Examples
Worked examples
- Is an instance
A trial comparing two already-approved antihypertensive drugs, where the clinical question is neither 'is drug A better' nor 'is drug A not unacceptably worse' -- the protocol sets a symmetric equivalence margin (for example, plus or minus 4 mmHg systolic) and requires the 95% confidence interval for the between-arm difference to fall entirely inside that margin before the trial can conclude equivalence.
- Is an instance
A trial testing whether a reformulated version of an established therapy (a new delivery device or dosing schedule, not a generic copy) produces a clinically indistinguishable effect on the primary efficacy endpoint compared to the original formulation, using an equivalence margin agreed with the relevant regulatory authority before enrollment begins.
Counter-examples
Looks similar, but isn't
- Not an instance
A pharmacokinetic bioequivalence study comparing a generic drug's rate and extent of absorption (AUC, Cmax) to the reference product. Despite the similar name, FDA bioequivalence studies use a distinct, narrower statistical framework: the entire 90% confidence interval for the test-to-reference ratio of AUC and Cmax must fall within 80.00%-125.00%. That is a fixed regulatory rule for pharmacokinetic parameters, not the general clinical equivalence-margin approach described here, and the two terms should not be used interchangeably.
- Not an instance
A trial designed to show a new drug outperforms placebo on the primary endpoint. This is a superiority trial -- the underlying hypothesis and the direction of the statistical test are the opposite of an equivalence trial's, even though both designs report a confidence interval around the treatment difference.
Editorial commentary
Before a comparative clinical trial enrolls a single participant, its protocol has to commit to one of three possible questions: is the new intervention better, is it not unacceptably worse, or is it close enough in either direction that the two are considered clinically the same? An equivalence trial is the design used to answer the third question. Getting this distinction right at protocol review matters well beyond biostatistics — it drives sample size (and therefore budget), the shape of the statistical analysis plan, and how a research administrator should read the trial’s own claim of a positive result.
The three-way distinction: superiority, non-inferiority, equivalence
| Design | Question it answers | Margin used | Typical use case |
|---|---|---|---|
| Superiority | Is the new arm better than the comparator (or placebo)? | None — any positive difference favoring the new arm counts | New drug vs. placebo; new drug vs. standard of care where an improvement is the goal |
| Non-inferiority | Is the new arm not unacceptably worse than an active comparator? | One-sided lower margin only | A new treatment offering a real non-efficacy advantage (safety, cost, convenience, access) where matching — not beating — the standard of care is an acceptable outcome |
| Equivalence | Are the two arms close enough, in both directions, to be considered clinically the same? | Two-sided symmetric margin (-delta, +delta) | Comparing two already-established treatments; reformulations; situations where neither arm is expected or intended to outperform the other |
All three share the same basic statistical machinery — estimate a treatment difference and build a confidence interval around it — but the design choice changes which side of that interval has to clear which threshold, and it has to be fixed in the protocol before the trial starts. Switching the framing after unblinding (for example, reinterpreting a failed superiority trial as an equivalence success) is a well-documented methodological problem, not a legitimate post hoc option.
What an ‘equivalence margin’ means, without the biostatistics deep dive
The equivalence margin (often written delta, or -delta to +delta for the two-sided case) is the largest difference between arms that the trial’s investigators and reviewers agree, in advance, is small enough not to matter clinically. Two things about it matter for anyone reviewing a protocol rather than running the analysis:
- It is a clinical judgment dressed in statistical language, not a purely statistical output. The margin has to be justified by prior evidence on the comparator’s effect size and by what a clinician or patient would consider a meaningful difference — it is not simply whatever value the available sample size happens to be powered to detect.
- It must be prespecified. Because an equivalence trial requires the full confidence interval to land inside the margin, a margin chosen or adjusted after seeing unblinded data would make the equivalence conclusion circular. Regulatory guidance and methodological standards (see below) treat prespecification of the margin as a core requirement, not a formality.
One practical consequence worth flagging in protocol or budget review: because an equivalence trial’s confidence interval has to clear a margin on both sides, equivalence trials generally require larger sample sizes than an equivalently powered non-inferiority trial testing the same comparator, which only has to clear one side.
Not the same as bioequivalence
Equivalence trial and bioequivalence study are frequently confused, including in casual usage, but they are distinct regulatory concepts. A bioequivalence study (used for generic drug approval) compares pharmacokinetic parameters — typically AUC and Cmax — between a generic and reference product, using a fixed rule: the entire 90% confidence interval for the test-to-reference ratio must fall within 80.00% to 125.00%. That is a narrower, PK-specific regulatory standard, not the general clinical equivalence-margin framework described on this page, which can apply to any efficacy or safety endpoint and uses a margin set on clinical grounds rather than a fixed regulatory percentage band.
