“…We judged the risk of detection bias to be low in objective outcomes, and high in patient reported outcomes since lack of blinding can potentially introduce bias for this class of outcomes through multiple pathways (different expectations from the two groups and biased assessment of the effect) Higgins (13) . In seven studies, all the participants randomized were available for all follow-up duration with no drop outs or exclusions (15,16,18,19,20,21,25) . In four studies, the risk of attrition bias was high (14,17,22,24) .…”