The DOI matches its Crossref record; 16 of the 16 references it cites were found in a registry or an index; 3 of the 22 sources its claims cite were checked for support.
Understanding Equivalence and Noninferiority Testing
Open accessAbstract
Increasingly, the goal of many studies is to determine if new therapies have equivalent or noninferior efficacies to the ones currently in use. These studies are called equivalence/noninferiority studies, and the statistical methods for their analysis require only simple modifications to the traditional hypotheses testing framework. Nevertheless, important and subtle issues arise with the application of such methods. This article describes the concepts and statistical methods involved in testing equivalence/noninferiority. The aim is to enable the clinician to understand and critically assess the growing number of articles utilizing such methods.
Overview
CoChat Summary
TL;DR
This article explains equivalence and noninferiority testing, emphasizing that these require reversing traditional hypotheses and defining an equivalence margin δ in advance. It reviews the two one-sided test (TOST) procedure, sample-size implications, and common pitfalls such as interpreting a nonsignificant difference test as evidence of equivalence.
Key findings
- Equivalence is established at significance level α when a (1−2α)×100% confidence interval for the difference in efficacies is contained within (−δ, δ); for α=0.05, this means using a 90% confidence interval. Noninferiority is established when the lower limit of that interval is above −δ when higher is better, or when the upper limit is below δ when lower is better.
- In the HIV example, the equivalence margin was δ=12 percentage points; response rates were 50.8% for abacavir and 51.3% for indinavir, with a reported 95% CI for the difference of (−9, 8), which lies within (−12, 12), so equivalence was declared at the 0.025 significance level.
- In OASIS-5, the noninferiority margin for relative risk was 1.185; observed risks were 5.8% for fondaparinux and 5.7% for enoxaparin [RR=1.01, 95% CI (0.90, 1.13)], and because the upper CI limit was below 1.185, noninferiority was established. For major bleeding, fondaparinux showed superiority [RR=0.52, 95% CI (0.44, 0.61)].
- Barker et al. found contradictory results between traditional two-sided tests and TOST in 9 of 21 comparisons using α=0.05 and an equivalence margin of 5 percentage points. Le Henanff identified 162 equivalence/noninferiority reports from 2003–2004, with about 80% not justifying their equivalence margin and 28% not considering the margin in sample-size calculation.
Methods
- This is a conceptual/methodological review or tutorial, not a primary empirical study. It describes hypotheses, the TOST procedure, equivalence margins, sample size, measures of effect, intention-to-treat versus per-protocol analysis, and reporting recommendations.
- It uses published trial examples—the abacavir-lamivudine-zidovudine versus indinavir-lamivudine-zidovudine HIV trial (n=562), OASIS-5, and Barker et al.’s National Immunization Survey vaccination-coverage comparisons—to illustrate equivalence and noninferiority testing.
- Sample-size illustrations were calculated with PASS 8; for true proportions of 0.28 and 0.33, achieving 0.80 power at α=0.05 required 26,185 per group for δ=0.06 and 535 per group for δ=0.12.
Limitations
- The paper does not state limitations of its own review; it states that the literature contains much confusion, lack of uniformity, and lack of transparency in terminology, and cites deficiencies in reporting of equivalence/noninferiority trials.
- Evident from the text: this is a narrative educational article with selected examples, not a systematic review or meta-analysis, and it reports no formal search strategy or primary data collection.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
3 Valid · 12 Not valid · 15 Not checked · 40 Own results and interpretations
Every source the paper cites for this claim was checked and supports it.
In the paper
“Using TOST, equivalence is established at the α significance level if a (1-2α) × 100% confidence interval for the difference in efficacies (newcurrent) is contained within the interval (-δ, δ) (Fig. 1).”
