The review had clear inclusion criteria for participants, intervention, study design and outcomes. Outcomes had to be evaluated using a particular scale, which meant that relevant studies using other scales would have been excluded. The authors searched a number of relevant sources, but limiting the search to English language studies meant that relevant studies could have been missed (language bias). Unpublished studies were not sought, but publication bias was assessed using standard methods. Two reviewers independently selected studies for the review, but it was unclear whether similar methods to reduce errors and bias were used in data extraction.
The authors did not assess validity of the included trials, which meant that the reliability of the included trials and the synthesis based on them was uncertain. Trials were combined in a meta-analysis; statistical heterogeneity was assessed and differences between trials were investigated using subgroup analyses.
The authors' conclusions reflected the evidence presented, but the lack of validity assessment and small numbers of trials and participants (particularly in the subgroup analyses), suggest that the conclusions should be treated with caution. A high proportion of patients in the included trials had comorbid conditions, such as hypogonadism or HIV infection, and the conclusions are unlikely to be applicable to the general population of patients with depression.