Use of Bayesian Methods in Budget Impact Analysis

• Activity Number: 263
• Type: Topic Contributed
• Date/Time: Tuesday, August 10, 2004 : 2:00 PM to 3:50 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Shu Han*+ and Ya-Chen T. Shih
• Companies: Rice University and University of Texas M. D. Anderson Cancer Center
• Keywords: inhomogeneous markov chain ; budget impact analysis ; Bayesian ; generic entry ; economic evaluations

Abstract

An inhomogeneous Markov chain model was proposed to include variations in patients (pts) mix and drug prices in BIA using Bayesian methods. Markov states were categorized by whether a pt was treated with a generic drug, an existing brand-name drug, or a new drug. Modifications were made at each cycle for newly diagnosed incident cases and exiting cases due to cure or death. Also considered is the difference in treatment preference b/t the current and newly diagnosed pts. A case study from simulated data was used to compare the budget impact of including vs. excluding a new drug in a health plan, taking a payer's perspective and a five-year time frame. Results were presented in a probabilistic plot similar to cost-effectiveness acceptability curve. Adding the new drug to the plan was found to increase the budget increase in the short run but not in the long run. The probability that including the new drug would increase in the budget by 10% is 9%, 26% in a one- and two-year time frame, and it becomes cost neutral in the five-year time frame. Our model provides a framework to examine time-varying parameters in BIA and generates estimates that better reflect the real healthcare market.