This review addressed a clear research question, with potentially reproducible inclusion criteria. The search included appropriate electronic databases, but was restricted to English. No apparent attempts were made to review reference lists of retrieved papers or retrieve unpublished studies. This meant that language and publication biases could not be ruled out. To minimise bias and errors during the review process, two reviewers selected studies and assessed the quality of the included trials, but it was unclear how many reviewers extracted data.
Study quality was assessed using standard criteria; the overall level of risk of bias in each study was reported. However, conclusions about how the overall risk of bias was classified were not reported. The included studies were considered to be generally poor quality. Data were pooled in a narrative synthesis, which was appropriate given the clinical heterogeneity.
Given the potential for missed studies, limitations of the review process, and methodological weaknesses of the included studies, the authors' conclusions that their results should be treated with caution is justifiable.