Non-Inferiority Clinical Trial Design, Drug Approval, and Implications for Clinicians in Prescribing Oral Anticoagulants

Clinical trials with a non-inferiority (NI) design are becoming more common, and serve as clinical evidence for approval of a new drug entity or new indication. This review examines published non-inferiority studies for anticoagulant therapies to 1) assess the rigor of design elements and the statistical validity of reported conclusions and 2) conformance with FDA guidance for industry regarding non-inferiority designs. METHODS The Cochrane Systematic Review protocol was used to identify research publications from 2000–2016 that utilized a non-inferiority design in clinical trials. To be considered for further review publications were required to be published in English, have complete Phase III or Phase IV clinical trial results, and identify NI as the primary analytic approach. 59 publications were identified using the initial search terms. During review, articles describing a proposed study, providing correction of a previously published study, or a commentary only were excluded. Fifteen articles met inclusion/exclusion criteria for final evaluation by two reviewers. 1 study failed to meet the study’s pre-specified non-inferiority margin; 5 (35.7%) stated incorrect interpretation (e.g., “as effective as”) and 1 study reported findings compared to a placebo. Clinical margins to establish non-inferiority were arbitrarily established by each investigator, making judgement about the best therapeutic choice ambiguous. The outcome measure of time in therapeutic INR range for active comparator also differed widely between the published studies, weakening the NI study design validity at the expense of patient safety. NI trials reported in clinical literature may be misidentified and present conclusions not justified by the analytic findings. The complexity of the statistical analysis of NI trials and the absence of the requirement to determine efficacy of a new drug as a measured outcome in NI trials, makes clinically meaningful application of NI trial findings difficult.