Risk of biasTable 2 shows the methodological quality assessment ofindividual studies. For one study the corresponding author wascontacted to resolve ambiguity in data. The initial agreement ofthe reviewers on the total risk of bias assessment was 88% (77of 88 items), and Cohen’s Kappa (95% CI) was 0.79 (0.73–0.85). All disagreements were resolved in a consensusmeeting. Three studies were assessed as having a high risk ofbias and five studies were assessed as having a low risk ofbias. The most prevalent limitations were found in items aboutblinding (patient, care provider, outcome assessor), allocationconcealment and compliance.