Participants and procedure
All participants were treated at the psychotherapeutic outpatient clinic of the Catholic University Eichstätt-Ingolstadt. The inclusion criteria were: (1) arrived in Germany unaccompanied and under the age of 18, (2) current age no older than 21, (3) PTSD diagnosis according to the Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-5) , (4) living in a facility run by the German child and adolescent welfare (CAW) agency, (5) stability of living situation (at least 4 weeks in the current group home), and (6) availability of a caregiver to take part in assessment and psychotherapy. Youths were excluded from study participation in the case of (1) acute suicidality or risk of harm to others, (2) acute life-threatening self-harm, (3) bipolar disorder, (4) psychotic disorder, and (5) acute substance abuse. Caregivers who accompanied participants to treatment were professionals (e.g. social workers), who worked in the CAW facilities where participants lived. They had to have known the patient for at least 4 weeks and the patient had to see them as trustworthy. To ensure that this was the case, we added the inclusion criteria 4, 5, and 6. Furthermore, as PTSD treatments are known to work best in persons with a PTSD diagnosis, we decided to include only URMs with a full-blown PTSD. The reason we included participants up to the age of 21 is that, in the German health care system, child and adolescent psychotherapists are allowed to treat young adults up to the age of 21.
Participants were generally referred by staff from the CAW facilities where they lived. Youths and their respective caregivers were invited to an initial meeting with the first author, where the treatment and the study were explained to them and a first screening took place. Interpreters were on hand to assist during the appointments whenever necessary. If screened positively, the next step was the pretreatment assessment (T1). If the inclusion criteria were confirmed, the youth was offered the intervention (Fig. 1). We conducted assessments 1 week (T2), 6 weeks (T3) and 6 months (T4) after the end of treatment. Participants received vouchers as an incentive for T3 (10€) and T4 (15€) assessments.
The study was conducted between March 2015 and July 2017 and was approved by the ethical review board of the Catholic University Eichstätt-Ingolstadt. Informed consent was given by the youth, the caregiver, and—in the case of minors—by their legal guardian.
Measures taken to reduce attrition
As shown above, URMs constitute a difficult target group for methodologically sound intervention research. To make it easier for URMs to engage and stay in treatment, we involved trusted caregivers from the initial interview onward, and made sure that interpreters were available where needed and seen as trustworthy by the participants. This also involved the participants being able to choose the interpreter’s gender. Furthermore, we took great care to educate participants about psychotherapy in general and about confidentiality in particular (information sheets were prepared in several languages to this end and handed out at the initial meeting). In addition, participants were given a 10€ (T3) or 15€ (T4) voucher as an incentive to participate in follow-up assessments.
We regularly asked for informal feedback on assessment and therapy. Formal feedback involved participants’ rating of assessment-related experiences (RARE; Rimane & Vogel, unpublished test) after baseline diagnostics which led for example to a reduction in the number of questionnaires. Please refer to the Measures section for further information.
TF-CBT consists of nine modules that can be illustrated within the acronym PRACTICE . The first five components, psychoeducation and parenting skills, relaxation, affective modulation, and cognitive processing, are trauma-focused stabilisation skills to prepare patients for describing their personal trauma experiences and to cope with their symptoms related to these experiences. This is followed by the trauma narrative and cognitive processing II (in sensu exposure), and in vivo exposure work. After the narrative has been processed, there is a conjoint child/caregiver session and a module focusing on enhancing safety and future skills in order to integrate the traumatic events into the child’s life . TF-CBT is trauma-focused; it emphasizes the need for caregiver involvement and skills, and works with graduated exposure from the very beginning. The TF-CBT manual suggests a 1:1 ratio of child and caregiver sessions. However, this can be modified according to the patient’s age. In this study, the level of caregiver involvement was flexible and modified to the individual participant’s age and need. Participants received a mean of 15 sessions of TF-CBT (100 min each). On average the therapists saw the caregiver in 8 sessions (53.3% of participants’ sessions). In all but one treatment case there was a conjoint session with patient and caregiver. Treatment cases were conducted by eight therapists (one male) who were licensed in Germany or undergoing training to become licensed psychotherapists. All therapists completed the TF-CBT online training in English or German and attended a 2-day TF-CBT training run by a licensed TF-CBT trainer (RR). Therapists underwent in-house supervision biweekly (RR). In addition, they had case consultation calls with one of the treatment developers, Anthony Mannarino, once a month. If therapists missed more than 30% of supervision sessions and/or failed to record any treatment session on videotape, the case was excluded from the trial as adherence to TF-CBT could not be verified (“no TF-CBT”, Fig. 1). Treatment fidelity was checked by two independent raters who randomly viewed three videotaped sessions of each participant. Therapists completed treatment checklists after each session as a self-report measure of adherence and to document changes in the manual course (mean adherence was rated as 82% in URM and 62% in caregiver sessions). An interpreter was present in 55% of treatment cases.
