CBT is usually provided for 12-16 sessions of individual and parent-child therapy; however, it is often used flexibly in clinical settings, provided in a longer-term or group therapy format. Treatment was delivered by 10 clinical staff from the CAMHS service. The majority were female (n = 9, 90%) and comprised of psychologists (n = 8, 80%) and social workers (n = 2, 20%). The majority of therapists had more than 6 years clinical experience (n = 6, 60%) (range: 1-20 years). Therapists were trained in TF-CBT including completion of the online training (Medical University of South Carolina MUSC, 2017). This was followed by two days of face-to-face training sessions with Dr Laura Murray, an international expert in adapting TF-CBT to clinical settings (Murray & Skavenski, 2012). Six senior, experienced clinicians were trained by Dr Murray to provide group and/ or individual TF-CBT supervision to CAMHS clinicians. Supervisors were clinical psychologists and allied health workers with more than 15 years clinical experience in CAMHS. Adherence was addressed during supervision by reference to the treatment manual and by reviewing clinical notes.

Clinical Measures
Child PTSD Symptom Scale (CPSS) The CPSS (Foa et al., 2001) is designed to assess PTSD diagnosis and symptom severity in children aged 8-18 years who have experienced a traumatic event. The questionnaire consists of 24 items, of which 17 items correspond to the Diagnostic and Statistical Manual of Mental Disorders -4th Edition (DSM-IV) symptoms of PTSD (American Psychiatric Association AMA, 2000). The 17-item scale used in this study ranges from 0-3 depending on how often the symptom bothers the young person. The higher the score, the more severe the symptoms of PTSD. A clinical cut-off score of 15 or more is recommended for PTSD. The scale consists of three subscales to measure trauma symptoms of re-experiencing, avoidance/numbing, and hyperarousal. The CPSS has demonstrated strong psychometric properties (reliability, validity and internal consistency) in samples of children and adolescents with both single-incident traumas and complex trauma histories (Foa et al., 2001;Gillihan et al., 2013). Cronbach's alpha for this sample was 0.86. Clinically significant change for the CPSS has been assessed as a reduction of 11 points or more (CHDI, 2021).

Health of the Nation Outcome Scales for Children and Adolescents (HoNOSCA)
The clinician-rated HoNOSCA (Gowers et al., 1999) focuses on clinically significant problems and symptoms in children and adolescents. It consists of four subscales: impairment, behavioural, symptomatic and social problems, each rated from 0 (no problem) to 4 (severe to very severe problem). The HoNOSCA total score is the sum of the first 13 items and indicates the severity of the mental health problems. With a possible total score of 0-52, the higher the number, the more severe the problems associated with mental illness. HoNOSCA has been found to be a sensitive, valid measure (Garralda et al., 2000), with adequate reliability and utility (Pirkis et al., 2005) to assess clinical outcomes within children and adolescents' mental health services. The HoNOSCA showed acceptable internal consistency for this sample: Cronbach's = 0.77.
Strengths and Difficulties Questionnaire (SDQ): self-report (SDQ-self) and parent report (SDQ-parent) The SDQ is a brief screening questionnaire for child mental health problems which can be completed by parents, teachers and children (Goodman, 2001). The self-report scale can be completed by 11-16-year-olds, while the parent report can be completed for children aged 4 years and older. The SDQ consists of five subscales: emotional problems, conduct problems, hyperactivity, peer problems and prosocial behaviours. Each item is scored 0 (not true), 1 (somewhat true) or 2 (certainly true). A total difficulties score is formed by adding all subscales, except for prosocial. Children with higher total difficulty scores indicate the presence of a clinical disorder. The SDQ is used extensively in child and adolescent mental health clinical and research settings. Research supports its use as a dimensional measure for comparing mean scores in child mental health settings after an intervention (Goodman & Goodman, 2009), with satisfactory reliability and validity (Goodman, 2001) and moderate to strong internal reliability and stability (Hawes & Dadds, 2004) for both self-and parent-reports. Cronbach's alphas for the self-report and parent-report scale used in this sample were 0.80 and 0.81, respectively.

