Risk of bias for each study based on the Cochrane risk of bias tool
| Vertommen et al 53 | Stafford et al 55 | Stafford et al 54 | McPherson et al 58 | McHugh and Howard et al 57 | Haegele and Zhu63 | de Schipper et al 51 | Dane-Staples et al 56 | |
| Selective reporting (reporting bias) |
|
|
|
|
|
|
|
|
| Incomplete outcome data (attrition bias) |
|
|
|
|
|
|
|
|
| Allocation concealment (selection bias) |
|
|
|
|
|
|
|
|
| Random sequence generation (selection bias) |
|
|
|
|
|
|
|
|
| Blinding of outcome assessment (detection bias) |
|
|
|
|
|
|
|
|
| Blinding of participants and personnel (performance bias) |
|
|
|
|
|
|
|
|
| Other bias |
|
|
|
|
|
|
|
|
The Cochrane r isk of b ias t ool was used to assess the possibility of bias across individual studies. Studies were categorised as having ‘ high ’ , ‘ unclear ’ and ‘ low ’ risk of bias across five domains. High risk of bias is represented by a black dot, unclear risk of bias by a grey dot and low risk of bias by a white dot. The five domains of bias are ‘Selection bias’, which included random sequence generation and allocation concealment; ‘performance bias’, which included blinding of participants and personnel; ‘detection bias’, which included blinding of outcome assessment; ‘attrition bias’, which included incomplete outcome data; ‘reporting bias’, which included selective reporting; ‘other bias’ sources were also addressed. Studies with predominantly ‘high’ and ‘unclear’ risk of bias (black and grey dots) were considered lower quality when compared with studies with predominantly ‘low’ risk of bias (white dots) across domains.