When analyses were saturated, 1430 articles had been evaluated for inclusion in the meta-analysis. After exclusion due to criteria A-G, 346 studies, describing 502 control groups, remained. Due to lack of certain pieces of information in some of the articles, not all 502 control groups could be included in all four multiple regression analyses, as shown in the thick-boarded boxes to the right in the figure.
Ström et al. BMC Neuroscience 2013 14:41 doi:10.1186/1471-2202-14-41