Please note that in the abstract we state that we follow the PRISMA guidance for systematic reviews; we would like to clarify some details. Firstly, we note that although our protocol was not published it was formally lodged in advance of the study with the University of Bath. Secondly, we did pre-specify all sub-group analyses in the protocol. We also note that we did not publish the full details of the search strategy for one specific database as per PRISMA guidance.

Unrelated to the PRISMA checklist, we stated that Olatunji et al (2014) did not employ a systematic search strategy. To clarify, that study employed a search strategy on two databases. There were a number of risks of bias concerning various areas of Olatunji et al's method, including identification and selection of studies (e.g. a small number of databases searched, no reported inter-rater checking of the selection process) and data collection and study appraisal (e.g. there was no reported checking of the data extraction procedure and there were no reports of an assessment of the methodological quality of the included studies, for example by using a recognised quality assessment tool). Based upon these methodological limitations it was our judgment that a review adopting a more rigorous methodological approach was warranted.