Study design

This 2-arm, single-blind RCT was conducted in Amman, Jordan. There are an estimated 1.4 million Syrians residing in Jordan [14]. The study was conducted in collaboration with the Institute for Family Health (IFH), a national humanitarian agency. The project was prospectively registered (Australian and New Zealand Clinical Trials Registry, no. ACTRN12619000341123) and was approved locally by the Ethics Committee of Al Basheer Hospital in Jordan, the University of New South Wales Human Research Ethics Committee, and the WHO Ethical Review Committee. The trial protocol is available in the Supporting information (S1 Text). This study is reported as per the Consolidated Standards of Reporting Trials (CONSORT) guideline (S1 Consort Checklist).

Participants

Participants were recruited by inviting adolescents and their caregivers to participate during door-to-door visits in regions in Amman known to host large numbers of Syrian refugees. Participants were enlisted in the trial if they met the following inclusion criteria: (a) Syrian refugee; (b) aged 10 to 14 years; (c) resided with a related caregiver who could provide legal consent; and (d) scored ≥15 on the Paediatric Symptom Scale (PSC-17) [15] (see Table A in S1 Appendix). The PSC-17 is a 17-item questionnaire that assesses psychological distress in children, with a range of 0 to 34; a cutoff ≥15 has been shown to indicate psychological distress [15]. Exclusion criteria were: (a) unaccompanied minor; (b) minors with an unrelated caregiver; (c) significant developmental, cognitive, or neurological impairments as determined by 4 items from an adapted version of the 10 Questions instrument [16]; or (d) imminent risk of suicide. One child, selected by the caregiver, was recruited per household in order to minimise the burden on the family and reduce likelihood of contamination of the intervention between participants. Any potential participants who met the exclusion criteria were referred to specialised services within IFH or another organization according to Inter-Agency Standard Operating Procedures. Informed consent was obtained from caregivers and assent from the adolescents in 2 stages to participate in (a) the screening; and (b) the EASE trial; participation required written informed consent, except oral consent was accepted for illiterate participants.

Randomisation and masking

Randomisation was conducted at the University of New South Wales (Australia) by staff who were independent of the trial using computerised software that generated random number sequences. Eligible adolescents were randomised without stratification allocated to either the EASE program or EUC (on a 1:1.6 ratio). Assessors who identified and enrolled participants into the trial and who conducted all assessments were masked to treatment condition allocation. Assessors were managed separately from the rest of the research team and at no point interacted with EASE or EUC facilitators. At the commencement of each postintervention assessment, assessors instructed participants to not inform them about their allocated condition. Further, at each assessment, assessors were instructed to guess which treatment arm the person was assigned to.

Before the trial was conducted, a comprehensive translation and cultural adaptation process of EASE was undertaken for the context of Syrian adolescent refugees in Jordan. Initial adaptation workshops were conducted with Syrian adolescents in Lebanon to agree upon necessary amendments [17], and subsequent adaptation workshops were conducted in Jordan with community representatives and health providers to determine the need for local adaptations. A pilot trial was then conducted to determine feasibility and acceptability of the EASE intervention [18]. All assessment instruments and the EASE protocol were adapted to be acceptable for cultural appropriateness, language, metaphors, and context.

As prescribed by the EASE protocol [12], the intervention comprised 7 weekly 1.5-hour group sessions for adolescents (8 to 10 people per group). In accordance with the findings of the adaptation work, all groups were gender specific and the groups comprised adolescents of any age from 10 to 14 years of age. Session 1 established group rules, provided psychoeducation about the effects of stressful events, and provided strategies on how to identify emotions. Session 2 taught strategies on reducing arousal and focused on relaxation breathing as a core stress management strategy. Sessions 3 and 4 focused on behavioural activation. Sessions 5 and 6 taught problem solving strategies, which included seeking social support as a means of dealing with problems. Session 7 focused on relapse prevention to manage future stressors. One caregiver of each adolescent was invited to three 2-hour group sessions (8 to 10 people per group) that occurred at 2 weekly intervals concurrently with the adolescent sessions. The goal of caregiver sessions was to promote supportive parenting to assist adolescents to manage stress. In session 1, caregivers were provided with psychoeducation and skills to help their child cope when they are distressed. Session 2 promoted positive parenting skills, including increasing praise and reducing physical discipline and promoting communication and relationship skills. Session 3 focused on skills for the caregiver to manage their own stress, including advice about sleep, nutrition, stress reduction exercises, and utilisation of social support. In addition, each session included informing caregivers about the strategies being taught to the adolescents in their sessions (see Table B in S1 Appendix). Sessions were conducted in various locations in Amman to facilitate attendance by participants who received reimbursement for travel costs. A Jordanian supervisor with >15 years in psychosocial programs and who participated in a training of trainer course on the EASE intervention attended 19% of sessions and used a standardised checklist to assess treatment fidelity and facilitator competency (rated on a 4-point scale: 0 = not satisfactory, 3 = very good).

