On the Exact Interval Estimation for the Comparisons in the Probability of Cost-Effectiveness

Abstract

In health economics, public health and health science, the cost-effectiveness analysis is a type of economic evaluation that examines the costs-effectiveness of two competing treatments or groups. However, limited statistical approaches have been developed for the evaluation comparing more than two treatments. The probability of cost-effectiveness is a new measure, which is defined as the chance that the cost spent per effect for a case is cheap enough to overcome that for a control. Unlike the incremental cost-effectiveness ratio (ICER) and its derivative measures that are constructed by mean cost and mean effectiveness, the probability of cost-effectiveness is not sensitive to extreme value. In this study we propose to use the concept of generalized pivotal quantities to construct an exact interval estimation for the several specific comparisons in the probability of cost-effectiveness. Simulation results demonstrate that our proposed interval estimation provides not only sufficient probability coverage but reasonable expected length. Key words: generalized confidence interval, many-to-one comparison, skewness, area under the ROC curve