The three pillars of equivalence under MDR are technical characteristics, biological characteristics, and clinical characteristics. Annex XIV Part A Section 3 of Regulation (EU) 2017/745 requires all three to be demonstrated in full for an equivalence claim to hold. MDCG 2020-5 (April 2020) translates each pillar into specific, feature-by-feature criteria. A strong argument on two pillars does not compensate for a weak argument on the third — the pillars are independent gates, and a claim that fails on any one of them fails entirely.

By Tibor Zechmeister and Felix Lenhard. Last updated 10 April 2026.

TL;DR

  • The three pillars of equivalence under MDR are defined in Annex XIV Part A Section 3 of Regulation (EU) 2017/745: technical, biological, and clinical characteristics.
  • MDCG 2020-5 (April 2020) operationalises each pillar into specific criteria that must be compared feature by feature between the subject device and the predicate.
  • Technical characteristics cover design, specifications, properties, deployment, principles of operation, and conditions of use.
  • Biological characteristics cover materials in contact with the body, the tissue and duration of contact, and release characteristics.
  • Clinical characteristics cover the clinical condition, severity, site, population, user profile, and performance for a specific intended purpose.
  • All three pillars are independent gates. Failing any one pillar fails the equivalence claim, regardless of how strong the other two are.

Why one pillar is never enough

Equivalence is the single highest-leverage subtraction move available in an MDR clinical evaluation. If the claim holds, the clinical data of an already-marketed device becomes your clinical evidence, and the expensive new clinical investigation drops out of the plan. The stakes are why the MDR defines equivalence narrowly and why Annex XIV Part A Section 3 lists three independent characteristic groups instead of a single judgement call.

The pillars are independent for a reason. Two devices can look mechanically identical and still behave differently in the body because the materials differ. Two devices can share materials and geometry and still be clinically different because they are used in different patient populations. Two devices can be clinically identical for the same condition and still fail technical equivalence because their principles of operation diverge. Collapsing the three into one overall judgement hides exactly the failure modes the regulation is trying to catch.

The frame this post uses is simple. Each pillar is a gate. Every gate must open. A manufacturer who has built a strong technical and clinical case but a weak biological case has not built an equivalence argument — they have built two-thirds of one, and two-thirds is zero in this context.

What Annex XIV Part A Section 3 actually says

Annex XIV Part A of Regulation (EU) 2017/745 contains the clinical evaluation procedure. Section 3 of Part A is the section that defines equivalence. It lists the three characteristic groups and sets out the criteria for each. The relevant text is precise:

"Technical: the device is of similar design; is used under similar conditions of use; has similar specifications and properties including physicochemical properties such as intensity of energy, tensile strength, viscosity, surface characteristics, wavelength and software algorithms; uses similar deployment methods, where relevant; has similar principles of operation and critical performance requirements." — Regulation (EU) 2017/745, Annex XIV Part A Section 3.

"Biological: the device uses the same materials or substances in contact with the same human tissues or body fluids for a similar kind and duration of contact and similar release characteristics of substances, including degradation products and leachables." — Regulation (EU) 2017/745, Annex XIV Part A Section 3.

"Clinical: the device is used for the same clinical condition or purpose, including similar severity and stage of disease, at the same site in the body, in a similar population, including as regards age, anatomy and physiology; has the same kind of user; has similar relevant critical performance in view of the expected clinical effect for a specific intended purpose." — Regulation (EU) 2017/745, Annex XIV Part A Section 3.

The language is specific. It is not "broadly similar." It is not "same category." It is an enumerated list of criteria, each of which must be addressed. MDCG 2020-5 then turns each criterion into an operational expectation for how it should be evidenced.

Pillar one — technical characteristics

Technical equivalence is usually the pillar that manufacturers reach for first, because it is the most tangible. The criteria in Annex XIV Part A Section 3 cover design, conditions of use, specifications and physicochemical properties, deployment method, principles of operation, and critical performance requirements. MDCG 2020-5 expects each of these to be documented for the subject device and the predicate and compared directly.

For a physical device, this means dimensions, geometry, materials of construction, mechanical properties, surface treatment, sterilisation method and its effect on the material, and the intended use environment. For an active device it additionally means energy source, output parameters, and any programmable behaviour. For software it means the algorithms, the input data types, the output formats, the user interface for clinical decisions, and the underlying model or rule set — the Annex text explicitly names "software algorithms" as a specification.

The trap in the technical pillar is not missing a criterion, it is treating "similar" as a synonym for "close enough." MDCG 2020-5 requires that every difference be identified and that each difference be justified against potential impact on safety and performance. A small difference that does not affect either is acceptable; a small difference that does affect either is not. The burden of proof sits with the manufacturer. Assertion is not justification. Evidence is.

The second trap is scope. Teams compare against the product family instead of against a specific version of a specific device. Annex XIV Part A is about device-level equivalence, not category-level equivalence. The comparison must name a specific predicate, identify its regulatory status and version, and demonstrate equivalence against that exact device.

