Online Program Home
My Program

Abstract Details

Activity Number: 254
Type: Contributed
Date/Time: Monday, August 1, 2016 : 2:00 PM to 3:50 PM
Sponsor: Biopharmaceutical Section
Abstract #318683 View Presentation
Title: Randomization does not prevent bias when we assess treatment by covariate interaction
Author(s): Lei Nie* and Zhiwei Zhang and Jialu Zhang
Companies: FDA and FDA/CDRH and FDA
Keywords: Bias; ; Treatment by covariate interaction; ; subgroup analysis
Abstract:

Inference for the overall treatment effect resulted from a randomized clinical trial has the best credibility. In a traditional paradigm, where a clinical trial is primarily designed to answer a single question regarding the average (overall) treatment effect, randomization effectively prevents systematic confounding and bias when we assess the overall treatment effect. In an increasingly important new paradigm, inference of treatment heterogeneity is key to comparative effectiveness. In this new paradigm, randomization does not prevent systematic bias when we assess treatment by covariate interaction. Through a case study, this paper explains how this phenomenon occurs and examines whether and how this problem could be solved.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2016 program

 
 
Copyright © American Statistical Association