Berger 2009.

Methods	Cluster‐randomised trial of ERASE Stress Sri Lanka group therapy vs wait list
Participants	Included (n = 166) Children 9 to 14 years of age exposed to the 2004 tsunami Female: 73. Trauma exposure: physically hurt during the tsunami 139, knew someone close who had died during the tsunami 100, exposed to another major traumatic incident 148 Excluded Not reported Setting School in a small town that suffered a direct hit from the tsunami in Welligama Sri Lanka, 2006
Interventions	Group therapy (n = 84) Twelve weekly 90‐minute sessions of 12 to 16 participants (grouped by age and gender for the older 2 groups). Key components: CBT, psychoeducation, meditation, bioenergetic exercises, coping skills, narrative techniques including art therapy, planning for the future Wait list control (n = 82) Children in this group attended a religious class Therapists Teacher training was given to all 12 homeroom teachers over 3 days. Each teacher was responsible for 1 class only
Outcomes	Diagnosis Scale: score ≥ 3 on 1 item of the UCLA PTSD Index (17‐item) Rater: child Trauma symptoms Scale: UCLA PTSD Index Rater: child Depression Scale: Beck Depression Inventory (7‐item) Rater: child Functional problems Scale: Child Diagnostic Interview Schedule (7‐item) Rater: clinician When At 3‐month follow‐up
Notes	So they could be added to the meta‐analysis of function, functional impairment scores were converted to negative values
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	One class in each of the 6 age groups were randomised by coin toss to group therapy or wait list control
Allocation concealment (selection bias)	Unclear risk	Not reported
Blinding of participants (performance bias	High risk	Participants probably were aware of whether they were in the active or wait list group
Blinding of outcome assessment (detection bias) All outcomes	High risk	Although 'volunteers blinded to the experimental condition administered questionnaires', all measures were self reported
Incomplete outcome data (attrition bias) All outcomes	Low risk	No loss to follow‐up: "we had no dropouts among those who completed the questionnaires. There were no missing data"
Selective reporting (reporting bias)	High risk	Only diagnosis was reported
Other bias	Low risk	No other bias was apparent