Analysis of Overall Survival When Length of Treatment and Treatment Switching Are Time-Dependent Outcomes of Oncology Trials

• Activity Number: 256
• Type: Contributed
• Date/Time: Monday, August 2, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Jinwei Yuan*+
• Companies: ICON Clinical Research
• Keywords: survival analysis ; Cox regression ; oncology trials ; treatment switching

Abstract

Some oncology trials that target overall survival involve a design adaptation based on an intervening event, such as progressive disease (PD). After the event, patients stop receiving their assigned treatment and some control patients switch treatments ("crossover"). Therefore treatment received and the length of exposure to treatment become outcomes of the study as it unfolds. The impact of this change can be substantial. We propose two methods based on the Cox regression model with time-dependent covariates by using programming statements and the counting process style of input in SAS PHREG to deal with this problem. An application to a hypothetical data from a clinical trial of a solid tumor treatment is described. It is helpful for analysts to follow closely the assumptions and limitations of the various statistical approaches available.