This review assessed a well-defined review question and searched a number of relevant sources. Both published and unpublished data were sought. Although this suggested that the risk of publication bias would be low, the authors acknowledged there may be some risk of bias and that their tests for publication bias were unlikely to be reliable given the small number of included trials. The authors made some attempt to reduce the risk of reviewer error and bias when extracting the study data, but it was unclear whether similar precautions were taken when selecting the studies for inclusion in the review. Also, the authors did not report how many reviewers assessed the quality of the trials.
Although appropriate criteria appear to have been used for quality assessment, the findings were not reported. Therefore, the reliability of the data was unclear. However, the methods used to synthesis the data, including the time to event data such as survival, appeared to be appropriate.
Overall, the authors' conclusions appeared to reflect the evidence available, but the paucity of trials and the risk of missing data should be considered when interpreting the review conclusions.