Inclusion criteria for participants, intervention and study design were clear. A wide range of comparators and outcomes were included. The authors searched a range of relevant sources, but some language restrictions were imposed and unpublished studies were not specifically sought, so the review could be at risk of language and publication bias. The authors did not formally assess validity of the included studies; some relevant information was presented in the text, but it was not clear whether this was used in the synthesis. Data were extracted by two reviewers, reducing the risk of errors and bias, but the methods used for study selection were not reported. Adequate details of included studies were presented. Studies were combined in a narrative synthesis, which seemed appropriate in view of the clinical heterogeneity of the included studies. However, the synthesis was brief and largely consisted of counting the number of studies reporting a statistically significant difference between treatments. The authors' conclusions were based mainly on differences reported in one or two small trials of uncertain quality and should, therefore, be treated with caution.