|
Activity Number:
|
522
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 7, 2008 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
| Abstract - #302531 |
|
Title:
|
Permutation Test Approach To Compare Two Groups of Censored Lognormal Data with Multiple Detection Limits
|
|
Author(s):
|
Wei Zhong*+ and Jeffrey Welge and Linda Levin and Paul Succop and Rakesh Shukla
|
|
Companies:
|
ICON Clinical Research and University of Cincinnati and University of Cincinnati and University of Cincinnati and University of Cincinnati
|
|
Address:
|
, , 95134,
|
|
Keywords:
|
Lognormal ; Permutation Test ; MLE-EM ; Type I Error Rate ; Power
|
|
Abstract:
|
Censored lognormal data with multiple lower detection limits frequently arise in safety laboratory assessment. A common practice to compare the means between two groups of such data is to perform the two sample t-test on the log-transformed outcome and to report the resulting P-values for the null hypothesis based on the original data. The problem with such a practice is that it essentially tests the equality between the two group medians. The permutation test approach with the mean estimation obtained from the MLE-EM algorithm was studied in this paper. The proposed method was evaluated through a series of simulation studies and its performance was compared to the two sample t-test and Z-score test. Simulation results consistently showed that the proposed approach provides the minimum Type I error rates and maximum power.
|
