However, the results from these subjects cannot be separated from those from other participants who met the inclusion criteria of this review, possibly influencing the homogeneity of participants. Thirdly, this study had a language bias because only articles in English were included, possibly distorting the results of this meta-analysis by failing to identify relevant findings published in languages other than English. Future studies might systematically search for relevant studies outside the English literature.Fourthly, not all existing studies may have been included in this meta-analysis with the search method used. This problem may be overcome by searching the grey literature (i.e. literature that has not been formally published). Fifthly, heterogeneity was found across studies (p