Cost-effectiveness analysis (CEA) is extensively employed by healthcare policymakers to guide funding decisions and inform optimal design of medical interventions. In the CEA literature, willingness to pay (WTP) serves as a common metric for converting health benefits into monetary value and defining the net monetary benefit of an intervention. However, there is no universally accepted value for WTP. To address this issue, we propose presenting policymakers with a comprehensive menu of strategies that are proven cost-effective across a reasonable range of WTP values. In our approach, we consider a setting where the medical decision-making process can be formulated as a parametric linear programming model. We have developed a novel algorithm aimed at efficiently constructing the menu of cost-effective policies. Our algorithm is particularly suited for Constrained Markov Decision Process (CMDP) and Constrained Partially Observable Markov Decision Process (CPOMDP) models, which are commonly utilized modeling frameworks for addressing sequential medical decision-making problems. We have applied our modeling framework to design hearing loss screening strategies for cystic fibrosis patients. Informed by a validated, data-driven model, we have developed several heuristic and approximate policies, allowing policymakers to balance between performance and ease of implementation.
Article
View Efficient Discovery of Cost-effective Policies in Sequential, Medical Decision-Making Problems