A Likelihood-Based Approach to Responder Analysis When Data Are Grouped

• Activity Number: 132
• Type: Contributed
• Date/Time: Monday, July 30, 2012 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Daniel Bonzo*+ and Erold Ryan Bonzo
• Companies: XenoPort, Inc. and University of the Philippines
• Keywords: responder analysis ; likelihood-based ; grouped data ; GERD

Abstract

Responder analysis is typically employed to further assess the therapeutic effect of a given treatment. The interest in responder analysis is the identification of an optimal threshold that will determine whether a clinically meaningful improvement for a patient has been achieved. Such a clinically meaningful improvement is usually equated to treatment success. Furthermore, the threshold should result in a differentiation of the proportion of patients meeting the success criterion between treatment groups.

A likelihood-based approach is formulated in a situation when the data involved is grouped. Formally, suppose one observes Y = y, whenever x belongs to I(y), where I is an interval in the support of the distribution of the underlying variable X. The underlying variable X is called a latent variable as it is unobserved. In this case, information on the underlying distribution is provided for by the grouped data I(y).

Application of the method is demonstrated using simulated data based on a clinical trial on patients with Gastroesophageal Reflux Disease (GERD) who were incomplete responders to a Proton Pump Inhibitor (PPI).