The review addressed a clear question, and was supported by reproducible eligibility criteria. Several relevant databases were searched for studies in any language, but some relevant studies may have been missed since unpublished studies were not sought. No reason was given for seeking studies published after 2008. Duplicate processes were used to reduce the risk of reviewer error and bias, except they were not reported for the assessment of study quality.
Study quality was evaluated using the QUADAS, but very few results were reported, meaning it was not possible to evaluate the likely reliability of each study's results. The results were pooled in meta-analyses, but separate pooling of sensitivity and specificity is not generally recommended, and the results from the summary receiver operating characteristic curves may be more reliable. Not all the methods were clearly reported. Possible sources of variation were discussed.
The review had several reporting limitations, but its cautious conclusions are likely to be appropriate.