Parameter Estimation and Power Calculation in Zero-Inflated Regressions Models for Clinical Trial with Over-Dispersed Count Data

• Activity Number: 657
• Type: Contributed
• Date/Time: Thursday, August 7, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Jiang Hu*+
• Companies: FDA
• Keywords: Zero-Inflated Models ; Over-dispersed data ; Clinical trial

Abstract

Zero-inflated regression models, such as Zero-Inflated Poisson (ZIP) and Zero-Inflated Negative Binomial (ZINB), have been applied in observed clinical count data with an excess of zeroes. However, parameter estimates may be seriously biased if the nonzero observations are over-dispersed due to the nature of the data. In this paper we investigate the effect of the distribution of data on the parameter estimation, as well as sample size and power calculation for both ZIP and ZINB regression models. Numerical simulations are implemented to compare the performance of ZIP and ZINB. The purpose of this paper is to illustrate the differences between data distributions and models and to explore how to optimize the study design based on the available information.