Abstract:
|
It is a common practice in oncology study to let patients cross over to another treatment arm which may be benefit to patients. Crossover includes but not limited to, patients reached progressive disease (PD), AE, or don't reach to PD, but there is inadequate response for the current treatment based on the investigator discretion, etc. Due to the cross over, a comparison of the treatment effect cannot be made as randomized. This leads to statistical challenges in the analysis of overall survival and cost-effectiveness because crossover leads to information loss and dilution of comparative clinical efficacy. Conventional methods cannot fully adjust for the bias caused by crossover and clinical effect may be underestimated. In this talk, several statistical methods will be discussed for handling treatment arm cross over. It includes Iterative Parameter Estimation (IPE), Inverse Probability of Censoring Weight (IPCW), Rank-Preserving Structural Failure Time (RPSFT). A real study will be discussed for testing those methods.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.