Inclusion criteria for the review were broadly defined. Several relevant data sources were searched. There was potential for language bias, as only articles in English were included. Publication bias was not assessed and could not be ruled out, but the extensive search should have limited the risk. Attempts were made to reduce reviewer error and bias throughout data extraction and quality assessment; it was unclear whether similar attempts were made for study selection. Quality assessment was undertaken using a standard checklist and indicated the variable quality of the included trials (acknowledged by the authors).
The authors noted that types of intervention, cohort characteristics, outcomes and outcome measurement tools varied across trials. Given such differences it seems appropriate that the authors chose to narratively synthesise the evidence.
The quality of the included evidence and potential differences across the trials limit the reliability of the estimates of effectiveness, but the authors’ conclusions are suitably equivocal and seem reasonable but broad.