The research question was supported by inclusion criteria for participants, intervention, outcomes and study design. Two databases were searched for published studies, but abstracts and unpublished studies were excluded, so publication bias could not be ruled out. All languages were eligible for inclusion, which reduced the possibility of language bias. The review process was not reported, so it was not known whether steps were taken to reduce error and bias.
Study quality was assessed, but the quality assessment criteria and results were poorly reported. A narrative synthesis appeared appropriate given the heterogeneity in interventions and outcomes.
Some aspects of the review process were poorly reported and publication bias may be possible, but the authors' conclusions adequately reflect the paucity of evidence available and the limited conclusions that can be drawn.