The research question was supported by well-defined inclusion criteria. Several relevant databases were searched without language restrictions. It appeared that conference abstracts (studies that were not fully published) were eligible for inclusion. Two reviewers were involved in study selection; it was unclear whether similar steps to reduce reviewer error and bias were employed in the data extraction process.
Trial quality was not systematically assessed, but only double blinded RCTs were included and all of these trials reported allocation concealment, suggesting that they were of reasonable quality. It did not appear that between-trial heterogeneity was assessed, but the forest plots suggested possible heterogeneity for pain scores. The meta-analyses generally included small numbers of trials with few participants; several outcomes had wide confidence intervals.
The authors' conclusions reflect the evidence presented, but the heterogeneity and the low numbers of small trials should be taken into consideration when interpreting their conclusions.