|
Activity Number:
|
687
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Thursday, August 4, 2016 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
|
Abstract #320396
|
View Presentation
|
|
Title:
|
Utilization of Historical Patient-Level Data for Efficient Trial Design
|
|
Author(s):
|
Zachary Thomas* and Tianle Hu and Nathan Enas and Honglu Liu
|
|
Companies:
|
Eli Lilly and Company and Eli Lilly and Company and Eli Lilly and Company and Eli Lilly and Company
|
|
Keywords:
|
design ;
clinical trials ;
historical ;
borrowing ;
matching
|
|
Abstract:
|
A variety of statistical methods have been proposed for incorporating historical controls into the design and analysis of randomized controlled trials of potential new medicines. These methods often involve techniques to allow adaptive borrowing of information (or perhaps model-based covariate adjustment) at the study level (e.g., from literature reports of past trials). However, it is frequently the case that the number of available historical controls is too small (often fewer than three) to adequately inform inference in such adaptive methods without specification of highly informative prior distributions (or other restrictions). As an alternative, we explore the potential utility of simple matching-based procedures for incorporating patient-level data from historical controls in the circumstance that such data (for patient response and prognostic factors) are available (while acknowledging that such methods are also not immune to issues of confounding due to study-effect). We will share numerical results from several real and simulated datasets and discuss potential challenges/benefits.
|
Authors who are presenting talks have a * after their name.
