What next for Cochrane in systematic reviews of prognosis?

Article type
Authors
Beller E1, Glasziou P1, Del Mar C1
1Bond University, Australia
Abstract
Background: Cochrane Reviews are usually about interventions; and a few address diagnostic test accuracy. Although there are Cochrane Methods Group for prognosis, non-randomised studies, and bias in reviews, (and even one published exemplar review on prognosis) there is no established method for conducting reviews of these, in contrast to interventions, where Cochrane has led methods development. We wanted to quantify different types of systematic reviews (SRs) published outside the Cochrane Library.
Objectives: To quantify: 1) the proportion of non-intervention to intervention SRs; and 2) to quantify it over time; and 3) to find published methodologies for SRs of prognosis and association.
Methods: A validated search for SRs published in the 118 Core Clinical Journals (CCJs) in each calendar year from 2009 to 2013. SRs were defined as: 1) a documented search strategy; 2) clinical outcome; and 3) clear inclusion and exclusion criteria for studies. Reviews were then classified by aim: 1) intervention (e.g. meta-analyses of RCTs), 2) prognosis or association (e.g. meta-analyses of cohort studies), 3) frequency or prevalence (e.g. meta-analyses of cross-sectional studies), or 4) diagnostic test accuracy.
Results: The number of systematic reviews in the CCJ set has increased each year from 521 (2009) to 931 (2013; Figure). Only 53% of systematic reviews published in the CCJs are about interventions; 26% address prognosis or association, 15% prevalence or frequency, and 6% diagnostic test accuracy. Although the number of SRs is increasing, the proportion of SRs with each of the four aims is relatively constant from year to year. We found several papers addressing conduct of primary prognosis studies, and statistical methods that may be used in prognosis reviews, however we found no unified guidelines incorporating them.
Conclusions: Nearly half of the SRs published in the leading journals outside the Cochrane Library have aims other than evaluating interventions. One-quarter of the reviews are about prognosis or association. If Cochrane is to incorporate this type of SR, funding and time to address methodology of prognosis reviews is required.