Why this distinction matters for research administrators
- Protocol and budget review: confirm the trial design (superiority, non-inferiority, or equivalence) matches the stated study objective, and that the sample size justification in the protocol is consistent with the chosen design — an equivalence design mislabeled as non-inferiority (or vice versa) usually means the wrong margin, and therefore the wrong sample size, was used.
- Statistical analysis plan (SAP) alignment: the SAP should state the margin, the confidence level, the analysis population (equivalence and non-inferiority trials commonly rely on both intention-to-treat and per-protocol analyses, since either alone can bias the result toward a false-positive equivalence conclusion — see Intent-to-Treat vs. Per-Protocol Analysis), and how the margin was justified, before the trial reports results.
- Ethics/IRB review: an equivalence trial’s justification for exposing participants to a treatment not expected to outperform the comparator rests on the real value of demonstrating equivalence itself (supporting a new formulation, a lower-cost option, or expanded access) — that rationale should be explicit in the protocol.
- Reporting and publication: a properly reported equivalence trial should follow the CONSORT extension for non-inferiority and equivalence trials, which requires the margin, its justification, and both ITT and per-protocol results to be reported transparently.
Related standards and guidance
- ICH E9, Statistical Principles for Clinical Trials, sets out the equivalence/non-inferiority/superiority framework referenced throughout this page and is the primary international methodological reference.
- FDA’s guidance Non-Inferiority Clinical Trials to Establish Effectiveness discusses the relationship between non-inferiority and equivalence designs and the prespecification requirement for margins in FDA submissions.
- The CONSORT extension for reporting non-inferiority and equivalence randomized trials sets the reporting standard referenced above; see CONSORT 2010 for the parent reporting standard it extends.
Related terms
- Randomized Controlled Trial (RCT)
- Intent-to-Treat (ITT) vs. Per-Protocol Analysis
- Placebo-Controlled Study Design
- CONSORT 2010
Frequently asked questions
Is an equivalence trial the same as a non-inferiority trial?
No. A non-inferiority trial only has to rule out the new arm being unacceptably worse (a one-sided lower margin); an equivalence trial has to rule out meaningful difference in either direction (a symmetric two-sided margin), which is a stricter requirement and generally needs a larger sample size.
Who decides the equivalence margin?
The trial’s investigators and biostatisticians propose it based on prior evidence about the comparator’s effect and what difference would be clinically meaningful, and for regulated trials it is typically discussed with or set in consultation with the relevant regulatory authority before the trial starts. It cannot be adjusted after unblinding without undermining the trial’s conclusions.
Does an equivalence trial need a placebo arm?
Not necessarily. Equivalence trials most often compare two active treatments (or a reformulation against the original), so a placebo arm is frequently absent; see Placebo-Controlled Study Design for when a placebo arm is and is not ethically and scientifically appropriate.
See also the Clinical Research cluster overview for related trial-design and regulatory-compliance topics.
Machine-readable encodings
Use in your systems
<role vocab="credit"
vocab-identifier="https://casrai.org/dictionary/"
vocab-term="Equivalence Trial"
vocab-term-identifier="https://casrai.org/dictionary/term/equivalence-trial" />{
"@context": "https://schema.org",
"@type": "DefinedTerm",
"@id": "https://casrai.org/dictionary/term/equivalence-trial",
"name": "Equivalence Trial",
"identifier": "https://casrai.org/dictionary/term/equivalence-trial",
"description": "An equivalence trial is a clinical trial designed to show that a new intervention's effect is neither meaningfully better nor meaningfully worse than an active comparator -- the treatment difference must fall within a pre-specified equivalence margin in both directions. It is one of three ways a comparative trial's primary hypothesis can be framed at the protocol design stage: a superiority trial sets out to show one arm is better than another (or than placebo); a non-inferiority trial sets out to show a new arm is not unacceptably worse than an active comparator, using only a single lower margin; an equivalence trial requires the treatment difference's confidence interval to fall entirely within a symmetric two-sided margin, written (-delta, +delta), that was fixed before the trial started as the largest difference still considered clinically unimportant. Which of the three applies is a design decision documented in the protocol and statistical analysis plan -- it is not something inferred after the fact from a trial's results.",
"inDefinedTermSet": "https://casrai.org/dictionary/domain/clinical-research#set",
"url": "https://casrai.org/dictionary/term/equivalence-trial",
"sameAs": [],
"license": "https://creativecommons.org/licenses/by/4.0/",
"publisher": {
"@id": "https://casrai.org/#organization"
},
"dateModified": "2026-07-23T04:57:25",
"inLanguage": "en"
}
