|
Activity Number:
|
193
- Modeling, Design Strategies and Assessments of Biomarkers
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 4, 2020 : 10:00 AM to 2:00 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
|
Abstract #313605
|
|
|
Title:
|
Should We Check the Normality? The Impact of Model Assumptions in Early Phase Clinical Studies
|
|
Author(s):
|
Qianyu Dang* and Ran Bi
|
|
Companies:
|
FDA Center for Drug Evaluation and Research (CDER) and FDA/HHS
|
|
Keywords:
|
linear mixed-effects model;
paired t-test;
type I error;
HAP studies
|
|
Abstract:
|
Human abuse potential (HAP) studies are early phase clinical trials used to evaluate the abuse potential of a new drug product containing central nervous system (CNS) active drug substance. The outcome measures we are interested in are the pairwise differences between different arms and the sample sizes are usually 30 to 50 with cross-over design. Currently the statistical model we use is a linear mixed-effects model and then t test for paired differences if normality assumptions of mixed-effects model do not hold. Nonparametric approach will be applied if paired difference is not normal as well. However, minor deviation in skewness may not have significant impact on type I error. Plus, multiple tests for normality may also inflate the type I error rate. In this study, we used numerical simulation to evaluate the overall type I error rates under different scenarios in the deviation from normality across the entire testing procedure. The results indicate that the type I error rate can inflate substantially only if the skewness is severe and the impact of multiple testing for normality is minimal. This result may also be applied to other studies with relatively small samples.
|
Authors who are presenting talks have a * after their name.
