Understanding & Interpreting Non-Inferiority Trials

Preparation for the P506 (15NOV16) & P535B (2FEB17) sessions

By the end of the session, upon study and reflection, students should be able to

  1. Name scenarios where non-inferiority trials are preferred over superiority trials, and scenarios where they are not.
  2. Explain what an MCID/non-inferiority margin is and roughly how they are determined, especially the “setting the margin” and “imputed placebo” approaches.
  3. Graphically depict the non-inferiority assumptions and the primary endpoint result for a clinical trial in terms of whether the intervention is non-inferior or not.
  4. Explain the role of ITT analysis vs. on-treatment analysis in superiority vs. non-inferiority trials.
  5. Describe the difference between superiority and non-inferiority trials in terms of what it means to be “conservative” and in terms of Type 1 and Type II errors.
  6. Be able to calculate the confidence intervals around any non-significant trial endpoint and personally evaluate it from the viewpoint of non-inferiority. Use my [stats calculator] for this.
  7. ADVANCED: Explain what a p-value on a non-inferiority finding means.
  8. ADVANCED: Think about the methodologic, practical, and ethical issues in TECOS and EXAMINE

Required Prior Knowledge & Skills

Self-test that you can do these things before showing up for the session, because NI trial design/interpretation is a special case of these concepts.

  1. be able to accurately explain what a p-value is
  2. be able to state what the usual null and alternative hypotheses are in a superiority trial
  3. be able to explain what it means to make a Type I and Type II error both in lay-language, and in terms of the null & alternative hypotheses
  4. explain why ITT analysis is the gold-standard approach in superiority RCTs

During the Sessions

PHAR 506 - NOVEMBER 2016 We'll discuss the foundational concepts behind NI trials in terms of

  • clinical & design considerations
  • ethical appropriateness
  • statistical considerations

We’ll do roundtable discussion using roughly a CONSORT-NI approach, so please come prepared to discuss the assigned trial. We will concentrate on the issues unique to NI trial design and interpretation.


  • Bring vexing and/or interesting issues and questions for discussion.

Common Misconceptions and Sticking Points

  • how does the definition of conservative differ in a NI trial from a superiority trial?
  • precisely why (practically and philosophically) is the on-treatment (non-ITT) analysis so important in an NI trial?
  • what does it mean to report a “p-value for non-inferiority”

Challenge Yourself

  • how does the definition of conservative differ in a NI trial from a superiority trial?
  • what does it mean to report a “p-value for non-inferiority”?
  • What are the scientific and ethical issues here? EXAMINE
  • A million points if you can figure out how they devised their NI-margin of 1.46 in RE-LY & ROCKET-AF.

Supportive Literature & Further Reading

Other classic NI trials for self-study and skill testing