Understanding non-inferiority trials: Lessons from trials comparing thrombectomy with or without intravenous...

Area of Science:

• Neurology
• Clinical Trials
• Medical Research

Background:

• Non-inferiority (NI) trials assess new treatments with potential benefits (e.g., fewer side effects, lower cost) against established treatments without significant efficacy loss.
• NI trials present unique design and interpretation challenges, including selecting appropriate non-inferiority margins.

Purpose of the Study:

• To examine six non-inferiority trials comparing thrombectomy with or without intravenous thrombolysis (IVT) in acute stroke.
• To review fundamental problems and concepts crucial to understanding NI trials, such as assay sensitivity, constancy assumption, and NI margins.

Main Methods:

• Review of six non-inferiority trials comparing thrombectomy alone versus thrombectomy with IVT.
• Analysis of trial design, non-inferiority margins, and statistical outcomes.
• Examination of concepts like assay sensitivity, constancy assumption, and non-inferiority margins.

Main Results:

• Two of six trials found thrombectomy alone non-inferior to thrombectomy with IVT, often due to large margins or premature stopping.
• Four trials failed to demonstrate non-inferiority, with generally unjustified and clinically unacceptable margins.
• A meta-analysis showed non-inferiority at a 10% margin but not at a clinically relevant 5% margin.

Conclusions:

• Thrombectomy alone has not been adequately demonstrated as non-inferior to thrombectomy with IVT in acute stroke.
• The non-inferiority trial design appears poorly suited for acute stroke treatment evaluation.
• Assessing superiority using surrogate outcomes with established clinical relevance may be a more appropriate alternative approach.