The Children's Global Assessment Scale (CGAS)
The CGAS is a clinician-rated measurement of general functioning in children and adolescents (Shaffer et al., 1983). The scale is a continuum of descriptive parameters ranging from 0 to 100: 'needs constant supervision' to 'superior functioning in all areas'. The higher the score, the higher the level of functioning. Although CGAS is often used in child mental health services, it has been shown to have only moderate inter-rater reliability compared to other measures such as HoNOSCA (Hanssen-Bauer et al., 2007). Therefore, in our analysis we transformed the data into categories of lower functioning (CGAS = 1-40), sporadicmoderate functioning (CGAS = 41-60) and higher functioning (CGAS = 61-100). The CPSS was used as a measure of PTSD symptoms, HoNOSCA and SDQ (-self and -parent) were used as measures of general mental health, and CGAS was used as a measure of general functioning.

Data Analysis
The data were analysed using IBM SPSS Statistics Version 25 (IBM Corp, 2017) predictive analytics software. To test whether CAMHS clients experienced significant improvements in indicators of PTSD symptoms, general mental health and functioning, we conducted repeated measures analysis of covariance or paired samples t-tests using completed measures at two assessment points: pre-treatment (Pre) and either mid-or post-assessment (Post). Data were tested for normality using Kolmogorov-Smirnov tests. Outliers on any skewed scales were Winsorised (Ruppert, 2006) due to the clinical importance of including all participants in the sample. Bivariate correlations were used to explore relationships among the outcome variables and any potential covariates. Independent samples t-tests (continuous variables) and χ 2 tests of independence (categorical variables) were conducted to explore baseline comparability of study completers and non-completers for sample bias. Mixed between-within subjects analysis of variance was conducted to assess differences in mean measures, controlling for any potential covariates. Based on the preliminary analyses, specified moderators/covariates were examined for each outcome variable. Where interaction effects (moderator × time) were not significant, these variables were removed (to avoid reducing N). The analysis was then repeated to investigate main effects. Assumptions of homogeneity of variance between groups were met using Levene's test.

Sample and Participant Flow
Twenty-six youths consented to participate in the study and receive TF-CBT. Without research consent, no reliable data were available on those who were suitable for TF-CBT but declined to be involved. One participant dropped out early after completing three sessions and one did not return complete data (15 sessions). All 26 participants had a parent or caregiver who gave their written consent; only one participant (3.8%), a mature minor who attended the research interview without their parent, provided their own assent to participate. Of the 26 participants, 24 (92%) completed a pre-assessment HoNOSCA, 16 (61.5%) completed a midassessment, and 23 (88.5%) completed a post-assessment. There was a reduced sample size for paired pre-and postassessments for HoNOSCA (n = 23), SDQ-self (n = 21, valid for 11-16-year-olds only), SDQ-parent (n = 16) and CPSS (n = 7). See Fig. 1 for study flowchart. The time from pre to post treatment ranged from 8 to 56 weeks. The mean age of study completers was 15.6 years (range = 9-17), with more females (87%) than males (13%). The mean number of therapy sessions was 17.6 (range = 4-42), with 62.5% of clients receiving TF-CBT as their main therapy type, 33.3% receiving CBT without the trauma narrative/ exposure components and 4.2% receiving components of Dialectic Behavioural Therapy (DBT). A χ 2 test showed that there was no significant difference between completers and non-completers in terms of gender (females: 85.7% vs. males: 50%), χ 2 (1, n = 26), p = 0.289. Independent samples t-tests comparing pre-treatment outcome measures (HoNOSCA, SDQ-self, SDQ-parent, CGAS and CPSS) found there were also no significant differences between the two groups. There was a trend towards older completers at pre-treatment (M = 15.6, SD = 2.14) than non-completers (M = 12.0, SD = 4.24, t (24) = 2.17, p = 0.04).
Demographic information of study participants was collected by clinical audit and refers to trauma, mental health diagnoses and self-harm or suicidal behaviour occurring in the period prior to and during therapy. This information is presented in Table 1. It was more common for CAMHS clients to have experienced multiple traumatic events than single events. Participants reported an average of three traumatic events (SD = 1.3), with totals ranging from 1 to 6 Invited to participate in study n = 26  events. The most common events reported were sexual abuse (70.8%), emotional abuse (54.2%) and physical abuse (50%). All clients had a past or current history of suicidal thoughts (100%), 91.7% had self-harmed, 50.0% had attempted suicide and 45.8% had mental health-related hospital presentations (prior to commencing TF-CBT). In terms of diagnoses, clients had an average of three current mental health diagnoses recorded in their clinical file (SD = 1.1, range: 1-5). Anxiety disorders were the most frequently recorded (79.2%), followed by depressive disorders (75%) and PTSD (62%).

