The review question was clear and inclusion criteria were specified adequately to enable replication. The search strategy was limited to one database and had no details of search dates. A limited search, together with language restrictions, may mean that relevant studies were missed. Steps were taken to minimise reviewer error and bias at the study selection stage; it was unclear whether this was applied to the data extraction phase.
All the included studies were randomised controlled trials, which tend to be more reliable than other study designs, but the absence of any formal quality assessment meant that reliability could not be confirmed. Although only limited trial details were provided, the methods of synthesis seemed appropriate and helpful sub-group analyses were presented. Most included trials were small; the authors drew attention to one large trial that contributed disproportionately to the findings.
Given the limited search, inability to judge the quality of the included trials, and results dominated by one large trial, the reliability of the authors' conclusion is unclear.