Choosing from multiple alternatives in cost-effectiveness analysis with fuzzy willingness-to-pay/accept and uncertainty

Abstract

Cost-effectiveness analysis of medical technologies requires valuing health, an uneasy task, as confirmed by variability of published estimates. Treating the willingness-to-pay/accept (WTP/WTA) as fuzzy seems an intuitive solution. Based on this premise, I construct a framework allowing to compare multiple health technologies using choice functions. The final choice must be crisp, so I discuss various defuzzification methods and show that using indecisiveness point (IP) for WTP/WTA (the value the decision maker equally approves/disapproves) has desirable properties: satisfying the independence of irrelevant alternatives and not treating the Likert scale as interval. I suggest three approaches to infer about IP with Likert-based surveys in random samples (hypothesis testing, Bayesian or frequentist estimation). No difference between IPs for WTP/WTA is found, and an explanation of the WTP-WTA disparity is provided. Estimating IP results in stochastic uncertainty, and I show how to conduct sensitivity analysis in the framework and what new insight is gained.