Pre-Treatment Comparisons
Most variables were normally distributed with the exception of age (positively skewed), with a trend toward older adolescents in the sample. There was a non-significant trend showing the higher the total number of traumas experienced, the more PTSD symptoms experienced at pre-treatment (CPSS) (n = 11, r = 0.592, p = 0.055). All participants with pre-treatment CPSS scores were above the clinical cut-off of 15 or more for PTSD (M = 32.6, SD = 9.0, range 17-45). Pre-treatment clinician-rated HoNOSCA scores were negatively correlated to pre-treatment clinician-rated CGAS scores (n = 24, r = −0.466, p = 0.022), which was expected, as the more severe the mental health problems a clinician observed, the lower they rated the general functioning of the client. Furthermore, pre-treatment self-reported SDQ scores and parent-reported SDQ scores were positively correlated (n = 21, r = 0.599, p = 0.008), which indicates that the client and their parent designated similar scores for the client's strengths and difficulties. Pre-treatment CGAS scores were negatively correlated with CPSS scores (Post). Table 2 reports the pre-treatment means and standard deviations for all continuous variables and bivariate correlations among variables. CGAS raw scores were transformed into the following three categories: lower functioning (CGAS = 1-40), sporadic-moderate functioning (CGAS = 41-60) and higher functioning (CGAS = 61-100). Separate mixed betweenwithin analyses of variance (ANOVA) were conducted to assess for changes in study outcome variables over time (HoNOSCA, SDQ-self, SDQ-parent and CPSS) using CGAS category as the between-participants factor and Time as the within-participant factors. In the analysis exploring changes in SDQ-parent scores, we also controlled for pretreatment SDQ-self scores (which was a covariate).

Intervention Effects
In the initial analysis, separate mixed-design repeated measures (RM) ANOVAs were conducted to test for change in each outcome measure, including pre-treatment CGAS scores as a moderator (low/moderate/high) and child age as a covariate. Results showed that there were no significant Time x CGAS interactions for any of the outcome variables, thus, the analyses were re-run without these interaction terms. In other words, changes in outcome variables were similar for youth classified as lower, moderate or higher functioning. In terms of the RM ANOVA, mean changes for all outcome measures were in the expected directions. There was a significant reduction in clinician-rated HoNOSCA scores over time (main effect for Time: F (1, 22) = 20.33, p < 0.001, partial η 2 = 0.48) with a large effect size. HoN-OSCA subscales also significantly improved in three out of four scales: behavioural, symptomatic and social problems. SDQ-self scores also significantly improved with a large effect size (main effect for Time: F (1, 20) = 17.80, p < 0.001, partial η 2 = 0.47). Results all showed statistically significant reductions in all four SDQ-self subscales for emotional, conduct, hyperactivity and peer problems. There was a significant change in mean SDQ-parent scores over time, F (1, 15) = 0.7.74, p < 0.05, partial η 2 = 0.34 (large effect size) and for the peer problems subscale (p < 0.05, medium effect size). PTSD symptoms (CPSS) also significantly reduced over time (main effect for Time: F (1, 6) = 45.90, p < 0.001, partial η 2 = 0.88), with large effect size. There were significant reductions in all three CPSS subscales: re-experiencing, avoidance/numbing and hyperarousal symptoms (with large effect sizes). Given the small sample size for the CPSS, we examined change at the individual level following the recommendations for reliable change criteria. Results showed 71.4% (4/7) of participants showed clinically significant change in posttraumatic stress symptoms following TF-CBT, Pre and Post means, standard deviations and results of the mixed-design RM ANOVA main effects are reported in Table 3.