Each group was conducted by 2 facilitators. The EASE facilitators were from a variety of professional backgrounds and recruited from the community by the implementing partner (IFH) in Jordan specifically to be trained in EASE. Facilitators either had experience implementing psychosocial services and/or tertiary qualification in a variety of disciplines. No facilitators had specialist mental health qualifications, training, or experience. Facilitators received 8 days of training on the EASE protocol, as well as group facilitation skills. Classroom-based competency was assessed through standardised role plays using an adaptation of the ENhancing Assessment of Common Therapeutic (ENACT) framework to assess counsellor competency [19]. Only those who demonstrated competency were permitted to function as facilitators. Facilitators were required to complete 2 closely supervised practice cycles following training. Twenty-three facilitators participated in trainings prior to the start of the trial. Following the completion of the training, all 23 (100%) facilitators demonstrated competency in intervention strategies, basic counselling skills, and group management, and of these, 20 facilitators continued with the current study and conducted EASE groups. Training and weekly supervision was conducted by a local supervisor (MG) who had attended an EASE training of trainer course and was supported biweekly by an experienced lead supervisor (AM) who was involved in developing of EASE protocol.

Participants randomised to EUC received a single 30-minute family session conducted in the participant’s home by a community health worker. This session provided feedback about the adolescent’s responses on the assessment battery, instruction of simple coping strategies, and provided a list of local services that the family could attend for psychosocial support for the adolescent. This was considered enhanced care as there are typically no mental health services accessible for young adolescents in Amman. The EUC facilitators followed a standardised script, and both caregiver and child were invited to attend. Sessions were limited to 30 minutes and provided the family with information on psychological problems, self-care strategies, and local referral options.

Outcomes

Secondary outcomes

The following secondary outcome measures were completed by adolescents. The Patient Health Questionnaire, adolescent version (PHQ-A) assessed symptoms of depression [21]. The PHQ-A is a 9-item symptom checklist corresponding to symptoms of depression experienced in the past week and is scored on a 4-point scale (0 = not at all, 3 = nearly every day), with a score range of 0 to 36 and higher total scores reflecting more severe symptoms of depression; this scale has been validated in refugees in Jordan [22]. The Children’s Revised Impact of Events Scale (CRIES-13) assessed levels of posttraumatic stress [23]. The self-report scale consists of 13 items measuring intrusive memories, avoidance, and arousal. Items are scored on a 4-point scale (0 = not at all, 1 = rarely, 3 = sometimes, 5 = often), with a score range of 0 to 65 and higher total scores reflecting more severe symptoms of posttraumatic stress; this scale has sound psychometric properties with Middle Eastern youth [24]. Wellbeing was measured using the self-reported Warwick Edinburgh Mental Wellbeing Scale (WEMWBS), which consists of 14 items exploring the child’s thoughts and feelings in the past week [25] and has been previously used in Jordan with undergraduate students but not validated to date with Jordanian young adolescents [26]. Scores are recorded on a 5-point scale (1 = none of the time, 5 = all of the time), with a score range of 0 to 50 and higher scores indicating greater well-being. Adolescents attending school completed the Psychological Sense of School Membership (PSSM) scale to measure their sense of belonging and psychological engagement in school [27]. This self-report scale comprises 18 items that are measured on a 5-point scale (1 = not at all true, 5 = completely true), with a score range of 0 to 90 and higher scores indicating a greater sense of belonging. This scale has not been validated in an Arabic context. We assessed daily functioning with a scale developed for the EASE trial that followed methods prescribed for cross-cultural assessment ([28]; see Table C in S1 Appendix). Adolescents rated 9-items identified to be representative of their daily activities on a 4-point scale (1 = none, 4 = often), with total higher scores suggesting greater levels of impairment of daily functioning. This scale was purpose built for the EASE trial and lacks psychometric properties.

