Evaluating the efficacy of a combination therapy

Standard multiple comparison techniques do not adequately address the situation in which one must demonstrate a combination therapy's superiority to both of its components. This paper proposes several alternative tests to deal with this situation, and evaluates the tests with a Monte Carlo sampling experiment that compares the actual with the nominal significance levels of the tests. Data from a clinical trial that compares the efficacy of a combination analgesic/muscle relaxant to its components and placebo in the treatment of low back pain and spasm illustrate the tests.