Determining the non-inferiority margin for patient reported outcomes

Area of Science:

• Clinical Trials
• Biostatistics
• Patient-Reported Outcomes

Background:

• Non-inferiority trials rely on a pre-defined non-inferiority margin (delta).
• Delta is often subjectively negotiated between sponsors and regulatory agencies.
• Establishing a clinically relevant delta is challenging, especially for patient-observed outcomes.

Purpose of the Study:

• To investigate empirical methods for determining the non-inferiority margin (delta) for patient-observed endpoints.
• To compare anchor-based and distribution-based approaches for deriving delta.
• To assess the feasibility of using patient-derived minimal clinically important difference (MCID) to define delta.

Main Methods:

• Utilized an anchor-based approach with non-parametric discriminant and ROC analyses.
• Employed a distribution-based approach using Norman's half standard deviation rule.
• Applied these methods to three distinct patient-observed endpoints: endometriosis pain, facial acne, and hot flushes.

Main Results:

• All three methods produced comparable results for delta across the studied endpoints.
• Empirically derived MCIDs were often smaller than or similar to previously used deltas.
• In one case, the MCID-derived responder definition was accepted by the FDA.

Conclusions:

• Delta for patient-observed endpoints can be empirically derived using MCID.
• This empirical approach is preferable to subjective "negotiated" values for delta.
• Empirical derivation enhances the clinical relevance and validity of non-inferiority margins.