Lori Dodd, Ph.D., NIH/NCI
December 1, 2009 @ 3:30 pm - 4:30 pmLocation: BRB - 251
Biostatistics
Title: On the appropriate use of blinded independent central review in phase III oncology trials
Randomized clinical trials are the gold standard for shaping clinical practice by providing definitive evaluation of treatment efficacy. In oncology, overall survival (OS) is the most relevant end-point to measure patient benefit. In some settings where use of OS is unfeasible, progression-free survival (PFS) is used to assess treatment benefit. PFS evaluation, however, is not entirely objective, and considerable variability in its assessment exists. The potential for biased endpoint evaluation is a concern, particularly when PFS is a primary endpoint in an unblinded phase III trial. While the concern about bias and variability in assessments motivates the use of blinded independent central review (BICR), there is current debate about whether BICR adds much to the evaluation of a treatment's efficacy. An additional complication about BICR, is that the process may introduce its own source of bias from informative censoring, which may be difficult to overcome. We argue that BICR should not be the basis for a primary endpoint. Rather, it should be used as an auditing tool to confirm that the observed treatment effect cannot be explained by bias. In this talk, I discuss a strategy to evaluate the use of BICR on a subset of patients for auditing purposes. The method is based on an auxiliary variable model for time-to-event data. We present results from applying this methodology to a large phase III trial and from simulation studies.
