A Three-Objective Optimization Approach to Cost Effectiveness Analysis Under Uncertainty

摘要

1 Introduction One of the most important quantitative decision analysis tools for healthcare resource allocation is cost-effectiveness analysis (CEA). A known drawback of CEA lies in the fact that usually, neither cost nor effect of a health programme (medical intervention, treatment, screening programme etc.) are known with certainty in advance. Thus, a stochastic portfolio optimization problem for health programmes with the two different objectives "cost" and "health effect" arises. Several articles have studied healthcare resource allocation under stochastically modelled uncertainty [1, 3, 9, 11, 13]. However, up to now, no broadly accepted method for CEA under uncertainty seems to exist. Multiobjective stochastic optimization problems are notoriously hard (cf. [2, 7]). The possible dependence between the random objectives increases the difficulty, especially in the case where the decision maker is risk-averse. A broadly accepted analytical approach to decisions under risk aversion is expected utility theory, which is closely related to stochastic dominance concepts. This approach has recently also been extended to the multi-objective case [4, 8]. Nevertheless, the models in the last-mentioned articles assume that the objectives can be traded against each other, and this stipulates a perfect market which is usually not present in the health sector.