Randomized clinical trial design for assessing noninferiority when superiority is expected.

Abstract

The randomized clinical trial (RCT) is the gold standard for definitive evaluation of new therapies. RCTs designed to show that the therapeutic efficacy of a new therapy is not unacceptably inferior to that of standard therapy are called noninferiority trials. Traditionally, noninferiority trials have required very large sample sizes. Sometimes, a new treatment regimen with a favorable toxicity and/or tolerability profile is also expected to have some modest improvement in efficacy. In such specialized settings we describe a hybrid trial-design approach that requires a dramatically smaller sample size than that of a standard noninferiority design. This hybrid design can naturally incorporate a formal test of superiority as well as noninferiority.