“…A low risk was found in the two recent studies 42,50 whereas five older studies 31,32,34,38,40 were deemed to have high risk because outcomes were not assessed randomly (5/30, 17%) 31,32,34,38,40 or the outcome assessor was not blinded (2/30, 7%) 31,40 . The risk of attrition bias was high in 18 studies (18/30, 60%), 27–30,35,37,38,40,43,45–49,51,53,55,56 low in two studies (2/30, 7%) 33,36 and unclear in the others (10/30, 33%) 31,32,34,39,41,42,44,50,52,54 . In 11/30 (37%) articles, outcomes were not reported selectively, reducing the risk of reporting bias 31,32,34,36,39,41,42,44,50,52,54 .…”