Assessment of study quality
The majority of studies were judged to be ‘unclear’ across all domains of bias except ‘random sequence generation and allocation concealment’ which had equal ‘unclear’ and ‘low-risk’ judgements (Table S2).