Fig. 1. Flow diagram for identifying relevant studies. In total, 263 studies were identified by searching 5 databases and manually searching relevant bibliographies as follows: 37 studies from MEDLINE, 18 from EMBASE, 30 from Cochrane Library, 33 from Web of Science, 144 from SCOPUS, and 1 by manual searching. We excluded 39 duplicate studies, plus an additional 196 of the remaining 224 studies that did not satisfy the selection criteria. We reviewed the full texts of the remaining 28 studies, which resulted in a further 16 studies being excluded based on the selection criteria. The reasons for exclusion of these 16 studies were no outcome data (n=7), no or poor randomization (n=5), no control group (n=3), and duplication due to incorrect publication year (n=1). After reviewing the full texts, 12 studies were finally included in this study.