|
Activity Number:
|
243
- Contributed Poster Presentations: Biopharmaceutical Section
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, July 31, 2017 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
|
Abstract #323028
|
|
|
Title:
|
Constrained Longitudinal Model for Analyzing Pre- and Post-Treatment Count Data
|
|
Author(s):
|
Yongming Qu* and Junxiang Luo and Hong Zhao
|
|
Companies:
|
Eli Lilly and Company and Eli Lilly and Company and Abbott Laboratories
|
|
Keywords:
|
constrained longitudinal model ;
negative binomial ;
hypoglycemia
|
|
Abstract:
|
Hypoglycemia events are an important clinical outcome in diabetes clinical trials. Negative binomial regression is a common statistical model for analyzing such count data. To fully utilize the baseline hypoglycemia data, one option is to treat baseline hypoglycemia as a covariate [say covariate analysis model (CAM)]. Another option is to use constrained longitudinal model (CLM) to model the baseline and post-baseline hypoglycemia data as longitudinal repeated dependent variable and to pose constraints on baseline. CLM has been well studied for linear models. It has been shown that the two approaches provide equivalent results when there are no missing data, and CLM may be more efficient in the presence of missing baseline values. However, no research has been conducted to compare these 2 methods for negative binomial regressions. In this research, we evaluate the two analysis models for the case of negative binomial through simulation. Simulation showed that CLM has comparable or higher statistical power than CAM while preserving Type-1 error under multiple scenarios.
|
Authors who are presenting talks have a * after their name.