§ Understanding Equivalence and Noninferiority Testing > PROCEDURE
What the sources say
- Supportedcites [2] Schuirmann DJ. A comparison of the two one-sided tests procedure and the power approach for assessing equivalence of average bioavailability. J Pharmacokin Biopharm. 1987;15:657–680. source
In “The Two One-Sided Tests Procedure” (page unavailable), the paper states that TOST declares equivalence when both one-sided hypotheses are rejected at nominal level α and is operationally identical to requiring the ordinary 1−2α confidence interval to lie within the equivalence interval. The paper applies this rule to differences in average bioavailability; the citing sentence applies the same statistical rule to differences in efficacy. (LLM, conf=0.92)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The results show how small changes in the equivalence margin can cause large changes in the required sample size to achieve the same power.”
§ Understanding Equivalence and Noninferiority Testing > SAMPLE SIZE
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
In the paper
“The FDA requires the reporting of both types of analyses 11 .”
§ Understanding Equivalence and Noninferiority Testing > ANALYSIS OF DATA
What the sources say
- Not matchedcites [11] http://www.fda.gov/downloads/GuidanceComplianceRegulatoryInformation/Guidances/ucm073113.pdf . (Accessed 8/24/2010).
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Not matched
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Equivalence/noninferiority is clearly established when both ITT and PP analyses agree.”
§ Understanding Equivalence and Noninferiority Testing > ANALYSIS OF DATA
boostedstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: unclear — It is unclear why the paper says it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Using a traditional comparative test to establish equivalence/ noninferiority leads frequently to incorrect conclusions.”
§ Understanding Equivalence and Noninferiority Testing > NO DIFFERENCE DOES NOT IMPLY EQUIVALENCE
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
18 more claims
Create a free account to see all 70 claims and to verify the 1 reference still open.
- Probably not a substantial claim. 36
- It may state two separate claims as one. 19
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Understanding Equivalence and Noninferiority Testing > CONCLUSIONS
“Le Henanff 14 identified 162 reports of equivalence/noninferiority trials published in 2003 and 2004.”
§ Understanding Equivalence and Noninferiority Testing > References
“13. Le Henanff A, Girardeau B, Baron G, et al. Quality of reporting of noninferiority and equivalence randomized trials. JAMA. 2006;295:1147- 1151.”
Citations
Cites
Schuirmann, D. J. (1987) · Journal of Pharmacokinetics and Biopharmaceutics · doi:10.1007/bf01068419 · Cited by 2,696
Piaggio, G., Elbourne, D. R., Pocock, S. J., Evans, S. J. W., Altman, D. G., & CONSORT Group, F. T. (2012) · JAMA · doi:10.1001/jama.2012.87802 · Cited by 2,428
Comparison of Fondaparinux and Enoxaparin in Acute Coronary Syndromes
(2006) · New England Journal of Medicine · doi:10.1056/nejmoa055443 · Cited by 1,228
Reporting of Noninferiority and Equivalence Randomized Trials: An Extension of the CONSORT Statement
Piaggio, G., Elbourne, D. R., Altman, D. G., Pocock, S. J., Evans, S. J. W., & CONSORT Group, F. T. (2006) · JAMA · doi:10.1001/jama.295.10.1152 · Cited by 1,153
Temple, R., & Ellenberg, S. S. (2000) · Annals of Internal Medicine · doi:10.7326/0003-4819-133-6-200009190-00014 · Cited by 871
Ellenberg, S. S., & Temple, R. (2000) · Annals of Internal Medicine · doi:10.7326/0003-4819-133-6-200009190-00015 · Cited by 331
Testing Statistical Hypotheses of Equivalence
Wellek, S. (2002) · Chapman & Hall/CRC eBooks · doi:10.1201/9781420035964 · Cited by 318
Good Enough: A Primer on the Analysis and Interpretation of Noninferiority Trials
Kaul, S., & Diamond, G. A. (2006) · Annals of Internal Medicine · doi:10.7326/0003-4819-145-1-200607040-00011 · Cited by 272
Quality of Reporting of Noninferiority and Equivalence Randomized Trials
Le Henanff, A., Giraudeau, B., Baron, G., & Ravaud, P. (2006) · JAMA · doi:10.1001/jama.295.10.1147 · Cited by 263
Cited by
Rejecting or Accepting Parameter Values in Bayesian Estimation
Kruschke, J. K. (2018) · Advances in Methods and Practices in Psychological Science · doi:10.1177/2515245918771304 · Cited by 734
“Walker and Nowacki (2011) explained that one conventional setting of the noninferiority limit is at half of “the lower limit of a confidence interval of the difference between the current therapy and the placebo obtained from a metaanalysis” (p. 194).”