Pillar two — biological characteristics

Biological equivalence is where the largest share of equivalence claims break, and it is the pillar startups most often underestimate. The criterion is strict: the same materials or substances in contact with the same human tissues or body fluids, for a similar kind and duration of contact, with similar release characteristics including degradation products and leachables.

"Same materials" is doing a lot of work in that sentence. A polymer with the same nominal chemical name from a different supplier is not automatically the same material, because manufacturing processes, additives, residuals, sterilisation compatibility, and surface finish can all change the biological behaviour of a nominally identical material. The ISO 10993 series exists because biological safety is determined by what the body actually encounters, not by what the material is called on the datasheet.

The access problem is acute in this pillar. To demonstrate that your polymer is biologically equivalent to a competitor's polymer, you need the competitor's exact composition, processing information, and extractables/leachables profile. You generally do not have that access for a competitor device, and public information is rarely sufficient to close the gap. This is why equivalence to a competitor's implantable device is almost never workable in practice, even before the Article 61(5) data access contract requirement is considered.

Duration and type of tissue contact matter as much as the material itself. A material that is biologically acceptable for a 24-hour surface contact is not automatically acceptable for a 30-day implant in soft tissue. The predicate and the subject must match on both dimensions, and the release characteristics — what leaches, how much, over what time — must match too. MDCG 2020-5 treats the biological pillar as a technical, evidence-based comparison, not a narrative.

Pillar three — clinical characteristics

Clinical equivalence is the pillar most often skipped over in a single paragraph when it deserves the same depth as the other two. The criteria are the clinical condition or purpose, the severity and stage of disease, the site in the body, the population including age, anatomy and physiology, the user profile, and the critical performance for a specific intended purpose.

The temptation is to collapse clinical equivalence into "treats the same disease." The regulation does not allow that simplification. A device used for the same condition but in a different population — adults instead of paediatrics, chronic instead of acute presentation, general practitioners instead of trained specialists — is not clinically equivalent. A device with the same technology used at a different anatomical site is not clinically equivalent. A device with a different intended purpose inside the same broad condition is not clinically equivalent.

The user profile criterion in particular is frequently missed. An AI-assisted diagnostic tool intended for use by trained radiologists is not clinically equivalent to a tool intended for use by general practitioners, even if the underlying algorithm is the same, because the clinical effect depends on how the output is interpreted and acted on. MDCG 2020-5 expects the user profile to be explicit and compared directly.

Clinical characteristics are where the intended purpose statement written in the technical documentation meets the equivalence claim. If the two are inconsistent — if the claimed equivalence implies a clinical context different from the stated intended purpose — the inconsistency will surface in conformity assessment. The pillar is as technical as the others; it just looks softer because the criteria are human rather than mechanical.

How all three must hold at the same time

MDCG 2020-5 is explicit that the three pillars are independent and that equivalence requires all three to be demonstrated in full. This is the single point where MDR practice most sharply diverges from the MEDDEV 2.7/1 rev 4 tradition, where a strong argument on two pillars could sometimes carry a weaker argument on the third. Under MDR and MDCG 2020-5, that averaging is not acceptable.

The practical consequence is that the equivalence demonstration reads as three separate demonstrations, each with its own feature-by-feature comparison table, each with its own list of differences and evidence-based justifications, each with its own conclusion. The overall conclusion is a logical AND of the three individual conclusions. If any one pillar is "not demonstrated," the overall claim is "not demonstrated," and no amount of additional work on the other two pillars can change that.

This is not a bureaucratic quirk. It reflects how medical devices actually fail. A device can be technically sound, clinically appropriate, and biologically hazardous because of a material choice. A device can be technically identical, biologically identical, and clinically wrong because the population or user is different. The regulation separates the pillars because the failure modes are separable, and treating them as independent forces the analysis to surface the failure that a combined judgement would bury.

What MDCG 2020-5 adds to the Annex text

MDCG 2020-5 is the April 2020 guidance from the Medical Device Coordination Group that translates Annex XIV Part A Section 3 into operational expectations. The guidance is short but dense, and the key contributions are four.

First, MDCG 2020-5 requires feature-by-feature comparison. The expectation is a structured, tabular presentation of the subject device and the predicate against every criterion in the three pillars, not a narrative claim. Narrative-only equivalence arguments are no longer acceptable.

Second, the guidance requires that every difference between subject and predicate be identified and justified. The justification must be evidence-based — published literature, bench testing, risk analysis, not assertion. A difference that is acknowledged and unjustified is equivalent to an acknowledged failure on that criterion.

Third, MDCG 2020-5 is explicit about the distinction between equivalence and the use of data from similar but non-equivalent devices. Similar-device data has a legitimate place in clinical evaluation — in risk analysis, post-market context, and state-of-the-art assessment — but it is not clinical evidence for the subject device and cannot be used as such. The two must not be mixed in the CER.

