Determining the non‐inferiority margin for patient reported outcomes

Christoph Gerlinger, Thomas Schmelter
Pharmaceutical statistics · 2011 · vol. 10(5) , pp. 410–413 · doi:10.1002/pst.507 · PMID:21932298 · W1494653833
article OA: closed CC0 ⤵ 3 in-corpus citations
View on OpenAlex View on PubMed View at publisher
⚙ AI-generated summary by claude@2026-06, 2026-06-07 ⓘ

This paper describes methods to empirically determine the non-inferiority margin delta for patient-reported outcomes using minimal clinically important difference and compares these findings to previously used margins.

One-sentence paraphrase of the abstract; not a substitute for reading it. No clinical advice. How this works

Abstract

One of the cornerstones of any non-inferiority trial is the choice of the non-inferiority margin delta. This threshold of clinical relevance is very difficult to determine, and in practice, delta is often "negotiated" between the sponsor of the trial and the regulatory agencies. However, for patient reported, or more precisely patient observed outcomes, the patients' minimal clinically important difference (MCID) can be determined empirically by relating the treatment effect, for example, a change on a 100-mm visual analogue scale, to the patient's satisfaction with the change. This MCID can then be used to define delta. We used an anchor-based approach with non-parametric discriminant analysis and ROC analysis and a distribution-based approach with Norman's half standard deviation rule to determine delta in three examples endometriosis-related pelvic pain measured on a 100-mm visual analogue scale, facial acne measured by lesion counts, and hot flush counts. For each of these examples, all three methods yielded quite similar results. In two of the cases, the empirically derived MCIDs were smaller or similar of deltas used before in non-inferiority trials, and in the third case, the empirically derived MCID was used to derive a responder definition that was accepted by the FDA. In conclusion, for patient-observed endpoints, the delta can be derived empirically. In our view, this is a better approach than that of asking the clinician for a "nice round number" for delta, such as 10, 50%, π, e, or i.

My notes (saved in your browser only)

Condition tags

endometriosischronic_pelvic_pain

MeSH descriptors

Clinical Trials as Topic Endometriosis Pelvic Pain Research Design Self Report Acneiform Eruptions Acneiform Eruptions Acne Vulgaris Acne Vulgaris Clinical Trials as Topic Discriminant Analysis Disease Progression Drug and Narcotic Control Drug-Related Side Effects and Adverse Reactions Endometriosis Endometriosis Female Hot Flashes Hot Flashes Hot Flashes

Citation neighborhood (sparse)

Too few in-corpus citations on either side for a chart; here are the lists.

Cites (2)

Cited by (3)

References (14)

Cited by (3)

Source provenance

europepmc
last seen: 2026-06-11T06:19:48.454388+00:00
openalex
last seen: 2026-06-10T17:14:06.276822+00:00
pubmed
last seen: 2026-05-13T22:16:29.858026+00:00
unpaywall
last seen: 2026-06-04T02:00:05.705006+00:00
License: CC0 · commercial use OK