In terms of TF-CBT components and dosage, we carefully documented modifications with the help of treatment checklists and made the following observations. In addition to psychoeducation on PTSD and traumatic events, therapists provided psychoeducation on psychotherapy, working with translators, and a focus on the obligation to preserve confidentiality. In some cases the affective modulation played a major role in the first phase of treatment. For instance, skills had to be introduced already in the first session or more sessions were needed to practice naming and recognizing feelings. The trauma narrative was developed over several sessions. It always started with a time line to structure the traumatic experiences and identify the index event(s). Many URMs had lost family members or had missing persons in their families. Therefore, grief-specific components of TF-CBT  were added after the trauma narrative if necessary. In addition, we used grief specific material for the loss of homeland to address homesickness (e.g. “What I miss and what I don’t miss about Afghanistan”) and to resolve ambivalent feelings. All participants worked with their therapists on “Strategies for a good future” in the last treatment phase. This included helpful strategies learnt in treatment, helpful persons or sentences. In some cases, an emergency safety plan was developed and practiced in the event of a refusal of asylum (i.e. who to call, what actions to take). The involvement of translators did not present any issues in implementing TF-CBT.
The Diagnostic Interview for Mental Disorders in Childhood and Adolescence (Kinder-DIPS) in German  includes a child and caregiver interview. It is deemed to be a valid structured interview for mental disorders in children aged 6 to 19, with good psychometric properties of the German version . The Kinder-DIPS was used to determine PTSD diagnostic status according to the DSM-5  and comorbid diagnoses. We assessed current diagnoses only.
We used the German version of the Child and Adolescent Trauma Screen (CATS)  in the self-reports and caregiver reports. CATS is a screening questionnaire for exposure to potentially traumatic events and PTSD symptoms according to DSM-5. The reliability of the German version is good to excellent  and Cronbach’s alpha in this study was .82 (self-report) and .74 (caregiver report). The cut-off for clinically relevant symptoms is ≥ 21 (range of scores 0–60). In our study 4 events were added to the original 15-item event list, that proved to be relevant for URMs: “several days without enough water or food”, “dangerous transport/travel”, “kidnapping, imprisonment, deportation”, and “laid (forced to or voluntary) violent hands on someone”.
The Mood and Feelings Questionnaire (MFQ)  is a self-report and caregiver report questionnaire to assess depressive symptoms. We used the German short version with 13 items that measures symptoms on a 3-point Likert scale. Cronbach’s alpha in our study was .88 (self-report) and .77 (caregiver-report). The cut-off for clinical relevant symptoms was ≥ 12 (range of scores 0–26).
By using the Strengths and Difficulties Questionnaire (SDQ)  in the self-reports and caregiver reports, we measured 25 behavioural attributes divided into five subscales: emotional symptoms, conduct problems, inattention-hyperactivity, peer problems and pro-social behaviour. The total difficulties score comprises all but the last scale. The SDQ uses a three-point Likert scale. In a British sample reliability was good . In our sample where we used the German version of the SDQ, Cronbach’s alpha was .74 (self-report and proxy report).
The Patient Health Questionnaire Physical Symptoms (PHQ-15) , German version, was used to screen for physical symptoms. As our sample was all-male, we omitted the item on menstrual cramps. The total score ranges from 0 to 30. In this study Cronbach’s alpha was .74.
The Kinder-DIPS was administered by trained bachelor or master level psychologists for both youth and caregiver. Interpreters supported assessments when needed. The CATS, MFQ and SDQ were completed by patient and caregiver on tablet devices. Raters were on hand to assist both participants in case items were difficult to understand and interpreters to make sure all wording was sufficiently understood and could be translated correctly. Therapists did not take part in any of the assessments to avoid biased results. As there was no control group, we could not guarantee full blinding of raters. However, we tried to use different raters for each assessment (T1, T2, T3, T4) whenever possible to prevent them from drawing conclusions about the participant’s treatment status within the study. Originally, we were going to include the Adolescent Dissociative Experiences Scale (A-DES)  and the Screen for Child Anxiety Related Emotional Disorders (SCARED) . However, we dropped these measures due to insufficient validity and reliability, participants reporting difficulties in understanding the items and inappropriate questions (e.g. separation anxiety regarding parents for separated youths). Furthermore, participants gave the feedback that the assessment sessions lasted too long and this was confirmed by raters. Suicidality was assessed after every assessment by a licensed psychotherapist (JU). During treatment, the respective therapist was responsible for screening for suicidality in his/her patient after every session.
We used SPSS statistics version 25 for Windows for all analyses. We report descriptive data for demographic and baseline data and the number of reported traumatic events. The primary outcome (CATS) was analysed using multivariate analyses of variance (MANOVAs, for self-report and proxy report) for the comparisons T1–T2, T1–T3 and T1–T4 separately due to differing sample sizes. We tested changes in PTSD diagnostic status (Kinder-DIPS) using the McNemar test for dependent samples. We used a repeated measures MANOVA (without T4 data due to missing data) and post hoc t-tests to examine symptom reduction regarding secondary outcomes. Given the pilot nature of this study we conducted all analyses with available samples at each time point (“completer sample”) and we reported the sample size at each time point. Furthermore, we used an uncorrected significance level of .05 (2-tailed) for all analyses due to the exploratory nature of the hypotheses. Cohen’s effect size d was calculated for within group comparisons. On the individual level clinically meaningful symptom reduction for the primary outcome (CATS) was assessed using the reliable change index (RCI) . This resulted in changes > 13 points being considered as reliable changes.