The review question and inclusion criteria were clear. The authors searched a number of relevant databases and other sources and it appeared that no language restrictions were employed. This reduced the chances that relevant studies were omitted and that language and publication biases were introduced into the review. Assessment revealed that publication bias was present. The authors reported that they used methods designed to reduce reviewer bias and error in study selection, but not in the extraction of data, which included some limited appraisal of characteristics associated with study validity. There was considerable clinical heterogeneity between studies, although this was not accompanied by significant statistical heterogeneity.
Clinical heterogeneity, coupled with the probability of publication bias and the general low quality of the included studies means that the reliability of the authors' cautious conclusions is open to some doubt (which they acknowledged).