The inclusion criteria were clear in terms of the interventions, participants, outcomes and study designs. Several relevant sources were searched and authors attempted to locate unpublished data, thereby reducing the risk of publication bias. However, the restriction to studies reported in English might have resulted in the loss of some relevant data. Methods were used to minimise reviewer error and bias in the assessment of validity and extraction of data, but it was not clear whether similar steps were taken at the study selection stage. Validity was assessed using established criteria and methodological limitations of the included studies were discussed in the text of the review. In view of the differences between the studies, a narrative synthesis was appropriate. Characteristics of the included studies were presented in the tables. In general, the review reported only whether a study showed positive or negative results; the raw data on which these statements were based were reported for few outcomes. The included studies were conducted in Australia (2 studies), England (2 studies) and Germany (1 study) and, given the variation in practices between countries, the results from these studies may not be generalisable to other national health systems. The included studies were generally poor quality, with small sample sizes and high attrition rates. The authors appropriately considered the methodological weaknesses of the included studies and differences across the studies. Consequently, their cautious conclusions appear appropriate based on the evidence presented.
Implications of the review for practice and research