This paper deals with the handling of uncertainty in the context of cost-effectiveness analyses, through the building of confidence regions for the incremental cost-effectiveness ratio (ICER), to provide a statistical descriptive tool for helping medico-economic decision-making. However, the confidence region for the ICER, mathematically provided, cannot be directly usable for decision-making, because two identical ICERs can be associated with two totally opposite findings. To solve this problem, algorithms are developed in this paper, to provide on the one hand a directed confidence region usable for decision-making, and on the other hand a decision directly from the data.