Bias Correction in Subgroup Analysis with Survival Outcomes

• Activity Number: 473
• Type: Invited
• Date/Time: Wednesday, August 12, 2015 : 8:30 AM to 10:20 AM
• Sponsor: WNAR
• Author(s): Lu Tian* and LJ Wei and Fei Jiang
• Companies: Stanford University and Harvard University and Harvard University
• Keywords: bias correction ; subgroup analysis ; personalized medicine

Abstract

Subgroup analysis becomes a popular practice in analyzing clinical trial data for identifying subgroup of patients who may benefit from the treatment of interest. Main criticism of the subgroup analyses includes multiple testing and biased results caused by potential imbalance in baseline characteristics. In this talk, we will present a general method to correct the biases in estimating the treatment/subgroup interactions caused by imbalanced baseline covariates. The method does not depend on specific model assumption and can easily handle survival outcomes. We will also present ideas to improve the power of the subgroup analysis while controlling for the overall family-wise type one error. The proposal will be illustrated with real data examples.