Keywords: Bootstrap calibration, cost-effectiveness analysis, nonparametric tolerance limits
Unlike common population-level tools such as the incremental cost-effectiveness ratio (ICER) and the average cost-effectiveness ratio (ACER), we can tailor individual cost-effectiveness ratios and their tolerance limits to identify subgroups of interest, such as extreme regions, in the distribution of the potential cost effectiveness ratios. Previous methods only provided one-sided limits, could not be adapted to two-sided intervals, and had lower estimated coverage probabilities. In the talk, we will demonstrate and discuss a methodology to construct one- and two-sided tolerance limits using a parametric bootstrap approach, coupled with a bootstrap calibration in order to improve accuracy. The approach will be illustrated using an example from the health economics literature (Gardiner et al.2000).