2.6 Risk of bias
Risk of bias assessment was conducted by the two review authors (PA & SM) independently using the Cochrane methodology. Disagreements were resolved by discussion or arbitration by a third author (AM). Randomised controlled trials were assessed with the Cochrane risk of bias tool (29) and observational studies were assessed using the Non-Randomised Studies of Interventions (30) (ROBINS-I). For ROBINS-I, the classification ‘no information’ was used when there was insufficient data reported within the text to permit an accurate judgement of bias (e.g. when only an abstract was available). Therefore, an overall judgment was not given when it occurred due to the lack of available information potentially allocating an inaccurate overall score.