Analyzing Excessive No Changes in Clinical Trials with Clustered Data

• Activity Number: 213
• Type: Contributed
• Date/Time: Tuesday, August 13, 2002 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section*
• Author(s): Shou-En Lu*+ and Yong Lin and Weichung Shih
• Affiliation(s): University of Medicine and Dentistry of New Jersey and University of Medicine and Dentistry of New Jersey and University of Medicine and Dentistry of New Jersey
• Keywords: clustered data ; excessive zeros ; mixture model ; two-part model

Abstract

This paper considers clinical trials in which the efficacy measure is taken from several sites within each patient, for example, alveolar bone height of the tooth sites, or bone mineral densities of the lumbar spine sites. Since usually only a small portion of these sites will exhibit changes--hence, many zeros in the data set--the conventional method of using per-patient average gives a diluted result. Different methods have been proposed for this type of data in simple case without clusters, including the popular "two-part model" method (Lachenbruch, 2001), which is the same as the "composite approach" for discrete and continuous data in Shih and Quan (1997, 2001). In this paper, we model the data with excessive zeros and clusters using mixture of distributions and take into account of possible measurement errors. This mixture model includes the two-part model as a special case when one component of the mixture degenerates.