Good enough: a primer on the analysis and interpretation of noninferiority trials

Abstract

Active-control noninferiority trials are being performed with increasing frequency when standard placebo-controlled trials are considered unethical. Three attributes are optimally required to establish noninferiority: 1) The treatment under consideration exhibits therapeutic noninferiority to the active control; 2) the treatment would exhibit therapeutic efficacy in a placebo-controlled trial if such a trial were to be performed; and 3) the treatment offers ancillary advantages in safety, tolerability, cost, or convenience. Trials designed to show noninferiority require an appropriate reference population, a proven active control and dose, a high level of adherence to treatment, and adequate statistical power. However, the formal analysis of such trials is founded on several assumptions that cannot be validated explicitly. These assumptions are evaluated in the context of 8 recently published noninferiority trials. The analyses in this paper confirm the establishment of noninferiority in only 4 of the 8 trials. The authors conclude that if noninferiority trials are to be applied to clinical and regulatory decisions about the marketing and use of new treatments, these assumptions must be made explicit and their influence on the resultant conclusions assessed rigorously.