PLOS Medicine publish an up-to-date, cross-sectional study investigating the epidemiological and reporting characteristics of systematic reviews (SRs) from Page and colleagues and include Cochrane Reviews. A random sample of SRs published in February 2014 resulted in 300 included systematic reviews of which 15% (N=45) were Cochrane Reviews and 85% (N=255) non-Cochrane. This represented a threefold increase of published SRs from 2004 to 2014. Review authors compared method categories (eligibility, publication status, language, risk of bias and outcome, COI, funding source etc.) across both Cochrane and non-Cochrane SRs. The SRs included therapeutic/ treatment, diagnostic, prognostic and aetiology studies.
Page and colleagues found 40% of SRs to have misleading conclusions. They did not conduct bias assessments on the included reviews but noted poorly conducted and reported risk of bias. Poor reporting remains an issue even when authors report using PRISMA. Review authors propose developing a reporting tool based around PRISMA standards. RevMan integrates Cochrane’s own reporting standards (update and relaunch of all standards shortly). Thirty percent of non-Cochrane reviews report using Cochrane methods and cite the Cochrane Handbook. Rarely were unpublished data sought.
In conclusion, the increased SRs produced do not improve overall quality and suggests wasteful research.