Data collection and extraction
A data abstraction form was designed a priori with Excel (Microsoft Corp, Redmond Washington, USA) to standardise data collection. Two independent reviewers (E.B and T.P) scanned all abstracts identified by search cross-referencing. The full text was then identified for each study that potentially met the inclusion criteria. Two reviewers (E.B and T.P) independently reviewed the full-text eligibility. If no consensus could be reached by the two reviewers and after discussion, a third specialist author was consulted and made the decision (J.S and/or C.D). Articles that did not meet the inclusion criteria were excluded (Figure 1). Data extraction included general study information (see details in Table 1 and Supplemental Tables 2 and 3).