6. Assessment of risk of bias of quantitative studies using the Cochrane 'Risk of bias' tool (RoB 1).

Study:De Jong 2019
Biasa	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Cluster‐randomised trial. Lack of information about randomisation
Allocation concealment (selection bias)	Unclear risk	As above, lack of information about randomisation and concealment
Blinding of participants and personnel (performance bias)	High risk	Due to nature of intervention, participants were not blinded. Control group did not receive any attention control intervention
Blinding of outcome assessment (detection bias)	Low risk	"Twelve trained assessors, who were blind to the group participants were assigned to, administered the instruments."
Incomplete outcome data (attrition bias)	High risk	Authors state: "Seventy‐one (34.5%) people in the program group did not attend the PFA training, mainly due to practical reasons including heavy rainfall in Sierra Leone during the days of the trainings. In the control group, 4 (2.0%) people received PFA when they should not have received it. We performed both completers and intention‐to‐treat analysis, but we judged the completers analysis as the main outcome analysis since we considered it most relevant to examine the training effects of PFA training in individuals who were actually trained. In addition, we considered attrition bias unlikely since the reason for most people not having received the condition they were assigned to (PFA or control) was external (extreme weather conditions)." Although the study authors conclude that attrition bias was unlikely; there was a high proportion of dropouts, and dropouts were likely to involve whole clusters (or certainly some clusters were likely to be more affected than others) due to the impact of the weather. Geographical factors were likely to have affected the dropouts, and consequently the demographics of the dropouts and the completers could vary substantially. Furthermore the completers' analyses were all different from the intention‐to‐treat analyses, with the direction of difference the same. We therefore judged this to be high risk for attrition bias.
Selective reporting (reporting bias)	Unclear risk	No evidence of pre‐registered protocol, so not possible to know if there has been selective reporting. For example, only certain subscales of the Psychological Quality of Life scale are reported; it is unclear whether the decision to only collect these data was pre‐planned.
PFA: psychological first aid

^aAssessed using Cochrane 'Risk of bias' tool for randomised trials (RoB 1) (Higgins 2017).