3.2 Risk of bias assessment
Both randomised trials were assessed to have an overall low risk of bias. Performance and attrition bias were the two domains that the two trials had a common rating on low risk of bias. (31, 32) One trial had a high risk of bias for concealment of treatment allocation, as study sites became aware they were crossing over to the intervention phase four weeks prior. (31) The other RCT had a high risk of bias for method of randomisation, as the text did not state how participants were randomised. (32) Detection bias and reporting bias were judged as unclear due to a lack of information. The non-RCTs studies were not given an overall risk of bias as these observational studies were abstracts only and insufficient information was available to assess most domains. (27, 28) A summary of risk of bias for all studies is reported in Table 3.