“…97,100 These studies, as reported, therefore had a high risk of selection bias. Thirteen studies either clearly failed to conduct an intention-to-treat analysis or left it unclear whether or not they had done so, and so were categorised as not clearly conducting such an analysis; 42,53,55,77,79,80,88,91,102,114,117,122,124 six of these also failed to report adequate methods of randomisation and allocation concealment, 77,79,80,114,117,122 all but one of which were abstracts only. 114 Two further key criteria were assessed that had the greatest potential to confound the results of this report: the non-comparability of groups at baseline and non-identical care programmes post intervention.…”