Cost-effectiveness angles are an attractive measure of performance when comparing effects and costs of health-care therapies because they have a clear interpretation and are well suited for statistical inference. In clinical trials, a common setup is the comparison of multiple new therapies with a single control. If cost-effectiveness angles are calculated for each comparison, multiplicity issues should be taken into account when quantifying uncertainty of the point estimates.
Therefore, this paper proposes a parametric test for multiple cost-effectiveness angles that guarantees strong family-wise error rate control. The idea is to replace the test of m cost-effectiveness angles as a union-intersection test of 3m linear hypotheses. Considering the correlation structure of the individual test statistics for the linear hypotheses leads to a maximum-type test for the intersection hypothesis. Inverting these test decisions then gives simultaneous CIs of cost-effectiveness angles with the appropriate coverage probabilities. Copyright © 2012 John Wiley & Sons, Ltd.