WW11: USING MARKOV DECISION MODELS TO DEVELOP OPTIMAL TREATMENT PATHWAYS: AN ILLUSTRATION

Abstract

OBJECTIVES: The objective of this workshop will be to develop skills in the use of Markov decision models to develop treatment pathways that are optimal from a pharmacoeconomic perspective. PARTICIPANTS WHO WOULD BENEFIT: Analysts or decision-makers involved in the conduct or evaluation of pharmacoeconomic studies. Cost-effectiveness analysis of an intervention for a chronic condition requires estimates of cost and efficacy over the rest of the patient's life time. However efficacy estimates are commonly derived from clinical trials with limited duration of follow-up. This creates the need for modeling to estimate cost and efficacy beyond the follow-up period. Markov modeling is the most commonly used modeling technique to perform this estimation. However, as commonly applied, Markov models are used to determine if the new intervention is cost-effective if used instead an old one at one point in the patient's lifetime. These models do not attempt to determine the lifetime optimal treatment pathway for treating the disease. We show how Markov decision models can be used to develop an optimal treatment pathway, by assigning a treatment option to each health state so as to maximize overall net benefit. We will demonstrate how to derive the optimal treatment pathway using dynamic programming. We will present an illustrative example of the use Markov decision model for a chronic disease such as HIV. We will discuss some potential ethical issues that could be raised by the optimal policies derived from Markov decision models. We will conclude the workshop with an interactive discussion of the benefits and drawbacks of Markov decision models.