A Note on Confidence Intervals in Cost-Effectiveness Analysis

How to obtain confidence intervals for cost-effectiveness ratios is complicated by the statistical problems to obtain a confidence interval for a ratio of random variables. Different approaches have been suggested in the literature, but no consensus has been reached. In this note we propose an alternative simple solution to this problem. By multiplying the effectiveness units by the price per effectiveness unit, both costs and benefits can be expressed in monetary terms and standard statistical techniques can be used to estimate a confidence interval for net benefits. This avoids the ratio estimation problem and explicitly recognizes that the price per effectiveness unit has to be known to provide cost-effectibeness analysis with a useful decision rule.