The inclusion criteria were generally clear. The authors searched a range of relevant databases and made some limited attempts to locate unpublished trials. It appeared that there were no language restrictions, which minimised the risk of language bias. Publication bias was assessed but no results were reported, so the risk of this bias remained unclear. Appropriate methods were used to reduce risk of errors or bias affecting study selection, data extraction and validity assessment.
Quality was assessed using standard criteria and the results were used in the synthesis. Few details of participants in included trials were reported, which made it difficult to assess generalisability of the findings. Pooled estimates for dichotomous outcomes were derived using standard meta-analysis methods. Small numbers of participants and events in the trials may have limited the review's ability to detect differences for these outcomes. The decision not to pool studies for length of stay was appropriate given the reporting limitations of the included trials. Clinical heterogeneity of the included trials limited the conclusions that could be drawn about which components of enhanced recovery programmes were most effective.
The authors' conclusions and recommendations for further research reflect the limitations of the available evidence and seem appropriate.