BackgroundÖst, L.-G., Havnen, A., Hansen, B., & Kvale, G. (2015) · Clinical Psychology Review · doi:10.1016/j.cpr.2015.06.003 · Cited by 655
Power analysis of single-cell RNA-sequencing experiments
Svensson, V., Natarajan, K. N., Ly, L.-H., Miragaia, R. J., Labalette, C., Macaulay, I. C., Cvejic, A., & Teichmann, S. A. (2017) · Nature Methods · doi:10.1038/nmeth.4220 · Cited by 568
The efficacy of Acceptance and Commitment Therapy: An updated systematic review and meta-analysis
Öst, L.-G. (2014) · Behaviour Research and Therapy · doi:10.1016/j.brat.2014.07.018 · Cited by 476
National Cluster-Randomized Trial of Duty-Hour Flexibility in Surgical Training
Bilimoria, K. Y., Chung, J. W., Hedges, L. V., Dahlke, A. R., Love, R., Cohen, M. E., Hoyt, D. B., Yang, A. D., Tarpley, J. L., Mellinger, J. D., et al. (2016) · New England Journal of Medicine · doi:10.1056/nejmoa1515724 · Cited by 460
Metadata
- Authors
- Walker, E., & Nowacki, A. S.
- Affiliations
- Cleveland Clinic (US)
- Corresponding author
- Walker, Esteban
- ORCID
- Nowacki, Amy S.
- Venue
- Journal of General Internal Medicine
- Published
- 2010-09-21
- Volume
- 26
- Issue
- 2
- Pages
- 192-196
- ISSN
- 0884-8734 (ISSN-L)
- 1525-1497
- PMC
- PMC3019319
- OpenAlex
- W2086898699
- Cited by
- 751
- References
- 16
- Citations by year
- 2026: 52 · 2025: 34 · 2024: 63 · 2023: 51 · 2022: 52 · 2021: 67 · 2020: 66 · 2019: 82 · 2018: 83 · 2017: 68 · 2016: 55 · 2015: 46 · 2014: 21 · 2013: 6 · 2012: 2
- FWCI
- 3.00
- Type
- journal article
- Language
- en
- Open access
- bronze
- doi.org
- Licence
- www.springer.com/tdm (tdm)
- Locations
- Journal of General Internal Medicine (publishedVersion)
- Europe PMC (PubMed Central) (publishedVersion)
- PubMed Central (publishedVersion)
- PubMed (publishedVersion)
- Topics
- Physical Sciences › Mathematics › Statistics and Probability › Statistical Methods in Clinical Trials
- Physical Sciences › Mathematics › Statistics and Probability › Statistical Methods in Epidemiology
- Physical Sciences › Mathematics › Statistics and Probability › Probability and Statistical Research
- Keywords
- equivalence testing, non-inferiority testing, hypothesis testing, clinical trial design, treatment efficacy comparison, statistical interpretation
- MeSH terms
- Humans, Research Design, Therapeutic Equivalency, Sample Size, Comprehension
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-11-03, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-11-03 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Understanding Equivalence and Noninferiority Testing. https://app.cochat.ai/papers/10.1007/s11606-010-1513-8
- The paper
- Walker, E., & Nowacki, A. S. (2011). Understanding Equivalence and Noninferiority Testing. Journal of General Internal Medicine. https://doi.org/10.1007/s11606-010-1513-8
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.