In addition, caregivers completed the caregiver version of the PSC-35 to assess their perceptions of the psychological distress of their child. This scale is identical to the child version. The Kessler Distress Scale (K6), which has been used as a measure of distress in Arabic refugees [29], was administered to measure caregivers’ level of psychological distress [30]. The K6 is a 6-item self-report scale of distress measuring symptoms experienced in the past week; each item is scored on a 5-point scale (1 = all of the time, 5 = none of the time), with a total score range of 6 to 30 and higher scores indicating more distress. Validation studies indicate that scores ≥13 indicate probable anxiety or mood disorder [31]. Parenting behaviours were assessed using the Alabama Parenting Questionnaire (APQ), which measures 5 major parenting constructs: (a) parental involvement (10 items); (b) poor supervision and monitoring (10 items); (c) positive parenting (6 items); (d) inconsistent discipline (6 items); and (e) corporal punishment (3 items) [32]. Each item is scored on a 5-point scale, with subscale score ranges from 0 to 15 to 0 to 50 and higher scores indicating greater strength of the relevant subscale. The Arabic version of this scale has robust psychometric properties [33]. To measure adolescents’ exposure to potentially traumatic events, caregivers completed a 26-item traumatic events checklist. The items were determined in collaboration with local mental health professionals to develop a trauma checklist; items were scored dichotomously.

Assessments were conducted by Arabic speaking Jordanian assessors, who received 3 days of training in research ethics, the assessment battery, general interviewing techniques, and Psychological First Aid to support them to manage any distress in participants. Assessors additionally received 2 days of practical training in conducting assessments. Assessments were conducted on portable tablets to ensure that data could be reliably collected and uploaded. As a proportion of participants were inadequately literate, assessors administered the questions and entered responses on the tablets.

Statistical analyses

The required sample size for the trial was determined through a power calculation for individually randomised group treatment trials [34]. The estimated sample size was based on the estimated sample size required in the control arm, considering the number of EASE groups, the number of participants in each group with data at the 3-month follow-up time point, the effect size, the ratio of variance in the EASE arm versus the control arm, and the intracluster correlation coefficient [34]. In the absence of prior trials evaluating EASE, an estimate for the variance parameters was made on the basis of small datasets acquired in pilot studies in Jordan and Lebanon. This resulted in a conservative estimate of theta of 1.1 (the variance of the difference between EASE and EUC for the primary outcome) with an intraclass correlation of 0.13 (based on adolescents randomised to EASE groups). We calculated that with 20 EASE groups of 6 people each at the 3-month primary outcome time point (and assuming a 5% 2-tailed significance test and 80% power), data from 191 participants in the control arm would be needed at the 3-month follow-up to detect an effect size of 0.4 (based on the expected mean difference divided by the pooled standard deviation). This would correspond to an overall sample size of 311 at 3 months and an allocation ratio of EASE to EUC arms of 1:1.6. This ratio of allocation to EASE and EUC accommodated the effects of groups involved in EASE relative to the individually administered EUC. With a projected attrition rate of 33% loss to follow-up (based on previous scalable interventions using a 3-month follow-up [10]), we estimated the sample size required at enrolment would be approximately 470.

Baseline characteristics were compared for participants who were and were not retained at follow-up by planned comparisons with a Bonferroni adjustment to accommodate multiple comparisons (alpha = 0.002). The major analyses focused primarily on intention-to-treat. We employed linear mixed models to study differential effects of the treatment arms as this method allows the number of observations to vary between participants and thereby handles missing data by calculating estimates of trajectories using maximum likelihood estimation [35]. Fixed (intervention, time of assessment) effects and their interactions were included in the unstructured models, which provided an index of the relative effects of the treatments; time of assessment included baseline, posttreatment, and 3-month follow-up. Fixed effects parameters were tested with the Wald test (t test, p < 0.05, 2-sided) and 95% confidence intervals. Primary analyses focused on the subscales of the PSC-35 as EASE was designed with the intention to target internalising problems. The secondary outcomes (PHQ total scores, PHQ-A, CRIES-13, WEMWBS, PSSM, K6, APQ, PSC caregiver report) were analysed with the same analytic approach. Missing data was assumed to be random on the basis that participants completing the 3-month assessment and those who were missing did not differ in terms of any demographic or outcome measures at baseline. The following prespecified analyses were also conducted. We assessed the robustness of this statistical approach by conducting sensitivity analyses with only participants who completed the 3-month follow-up. To evaluate if the findings were influenced by exposure to prior traumatic experiences, analyses were repeated adjusting the models using caregivers’ baseline scores of the traumatic events checklist as a covariate. To determine the efficacy of the intervention on participants with more severe psychological disorders, we also conducted secondary analyses focusing on adolescents who scored ≥5 on the PSC internalising subscale.

To understand the potential role of EASE on parenting and adolescents’ mental health, a mediation analysis was conducted by assessing the direct and indirect effects of the intervention arm on change in the APQ subscale scores from baseline to 3-month follow-up and change in the PSC-35 subscale scores. The proposed mediation models were examined with 5,000 bootstrapped samples. These models examined the relationship between intervention arm on changes in child mental health (PSC subscale scores), with changes in parenting behaviours (APQ subscale scores) as the mediators. All analyses were overseen by an independent statistician who was blind to treatment condition.