Finding the evidence gaps in Acute Respiratory Infections: an analysis of systematic reviews and RCTs

Article type
Year
Authors
Clark J1, Alloo J2, Carter M3, Thorning S4, Vallath S2, Del Mar C1
1Cochrane Acute Respiratory Infections Group, Australia
2School of Medicine, Bond University, Australia
3Centre for Research in Evidence-Based Medicine, Bond University, Australia
4Library, Gold Coast University Hospital , Australia
Abstract
Background: Prioritising which reviews to accept and support is now a major requirement of Cochrane Groups. One important consideration (among many) is where the gaps in evidence are between what is published and systematically reviewed. To address this is at the Cochrane Acute Respiratory Infections Group, we attempted to describe any gaps.

Objectives: To identify gaps in Cochrane Reviews in one Cochrane Group by comparing randomized controlled trials (RCTs) held in the Group’s Specialised Register of Trials against those synthesised in Cochrane Reviews.

Methods: We generated a list of RCTs from the register, and similarly another list of Cochrane Reviews. For each item (RCT or Cochrane Review) listed, we derived the main disease and the main treatment (intervention). We then used graphic software to generate a single line that linked each intervention to its disease. This enabled us to compare both Cochrane Reviews and RCTs to highlight discordant areas of investigation.

Results: We screened 5329 RCTs; 141 were excluded, leaving 5188. We screened 162 Cochrane Reviews; five were excluded (withdrawn, or inaccessible), leaving 157 reviews.
Several areas exist in which RCTs have been performed, but there are no Cochrane Reviews: overall this was 3115/5188 (60%). These areas include vaccinations (e.g. there are 143 RCTs investigating vaccinations against Pneumococcus, but no Cochrane Reviews on it). This contrasts with vaccines for the common cold (a Cochrane Review but only five RCTs), and vaccines for acute bronchitis (a Cochrane Review but only two RCTs).

Discussion: Limitations of the method include the sometimes arbitrary decision about which is the main intervention and main disease in any RCT or review, and also in collapsing groups of similar diseases and interventions together. The classification stage of the analysis was lengthy and tedious. Nevertheless it provided us with an excellent assessment of review gaps in our Group.