The review question was defined in terms of study design, intervention, comparator, participants and outcomes. Although unpublished studies were eligible for inclusion, no attempts to uncover unpublished studies were reported and publication bias was suggested. Searching for studies in all languages reduced the possibility of language bias. The review process was performed in duplicate, reducing the possibility of reviewer error and bias.
The quality of the studies was assessed, but it was unclear how this was performed and how this was incorporated into the analysis. The results of the validity assessment were not reported. The meta-analyses were associated with high levels of heterogeneity, so the results of these may not be reliable.
In light of high levels of heterogeneity, the possibility of publication bias and the unknown quality of included trials, the authors' conclusions should be viewed with caution.