Evaluating Methods for Estimating a Treatment Effect During Treatment-Switching in Time-to-Event Clinical Trials

• Activity Number: 372
• Type: Contributed
• Date/Time: Tuesday, August 11, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Carl Dicasoli* and Martin Homering and Christian Kappeler and Harald Siedentop and Thomas Schmelter and Daniel Haverstock
• Companies: Bayer HealthCare Pharmaceuticals and Bayer HealthCare Pharmaceuticals and Bayer HealthCare Pharmaceuticals and Bayer HealthCare Pharmaceuticals and Bayer HealthCare Pharmaceuticals and Bayer HealthCare Pharmaceuticals
• Keywords: clinical trials ; treatment-switching ; proportional hazards ; accelerated failure time model ; drop-out ; intent-to-treat (ITT) analysis

Abstract

In cardiovascular trials, for a full ITT analysis, a situation may arise in which patients stop the investigational drug and switch to the active standard of care control. In most practical situations, the event hazard immediately after the switch is assumed to be similar to control. The ITT approach may not give a realistic treatment effect as it dilutes a potentially positive treatment effect under ITT analysis conditions. To better account for this situation, the Branson and Whitehead (2002) method will be explored which takes into account patients' switch times under an accelerated failure time (AFT) model framework, but assumes that the effect of the experimental treatment is the same at randomization in the test arm as those that switch. Moreover, this approach only examines drop-in, where control subjects start taking an investigational drug and does not examine the case of drop-out, where subjects stop taking the investigational drug. To assess drop-out, a comparison of the proportional hazards, AFT, Branson and Whitehead approaches will be conducted with both clinical trial data and a simulation study. Finally, more recent methods, e.g., Zeng et al (2012) will be studied.