|
Activity Number:
|
549
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 2, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #309110 |
|
Title:
|
Comparisons of Statistical Methods in Analyzing Clinical Data Through Empirical Simulations
|
|
Author(s):
|
Yongming Qu*+ and Lei Xu and Pandurang M. Kulkarni
|
|
Companies:
|
Eli Lilly and Company and University of Wisconsin-Madison and Eli Lilly and Company
|
|
Address:
|
Lilly Corporation Center, Indianapolis, IN, 46285,
|
|
Keywords:
|
Empirical simulation ; resampling with replacement ; statistical power ; type I error
|
|
Abstract:
|
With increasing requirements, from both regulatory and scientific community, for prespecification of details of all analyses prior to unblinding of data in clinical trials, it is critical that one pre-specifies the most appropriate statistical method and model. Selecting a model based on assumption checking (after the data has been unblinded) either inflates type I error or compromises the statistical power to detect a difference. Previous research is mainly focused on comparing various analysis models either through simulation or case studies. We propose an empirical procedure that utilizes the historical data, evaluates various models of interest, and provides an optimal choice of models to be used for pre-specification in future studies. We applies this method to several commonly used laboratory variables and gives guidance about how to analyze these variables.
|
