Assessing the Actual Treatment Benefit with Non-Adherence to Study Drug in a Large Randomized Trial

• Activity Number: 547
• Type: Contributed
• Date/Time: Wednesday, August 7, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Lulu R Sterling*+ and Yumi Kubo
• Companies: Amgen Inc. and Amgen Inc.
• Keywords: IPCW ; IPE ; cross-over ; ITT

Abstract

The ITT analysis is the gold standard in assessing the effectiveness of a study drug in randomized clinical trials. However, when non-adherence to study drug is substantial, the actual treatment effect may be underestimated. The EVOLVE trial (Evaluation of Cinacalcet Therapy to Lower Cardiovascular Events), the largest randomized trial ever conducted in the dialysis population, was marked by large proportion of patients withdrawing from randomized treatment but not to follow up(67% cinacalcet and 70% placebo). Various methods were used to assess the impact of non-adherence on the estimated treatment effect. The HR (95% CI) for the primary endpoint from ITT analysis was 0.93 (0.85, 1.02). Using lag censoring, the HR (95% CI) was 0.85 (0.76, 0.95). Using the IPCW method, the HR (95% CI) was 0.77 (0.66, 0.88). Accelerated failure time models with iterative parameter estimation were also used [HR (95% CI): 0.87 (0.75, 1.01)]. The best method to adjust for non-adherence remains debatable. The sensitivity analyses performed, while not providing definitive evidence that cinacalcet is more effective, suggest that the effect size is larger than that estimated by the ITT analysis.