Abstract 19575: Mortality Risk of Prolonged Dual Antiplatelet Therapy After Drug-Eluting Stent Implantation: A Bayesian Cross-Design Meta-Analysis of Completed and Prematurely Stopped Trials
Introduction: Meta-analyses of randomized controlled trials (RCTs) of dual antiplatelet therapy (DAPT) have produced varying conclusions about the relation between mortality and prolonged DAPT after drug-eluting stent (DES) implantation, but reasons for disagreement remain uncertain.
Hypothesis: RCTs achieving planned enrollments provide a less biased assessment of mortality rates than do prematurely stopped trials.
Methods: We used Bayesian cross-design methods to create models that incorporate parameters for completeness of trial enrollment and for trial outcome to determine the global relation between mortality and DAPT duration.
Results: A total of 12 RCTs (8 completed and 4 stopped trials) randomized 34,800 patients to short or prolonged DAPT after DES implantation (Fig. 1). Mortality rates in the 8 RCTs achieving planned enrollments were nominally lower after short than after prolonged DAPT (posterior median odds ratio [OR], 0.82; 95% Bayesian credible interval [BCI], 0.66-1.00), but mortality rates in 4 prematurely stopped RCTs were unrelated to DAPT duration (OR, 1.11; 95% BCI, 0.63-1.86). A global summary based on equal weighting of the evidence from RCTs with planned or incomplete enrollment showed no relation between mortality rates and DAPT duration (OR, 0.87; 95% BCI, 0.64-1.27), but a sensitivity analysis that weighted the RCTs by completeness of enrollment provided credible evidence for the alternative hypothesis that lower mortality rates were associated with shorter courses of DAPT.
Conclusions: Bayesian approaches, which contextualize mortality rates by the completeness of trial enrollment, help to explain the variable results of meta-analyses by demonstrating that RCTs achieving planned enrollments provide a less biased estimate of the mortality difference between shorter and prolonged DAPT than do prematurely stopped RCTs. Meta-analyses should differentially weight trial evidence by the completeness of trial enrollment.
Author Disclosures: J.A. Bittl: None. U. Baber: None. S.M. Bradley: None. D.N. Wijeysundera: None.
- © 2016 by American Heart Association, Inc.