|
Activity Number:
|
322
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Tuesday, August 8, 2006 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
| Abstract - #306093 |
|
Title:
|
Bayesian Modeling of Safety Data Using Databases of Placebo Patients
|
|
Author(s):
|
Jose Pinheiro*+ and Chyi-Hung Hsu
|
|
Companies:
|
Novartis Pharmaceuticals Corporation and Novartis Pharmaceuticals Corporation
|
|
Address:
|
One Health Plaza, East Hanover, NJ, 07936,
|
|
Keywords:
|
early development ; proof-of-concept ; mixed-effects models ; CRM ; adverse events
|
|
Abstract:
|
Evaluation of safety signals in early clinical development trials is often a difficult statistical problem because of the small number of patients involved and the short study duration typically observed. To make matters more challenging, safety variables often are measured in dichotomized form (e.g., occurrence of lab parameter elevations beyond a clinically relevant threshold). We will discuss and illustrate the use of Bayesian mixed-effects models to represent longitudinal continuous safety data and combine them with information from databases of placebo patients to implement a CRM-like algorithm to identify safety signals in early development data more efficiently. Simulation results, based on real clinical trial data, will be used to illustrate the proposed methods.
|