Fourth, the guidance reinforces that even a valid equivalence claim does not exempt the device from demonstrating that it meets the current state of the art. If the state of the art has moved since the predicate was certified, additional evidence is required on top of the equivalence demonstration.

Common failures across the three pillars

  • Addressing only the technical pillar. The most common failure mode. The technical table is detailed, the biological pillar is a paragraph, the clinical pillar is one sentence. MDCG 2020-5 explicitly rejects this structure.
  • Confusing "same category" with equivalence. Comparing against a product family instead of a specific named predicate device. Annex XIV Part A requires a specific predicate.
  • Nominal-material biological claims. Asserting biological equivalence because the polymer has the same ISO name, without accounting for supplier, processing, additives, or surface finish.
  • Skipping the user profile in the clinical pillar. Treating user profile as a soft parameter when MDCG 2020-5 treats it as a hard criterion.
  • Narrative instead of tabular comparison. Writing the equivalence section as prose rather than a feature-by-feature table.
  • Unjustified differences. Acknowledging differences without evidence that they do not affect safety or performance.
  • Ignoring the state of the art. A valid equivalence claim to an older predicate without checking whether state-of-the-art evidence has moved on.

The Subtract to Ship angle

The three-pillar discipline looks like more work, but it is the shortest path to the honest answer. The fastest way to discover that an equivalence claim will not survive conformity assessment is to run it against all three pillars at the MDCG 2020-5 level of detail early, before the CER is written. If one pillar fails, the claim fails, and the team can pivot to a targeted clinical investigation with the timeline and budget intact.

The Subtract to Ship framework (see post 65) is built on the principle that every subtraction must be defensible against the regulation itself. Equivalence is the textbook case. A subtraction that cannot be defended is not a subtraction — it is a delayed cost. The three-pillar check is the test that tells you which you have. Run it honestly, run it adversarially, and accept the answer the pillars give you. If all three hold, you have subtracted a clinical investigation. If one fails, you have learned it now instead of after six months of CER drafting, and that too is a subtraction — of wasted work.

Reality Check — Where do you stand?

  1. Have you identified a specific named predicate device — manufacturer, model, version, regulatory status — rather than a product family?
  2. Can you produce a feature-by-feature comparison table for the technical pillar covering design, specifications, properties, deployment, principles of operation, and critical performance?
  3. Can you produce a feature-by-feature comparison for the biological pillar covering materials, tissue contact, duration, and release characteristics — with evidence for any "same material" claim beyond the datasheet name?
  4. Can you produce a feature-by-feature comparison for the clinical pillar covering condition, severity, site, population, user profile, and critical clinical performance?
  5. Have you identified every difference on every criterion and produced an evidence-based justification for each?
  6. Would your equivalence claim survive if all three pillars were read independently by a sceptical Notified Body reviewer?
  7. If any one pillar fails, do you have a backup plan that does not break the project timeline?

Frequently Asked Questions

What are the three pillars of equivalence under MDR? The three pillars are technical characteristics, biological characteristics, and clinical characteristics, as defined in Annex XIV Part A Section 3 of Regulation (EU) 2017/745. All three must be demonstrated in full for an equivalence claim to hold, and MDCG 2020-5 (April 2020) specifies how each pillar is to be evidenced.

Can I claim equivalence if two of the three pillars are strong and the third is weak? No. Under MDCG 2020-5 the three pillars are treated as independent gates. A weak argument on one pillar is not compensated by strong arguments on the other two. Equivalence is demonstrated only if all three pillars hold.

What is the biggest difference between the technical and biological pillars? The technical pillar is about how the device is designed and operates. The biological pillar is about what the device is made of in the places where it touches the patient, and how the body responds. A device can be technically identical to a predicate and still fail biological equivalence if the materials, supplier, processing, or duration of tissue contact differ.

Does the clinical pillar just mean "treats the same disease"? No. Clinical equivalence under Annex XIV Part A Section 3 requires matching on clinical condition, severity and stage of disease, anatomical site, population including age and anatomy and physiology, user profile, and critical clinical performance for a specific intended purpose. "Same disease" is only one criterion among several.

How detailed does the comparison need to be? MDCG 2020-5 expects a feature-by-feature comparison, normally presented as a table, with every criterion addressed for both subject and predicate, every difference identified, and every difference justified with evidence. Narrative claims without feature-level comparison are not acceptable.

Sources

  1. Regulation (EU) 2017/745 of the European Parliament and of the Council of 5 April 2017 on medical devices, Article 61 (clinical evaluation) and Annex XIV Part A Section 3 (demonstration of equivalence). Official Journal L 117, 5.5.2017.
  2. MDCG 2020-5 — Clinical Evaluation — Equivalence: A guide for manufacturers and notified bodies, April 2020.

This post is part of the Clinical Evaluation & Clinical Investigations series in the Subtract to Ship: MDR blog. Authored by Felix Lenhard and Tibor Zechmeister. The three-pillar discipline is the test that tells you early whether an equivalence claim is real or wishful — run it honestly, and accept the answer the pillars give you.