The review question was broadly defined. The literature search was limited and restricted by language, which meant that potentially relevant data may have been missed. The authors did not state whether each stage of the review process was performed in duplicate, which meant that reviewer error and bias could not be ruled out.
Study quality was not formally assessed, but the authors did acknowledge that none were randomised controlled trial, most lacked comparison groups, and many were single case studies. The authors acknowledged that there was some overlap in study populations. The total number of included participants was unclear, but most studies appeared to include less than 15 patients. Study details were presented for 34 studies, so details on the remaining two studies were unknown. Given the differences between studies in type of intervention, intervention frequency, and outcome assessments, a narrative synthesis seemed appropriate. However, this was limited as the significance of the findings was generally unknown.
This was a poorly-reported review that was at risk of various forms of bias. The evidence base and synthesis were considerably limited, which suggested that definitive conclusions could not be drawn, therefore the authors’ conclusions cannot be considered reliable.