The objectives and inclusion criteria of the review were clear in most respects but the restriction of the review to studies published from 1999 onward did not appear to have a clear clinical rationale. Relevant sources were searched for studies. The restriction to published studies meant that the review was at risk of publication bias. It was unclear whether the search was restricted by language; all selected studies were in English. Steps were taken to minimise bias during study selection, quality assessment and data extraction. Insufficient information was provided about the quality assessment criteria used and the quality of individual studies (such as selection procedures, follow-up rates) and this made it difficult to assess the reliability of study findings. As the authors noted, most studies neither stated how participants were selected nor described the intervention and there was marked heterogeneity in their choice of outcomes. Results varied markedly across studies. The authors' conclusions referred to comparison with historical controls, but the data for medical groups were not included in the results section of the review. There was no evidence that the surgical and medical groups were comparable at baseline.
The authors' conclusions should be interpreted very cautiously in view of the lack of controlled data, failure to describe study quality, the restricted search and differences between the studies.