Secondary Outcomes: Children's Global Assessment of Functioning (CGAS)
To assess changes in clinician-rated CGAS scale scores, McNemar's tests were conducted to compare the proportions of clients in each functioning category (lower/ sporadic-moderate/ higher) at Pre and Post. Tests showed that compared to prior to the intervention, significantly more clients were classified as higher functioning after the intervention (Pre: 21.7 % vs. Post: 57 %, p = 0.001).

Short-and Long-Term Outcomes
To assess short-and long-term outcomes with TF-CBT, we conducted a chart audit to obtain clinical information at three time points: during treatment, at 3-month, and 2-year follow-ups (see Table 4). Clients who were discharged and had no evidence of further therapy, or a referral to another service were classified as 'no further contact' and those who were referred to another part of the service were classified 'attending public mental health services' (child and adolescent, or adult services). Finally, those referred to private therapists, headspace, women's health services, school psychologists, etc. were classified as 'referred elsewhere'. Results showed no clients were admitted to hospital during treatment with TF-CBT, although three adolescents presented in crisis due to thoughts of suicide or self-harm and HoNOSCA Health of the Nation Outcome Scales for Children and Adolescents, SDQ-self Strengths and Difficulties Questionnaire-Self report, SDQ-parent Strengths and Difficulties Questionnaire-Parent report, CPSS Child Posttraumatic Stress Symptom Scale (includes total scores and subscales) *p < 0.05; **p < 0.01; ***p < 0 0.001 Child and adolescent or adult services were managed in the community. At 3 months posttreatment, 33.3% of clients were still accessing public mental health services, with 41.7% at 2 years. More than one-third (37.5%) of clients at 3 months and 8.3% at 2 years were referred elsewhere. There was 'no further contact' at 3 months for 29.2% of clients and at 2 years for 50% of clients. Prior to and during therapy, 45.8% of clients presented to hospital for mental health reasons, with only 16.7% recording presentations at 3 months, increasing to 29.2% of clients at 2 years.

