Understanding non-inferiority trials: an introduction

Abstract

Non-inferiority studies are appearing in medical journals with increasing frequency. At first glance, one might think the intent of such a study is to publish negative results or, in the worst case scenario, to create a marketing ploy that would have us consider replacing a tried-and-true treatment with a new one. 1 Nevertheless, these trials can bring value by introducing new treatments that may be therapeutically similar to the standard treatment but with certain advantages. In this issue of the Journal, Merry et al. 2 show that clevidipine, a short-acting dihydropyridine calcium channel antagonist, is non-inferior to nitroglycerin (NTG) in the pre-bypass period of patients undergoing cardiac surgery. The information needed for a clinician to assess a non-inferiority trial is highlighted in the context of this study. The non-inferiority trial attempts to show that the new treatment is not an unacceptably worse alternative to the standard. 3 This is done by stating a margin of non-inferiority for the effect of the treatment. The non-inferiority margin describes the amount of reduction in efficacy that will be tolerated. As long as the new treatment is not worse than the standard treatment by this margin, the new treatment would be considered non-inferior. What is the rationale for performing a non-inferiority trial? In the regulatory setting, the efficacy of a new treatment is usually shown in a placebo-controlled trial. In some situations, however, it would be unethical to withhold treatment; hence, a non-inferiority trial is used to indirectly show the efficacy of the new treatment. Alternatively, a trial sponsor may have a new treatment that provides only a small advantage over the standard. Showing this superiority would require a very large study, while non-inferiority could be shown with a much smaller sample. In the non-regulatory setting, non-inferiority trials are most often used to show that an alternative treatment with advantages regarding safety, convenience, or cost has an efficacy similar to, or at least not much worse than, the standard treatment. The first order of business is to decide if a non-inferiority design is reasonable. The reader must question whether the purported advantages of the new treatment over the standard would justify adopting the new treatment should non-inferiority be shown. Typical advantages are price, ease of use, or a reduced number of side effects. If the answer is ''no'', then the non-inferiority design should be abandoned. If the answer is ''yes'', then the reader will need to consider the next issue, i.e., the assay sensitivity and the consistency of the control treatment. In a non-inferiority trial, two comparisons are made: the explicit comparison of the new treatment with an active control treatment and the implicit comparison of the active control treatment with placebo. Generally, there is no placebo control in these studies so the reader must be convinced that the treatment (active control) against which the new treatment is being compared would provide effective management (i.e., be superior to a placebo) in the context in which it is being used. This is referred to as the assay sensitivity. 4 If the efficacy of the standard is border-line, then showing that the new treatment is non-inferior is of little value as it may in fact be no different from a placebo. It is also important that previous trials using the standard