Finding the Right Distribution for Highly Skewed Zero-Inflated Clinical Data

• Activity Number: 127
• Type: Contributed
• Date/Time: Monday, August 2, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistical Consulting
• Author(s): Resmi Gupta*+ and Bradley S. Marino and James F. Cnota and Richard F. Ittenbach
• Companies: Cincinnati Children's Hospital Medical Center and Cincinnati Children's Hospital Medical Center and Cincinnati Children's Hospital Medical Center and Cincinnati Children's Hospital Medical Center
• Keywords: zero-inflated Poisson ; skewed data ; generalized linear models

Abstract

Discrete, highly skewed distributions with excess numbers of zeros may result in biased estimates and misleading inferences if the zeros are not properly addressed. A clinical example includes children with EP disorders in which many of the children are treated without surgery resulting in many zeros. The purpose of the current study was to identify the optimal modeling strategy for highly skewed, zero-inflated data often observed in a clinical setting by: (a) simulating skewed, zero-inflated count data; (b) fitting the simulated data with Poisson, Negative Binomial, Zero-Inflated Poisson (ZIP) and Zero-inflated Negative Binomial (ZINB) models; and, (c) applying the aforementioned models to actual, highly skewed, clinical data of children with an EP disorder. The ZIP model was observed to be optimal based on traditional fit statistics as well as bias, mean-squared error, and coverage.