Background: Adaptive designs are somewhat underused, despite prominence given to methodology in the statistical literature. Some concerns relates to robustness of adaptive designs in decision making, acceptability of trial findings to change practice, anxiety about early stopping of trials and worry about wrong decision making. These issues could be linked to inadequate reporting of the conduct of such clinical trials. We assess the reporting of group sequential randomised controlled trials (RCTs), which are one of the most well-understood adaptive designs in the confirmatory setting.
Methods: We undertake a systematic review searching Ovid MEDLINE from 1st January 2001 to 23rd September 2014 and including parallel group confirmatory group sequential RCTs that were prospectively designed using the Frequentist approach. Eligible trials are screened for completeness in reporting against the CONSORT 2010 checklist with some proposed modifications to capture issues such as statistical bias correction following early stopping. Descriptive statistics aided with forest plots on CONSORT compliance are presented.
Discussion: Reporting of the conduct of adaptive designs is an area which has not been fully explored. Hence, the findings from this study can enlighten us on the adequacy in reporting of well-understood group sequential RCTs as a class of adaptive designs and on ways to address some of the cited concerns. Most importantly, the study can inform policy makers on the adequacy of the current CONSORT statements in enhancing reporting of such adaptive designs.
CORE (COnnecting REpositories)
CORE (COnnecting REpositories)