The records were divided between two reviewers who read the title and abstract of each record for an evaluation of relevance. To assess inter-rater agreement, approximately 5% of the records (k = 315) were selected randomly and rated independently by both reviewers as ‘potentially relevant’ or ‘not relevant’ using the above inclusion criteria. Inter-rater agreement for the title/abstract review was 91.8%. The full-text articles of eligible records rated ‘potentially relevant’ by at least one reviewer (k = 160) were obtained, reviewed and rated independently by the two reviewers. Inter-rater agreement for the full-text review was 88.7%. Any discrepancies between the reviewers' ratings were discussed and a consensus was reached.
