Application of the Composite Quality Score (CQS-2) in systematic reviews of prospective, controlled, clinical therapy trials – a pilot study (Protocol)

Objectives: To establish, whether clinical conclusions from systematic reviews that applied the second version of Cochrane’s Risk of Bias tool (RoB 2) remain the same when the Composite Quality Score (CQS-2) is used, instead. Methods: PubMed will be searched for one systematic review that complies with the following criteria: At least 20 prospective, clinical, controlled therapy trials included into meta-analyses; computable data for test- and control group reported; trial appraisal using the RoB 2 tool; inclusion of at least 5 trials in at least one single comparison per measured outcome; published in English. All trial reports included in the systematic review will be traced in full copy and the following data extracted per trial: Full reference details; overall RoB 2 appraisal decision; computable datasets. All trial reports will be re-appraised using the CQS-2. All datasets will be stratified by overall bias risk (RoB 2 tool) and per corroboration (C-) level (CQS-2). Trial data for each bias risk and C-level will be statistically pooled using Mantel-Haenszel statistics with random-effects model. All pooled ‘low bias risk’ effect estimates (RoB 2) and all pooled effect estimates of the highest C-level with data (CQS-2) will in turn each be pooled by use of a random effects meta-analysis. DerSimonian and Laird’s method of moments estimator will be used to estimate the between-study variance. Statistical inconsistency will be quantified using the I2 statistic. The pooled effect estimates of both appraisal tools will be statistically compared using the Wald-test and the null-hypothesis tested that both are not significantly different. Clinical conclusions based on the pooled ‘low bias risk’ estimates (RoB 2) and the C4-level (CQS-2) will be qualitatively compared. Reporting: The final report will be made available online as preprint in one of the major preprint repositories and submitted to a peer-reviewed journal.