Bayesian Sample Size Determination for a Cost-Effectiveness Study with Censored Costs

• Activity Number: 80
• Type: Contributed
• Date/Time: Sunday, July 29, 2012 : 4:00 PM to 5:50 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Daniel Beavers*+ and James Stamey
• Companies: Wake Forest University Health Sciences and Baylor University
• Keywords: cost-effectiveness ; clinical trial ; sample size ; Bayesian ; censoring

Abstract

We propose a Bayesian approach for the design of time-to-event clinical trials in which the outcome of interest is the cost-effectiveness of the treatment. A common issue with such trials is that costs are typically right-censored within observations whose failure times are censored. Using a previously published model that assumes Weibull-distributed failure times and gamma-distributed costs, we develop a method for determining sample sizes for clinical trials using a Bayesian hypothesis test criterion for the primary outcome. The method incorporates parameter uncertainty in the design phase while both accounting for censored observations and allowing for the bivariate relationship between costs and time.