The review question and inclusion criteria were clear. Several bibliographic sources were searched. Only studies published in English were reported, so some relevant studies may have been missed. The authors reported appropriate measures to minimise error and bias for the quality assessment of the studies, but not at study selection and data extraction stages of the review.
Results of the quality assessment were reported. Most included studies were found to be appropriately randomised. Risk of performance bias was not assessed. However, blinding of participants and study personnel appeared unlikely given the nature of the interventions, so the risk of performance bias could not be ruled out. A relatively large number of trials were included. A wide range interventions and outcomes were assessed, making the appropriateness of pooling uncertain. Statistical heterogeneity was assessed but specific results were not reported.
Pooled effect estimates appeared relatively small, particularly for the outcome relating to caregiver's reactions. This, in addition to the potential risk of performance bias and the choice of synthesis methods, means that the reliability of the conclusions is unclear.