The review question and supporting inclusion criteria were broadly stated. The literature search was limited to two databases and by search dates. The search did not appear to include attempts to locate unpublished data, which meant that potentially relevant data may have been missed. Formal assessment suggested there was some evidence of publication bias. Each stage of the review process was performed in duplicate, which reduced the potential for reviewer error and bias.
Study quality was assessed using appropriate criteria and the results were fully reported in the review. The authors suggested that the included studies were of high quality, but the poor ability of the reference standards to correctly classify target conditions questioned the reliability of the findings.
The studies were pooled using appropriate meta-analysis methods, but the patient and study characteristics reported indicated both patient and methodological differences across the studies.
The number of studies used to calculate diagnostic statistics for different reference standards and cut-off values was unclear. For example, the pooled estimates at a cut-off value of 4 was reported to have been calculated from 64 studies (table 2, column N1a). It was not clear which studies this figure referred to as only 42 studies were eligible for inclusion or whether some of the 42 included studies had been double counted.
This review included a large number of studies and patients. However, given the lack of a consistent and reliable reference standard, the authors' conclusions should be considered carefully as the accuracy may be overestimated.