Interpreting the results of noninferiority trials—a review

Abstract

Noninferiority trials are becoming increasing common, but are often poorly reported and misunderstood. A better understanding of the new components of a noninferiority trial and their interpretation is needed. Noninferiority trials are an extension of conventional superiority trials, which provide a basis for determining if a new treatment, which may have advantages other than efficacy, has sufficient efficacy to be useful in certain situations. A key feature is the need to specify a clinical noninferiority margin above which the lower boundary of the confidence interval for the difference between the new treatment and the conventional treatment must lie. In most cases a nontreated control arm is not included, and when the efficacy of the new treatment is less than that of the standard treatment, determining its efficacy versus no treatment can be a major challenge. Treatments meeting a clinical noninferiority requirement can be statistically significantly superior to standard treatment, of similar efficacy (i.e., no significant difference), or even significantly inferior in a conventional analysis. Noninferiority comparisons are an important addition to the reporting of clinical trials, but require prior consideration of several factors that conventional superiority analyses do not address.