Discussion
The overall aim of the study was to evaluate the effectiveness of a trauma-specific intervention when implemented in a child and adolescent community mental health service. With respect to the first aim, this study outlined the demographics of the trauma-exposed clients at CAMHS. The complexity of the clients was evident in the number and type of traumas they experienced, as well as their consequent mental health outcomes. For example, the number of therapy sessions was high compared to research settings (Deblinger et al., 2011) and some clients took more than a year to complete TF-CBT treatment. Results showed that CAMHS clients were more likely to have experienced multiple traumatic events than single events. These findings support the findings of Reay et al. (2015), who studied the same mental health service in the ACT, as well as other external studies (Felitti et al., 1998;Sugaya et al., 2012). Our results further showed that 100% of the trauma-exposed CAMHS clients had experienced suicidal thoughts and the majority had selfharmed or attempted suicide. These data emphasise the necessity for trauma-specific intervention at CAMHS. Although the participants were assessed as eligible for TF-CBT, not all clients received the intervention as intended. Our chart audit revealed there was no evidence of the trauma narrative or exposure components for eight participants, and one participant received components of DBT. Our finding that 37.5% did not receive all components of the therapy as intended is consistent with prior research. For example, Allen and Johnson (2012) reported that only two-thirds of therapists used all of the TF-CBT components in practice.
Research has also shown that PTSD symptoms can improve whether or not the trauma narrative is included (Deblinger et al., 2011), indicating it may not be necessary to include all components of the therapy to achieve recovery. The second aim intended to determine whether mental health outcomes were improved following traumaspecific intervention, of which this study showed that treatment was associated with significant improvements in PTSD symptom severity (CPSS) from pre-to posttherapy with a large effect size. Improvement in all three CPSS subscales is evidence that the overall results were driven by improvement in trauma symptoms of reexperiencing, avoidance/numbing and hyperarousal. Furthermore, the finding that 71% of subjects experienced a clinically significant improvement in trauma symptoms suggests that treatment had a marked or clinically meaningful impact on the participants. On the other hand, the small sample size for the CPSS means these results should be treated with caution. The reduction in trauma symptoms following trauma-specific interventions in youth community settings supports previous literature (Cohen & Mannarino, 2015;de Arellano et al., 2014;Konanur et al., 2015;Rudd et al., 2019;. Overall, participants experienced improvements in their general mental health symptoms following therapy as reported by their clinicians (HoNOSCA), their guardians (SDQ-parent) and the youth themselves (SDQself). All effect sizes were large. These findings are also consistent with prior research into TF-CBT (Anda et al., 2006;Cohen & Mannarino, 2008;Jensen et al., 2014;Rudd et al., 2019). The finding that the majority of HoNOSCA subscales and all SDQ-self report subscales showed significant improvements provides evidence that treatment targets a number of symptoms and problems. Whereas the improvements on the SDQ-parent seems to be largely related to their perceptions of improvements in their child's peer relationships. Others have speculated that parents notice changes in this domain because they are more directly observable than others (Conn et al., 2016).
The general functioning of the clients was measured using clinician-rated CGAS. In our study, we categorised clients into lower, moderate and higher functioning, based on their pre-treatment CGAS score. It was evident that significantly more clients were classified as higher functioning after the intervention. This is a clinically important result as higher functioning children have more secure and meaningful relationships with their family, school and peers (Shaffer et al., 1983). In addition, they are more involved in extra-curricular activities and interests, and when they experience difficulties, these are more likely to be transient, sporadic or in a single area, with less impact on overall functioning. This study found participants with lower functioning pre-treatment showed significantly lower SDQself and higher CPSS scores than the higher functioning group at post-treatment. Supporting this, Wolpert et al., (2020) showed that children with higher quality of life at pre-treatment are more likely to show mental health improvement following intervention. This finding may be explained by the higher number of adversities and mental health issues that lower functioning clients experience. These clients may require more time and more therapy sessions in order to achieve the same result as higher functioning clients.
The final aim was to describe the therapy outcomes during treatment, at 3 months and 2 years following traumaspecific intervention for the CAMHS clients. No client was hospitalised during treatment, although clients experienced thoughts of suicide and self-harm, underscoring the importance of comprehensive crisis support services. It was determined that although mental health outcomes and PTSD symptom severity were improved following trauma-specific interventions, some CAMHS clients still required ongoing therapy. Three months following closure of therapy, 29.2% of clients had no known further contact with our mental health service (CAMHS or adult services), which increased to half of the clients after 2 years. Of those who were still receiving therapy, a higher proportion of clients re-entered the public mental health system at 2 years, which was mirrored in the number of mental-health-related hospital presentations increasing from 3 months to 2 years postclosure. These findings are to be expected due to the complexity of the trauma-exposure and young age of the clients. The clients' age is especially pertinent as they are still within emotional, social and cognitive developmental periods, and serious mental health issues for children and adolescents can result in delayed psychosocial development and increases their risk of long-term psychological morbidity (Skehan & Davis, 2017). PTSD symptom reduction has been demonstrated to be maintained 6-12 months following trauma-focused therapy (Cary & McMillen, 2012;Konanur et al., 2015), while reductions in other mental health symptoms, such as depression, are not maintained (Mannarino et al., 2012), providing a possible explanation for the return of CAMHS clients to further therapy. Parental support has also been shown to significantly improve mental health outcomes at 6-12 months following therapy (Cohen & Mannarino, 2015). The trauma-specific intervention that the CAMHS clients received was only one episode of care, which was shown to decrease PTSD symptom severity in this study. It is important to note that 42% of participants required follow-up care from public mental health services. Planning for episodic or maintenance treatment is an important aspect of follow-up care for CAMHS clients.

