Predict the Probability of Final Trial Success in Interim Analysis

• Activity Number: 423
• Type: Contributed
• Date/Time: Tuesday, August 2, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Rujun Teng*+ and Kaifeng Lu
• Companies: Merck & Co., Inc. and Forest Laboratories, Inc.
• Keywords: futility ; posterior probability ; predictive probability

Abstract

In clinical trials, enrolment of patients is a continual process staggered in time. In Phase II POC (proof of concept) trials, interim analyses are often performed to assess the available data, to find whether there is significant treatment difference, and to evaluate whether the difference is convincing enough to draw a conclusion. Since clinical trials are expensive, if superiority or futility conclusion can be reached at interim look earlier than completion of the whole trials, the cost will be less than planned. Based on the usual Bayesian framework, a posterior probability distribution for some parameter of interest (e.g. treatment effect) can be derived from the observed data at the interim analysis and a prior probability distribution for the parameter. For example, the conclusion treatment A is superior to treatment B is reached if the posterior probability is larger than some prespecified cutoff value. Alternatively, predictive probability can be used to calculate the conditional probability of success of final trial, given the available interim data. Simulation data will be used to compare the posterior probability and predictive probability approaches.