The review addressed a clear question and criteria for study inclusion were stipulated. Appropriate databases were searched, but as the review was restricted to published studies, there was a risk of publication bias. Although the authors evaluated publication bias, the small number of included trials meant that these tests were unlikely to be reliable. Steps were taken during most parts of the review process to minimise errors and bias, but were not explicitly reported for the selection of studies.
The included trials were all small, evaluated a wide range of medications, and few results could be pooled. It was unclear what the comparators in the included trials were from the trial details; the reviewers appeared to have pooled the mean differences in treatment arms for each type of adjunctive medication. Even with pooling of trial results, the maximum number of results for one treatment arm for comparison was still small (27 patients). There were no attempts to evaluate statistical heterogeneity across the results. The degree to which the adjunctive medications conferred additional benefits to patients was not clear.
Although the authors' conclusions are based on the evidence presented, the limitations of the data and review methods, particularly the flaws in the analysis, mean that the findings of the review are unlikely to be reliable.