The process followed in the selection of studies and theresults obtained from the search are shown in Figure 1.Across all searched databases (and reference listsreviewed) 302 abstracts were identified as potentiallyrelevant. Of these, 117 studies were identified as dupli-cates. In the case of duplicate publications, the originalpaper (or the oldest version) was used. A further 173were excluded, in 2 phases, for various reasons (Figure 1).Finally, twelve studies [19-30] met the aforementionedcriteria and were included. Although foreign languagestudies were excluded from this review, all potentiallyeligible studies reported in languages other than Englishwere documented for future assessment [31,32].