Limitations
The complex real-world context of the study increased the variable delivery of the intervention, thus there are several limitations to note. For example, we did not have a comparison group which received usual care, in order to investigate whether TF-CBT is more effective or beneficial. Clients were inconsistent in therapy attendance: some discontinued early or switched between therapy types. Outcome measures were not consistently recorded at certain time points during therapy and due to the novel trauma-specific intervention at CAMHS. At the time of the study, PTSD symptoms were not routinely measured via CPSS, thus we had a small sample size for this analysis. The study may have not been adequately powered to assess for interaction effects. Furthermore, this study aimed to replicate clinical practice by allowing the treating clinician to assess the HoNOSCA and CGAS measures; therefore, there were no independent measures undertaken. Although all clients were given trauma-specific interventions, they did not all receive TF-CBT in its entirety, and due to the small sample sizes, we could not determine differences between therapy types in client outcomes. On the other hand, these issues are common in real-world effectiveness studies, which involve more heterogeneous client populations and less-standardised treatment protocols. There are also less restrictions on therapists in clinical settings who tailor treatments (e.g., duration, dose, co-therapies), using their clinical judgement. The study reflects community delivery of trauma-informed interventions, supporting the external validity of our findings.
For the clients that did not complete therapy, there was no feasible way of following up to determine their reasons for dropping out or to obtain outcome measures. Although the 3-month and 2-year follow-up data was obtained via a retrospective clinical audit, the study design could have been strengthened by the use of follow-up questionnaires and outcome measures. The trauma-exposed CAMHS clients within this study were positively skewed towards an older age and were predominantly female (87%). However, the gender imbalance and skewed age range in this study reflects the demographics of CAMHS in general: using the same CAMHS service, Reay et al. (2015) determined 87% of all CAMHS clientele were 11-16 years old and only 13% were 6-10 years old, with 72% being female. It is also known that females are more likely to experience PTSD symptoms and associated risk factors (Christiansen & Hansen, 2015) and females are more likely to seek help for mental health issues than males (Liddon et al., 2018).

Future Research
Future research should aim to address the limitations of this study. Firstly, clinicians could have periodical reminders to obtain outcome measures, as well as ensure PTSD symptoms are routinely measured using CPSS. Follow-up data should be obtained during a therapy session with outcome measures recorded at 3 months and 2 years post-completion of therapy. An increased sample size would increase the number of male clients and younger clients; thus, comparisons could be made between gender and age groups in terms of mental health outcomes. Furthermore, determining differences in clinical outcomes between clients offered TF-CBT compared to other trauma-specific interventions would be noteworthy. Future research should also investigate which children do not respond well to the intervention. It would also be interesting to obtain information about the parents of the CAMHS clients in terms of their mental health and past traumas and adversities, as studies have shown children of parents with mental health problems are more at risk of developing their own mental illnesses (Rutter & Quinton, 1984). Our sample sizes were too small to determine the cumulative effect of trauma on mental health outcomes. It would be beneficial to add to the existing research that demonstrates cumulative trauma contributes to an increased risk of negative health outcomes in a dose-dependent manner (Dube et al., 2001;Felitti et al., 1998;Mock & Arai, 2010;Sugaya et al., 2012).

Conclusions
Following the implementation of a trauma-specific intervention to treat trauma-exposed children and adolescents in a community mental health service, clients showed significant improvement in PTSD symptoms, general mental health and general functioning. A significant proportion of clients had engaged in additional service usage at 2 years following completion of trauma-focused therapy, suggesting that additional supports are required long-term to fully maintain the improvement in mental health of traumaexposed clients following this implemented therapy. The high proportion of clients having experienced multiple traumas and adversities, and the consequent mental health issues, highlights the importance of further research into the use of trauma-specific therapies in community mental health settings.
Funding Open Access funding enabled and organized by CAUL and its Member Institutions.

Compliance with Ethical Standards
Conflict of Interest The authors declare no competing interests.
Consent to Participate Written and informed assent was obtained from all participants (and consent obtained from their parent/guardians). Mature minors (aged 16 or more) estranged from their caregiver could provide their written assent.
Ethics Approval This study was approved by the ACT Health Human Research Ethics Committee on 19 January 2015 (ETH.9.14.239).
Amendments to the protocols were approved on 3 November 2015 and 25 March 2020.
Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons. org/licenses/by/4.0/.