Same-Variety Comparison for NMPA Substantial Equivalence

1. Key takeaways

Comparison with a same-variety device is an evaluation activity built from three components, not a single table, and its outcome depends more on the comparator than on the drafting.

• It is an evaluation, not a table. Under Article 36 of Order 47 (SAMR Order No. 47, promulgated 26 August 2021, effective 1 October 2021) the file has three parts: the comparison, the analysis and evaluation of same-variety clinical data, and the scientific evidence and conclusion for differences.
• Choice of comparator decides the outcome. It must be the same variety, with clinical data that can lawfully be cited and sufficiently analysed; a similar name is not enough.
• Differences must be identified across five dimensions: working principle and structure, materials and key components, performance and test methods, intended use and contraindications, method of use and environment.
• Absence of evidence for a difference is a frequent deficiency ground. A difference with no data, or no reasoning link between data and conclusion, will be sent back.
• Three evidence sources exist and cannot replace one another: published clinical literature, post-market clinical data for the same variety, and non-clinical research data under Article 28 of the order.
• The correction clock is hard. Under Article 41 deficiencies are notified in full at one time and must be answered within one year, failing which review is terminated without approval.

2. Legal basis

Regulations on the Supervision and Administration of Medical Devices (State Council Decree No. 739, promulgated 9 February 2021, effective 1 June 2021)

• Article 14: seven categories of registration and filing material, the fourth being clinical evaluation material.
• Article 24: the two situations exempt from clinical evaluation.
• Article 25: the two routes, one of which is comparative analysis of same-variety clinical data.

Provisions for the Administration of Medical Device Registration and Filing (SAMR Order No. 47, promulgated 26 August 2021, effective 1 October 2021)

• Article 17: materials must be in Chinese with translations attached, and unpublished literature requires a licence from the rights holder.
• Article 28: scope of non-clinical research: physicochemical, electrical safety, radiation safety, software, biological characteristics, biological material safety, sterilisation, animal studies and stability.
• Article 29: requirements for method validation and statistical analysis.
• Articles 33, 34, 35: principles of clinical evaluation, the two exemption situations and catalogue authority, and clinical evaluation guidelines.
• Article 36: the material required under each route.
• Article 41: all deficiencies notified at one time, answered within one year, otherwise review is terminated and approval refused.

Normative and technical documents

• Catalogue of Medical Devices Exempt from Clinical Evaluation (2025) (NMPA Notice No. 19 of 2025, published 12 May 2025, effective on publication): 1,047 items, 28 added, 25 descriptions revised.
• Announcement on Registration Application Materials and Approval Document Format (NMPA Announcement No. 121 of 2021, published 29 September 2021, effective 1 January 2022).
• Fees: under the NMPA licensing fee catalogue, a high-risk device trial application costs RMB 43,200 and a domestic Class III initial registration RMB 153,600.

Practical estimate. No official figure exists for retrieval channel costs, the drafting cycle for comparison material, or provincial Class II fees. Some provinces publish RMB 40,670 per registration unit; the current rate must be verified with the provincial medical products administration and finance department at project initiation.

3. Expert insights

Decide same-variety first, comparison second. Failures fall into two groups: a wrong comparator and insufficient difference evidence. The first is more damaging, because everything built on it loses its foundation. Three questions settle status: do both share a working principle and structural type; are main performance indicators comparable; do intended use and setting overlap. A negative answer means a new comparator, a second comparator for partial comparison, or a trial. Forcing a near-equivalent into the category inflates the list until no evidence can cover it.

The difference list is built on five dimensions and must allow a not-determinable result.

Each item is marked consistent, different or not determinable. A not-determinable entry may appear but not be left blank: it needs a reason (missing data, unsuitable comparator) and a next step.

Each of the three evidence sources has its own boundary.

Literature and post-market data support clinical performance directly; non-clinical data show a difference adds no new clinical risk. Using the latter to replace clinical evidence for a performance claim is common and ineffective.

Deficiencies for missing difference evidence take three forms. The first identifies differences without evidence: ten are listed, three are supported, and the rest are dismissed as insignificant. The second supplies data without an argument: twenty papers are attached, nothing explains how they support the conclusion, and the conclusion merely restates the table. The third mismatches evidence and difference, using data for material A to justify material B, or in vitro data to justify in vivo behaviour. Each difference should follow a four-part structure of description, clinical impact analysis, evidence citation and conclusion.

Data legality and methodological quality are two separate thresholds. Article 17 sets the legal threshold: Chinese materials, translations attached, and a licence for unpublished literature, so downloading a paper is not the same as citing it. Article 29 sets the second: a small sample, unsuitable statistics or short follow-up will not support the conclusion even where citation is lawful. The two are checked separately, in a data legality table covering source, licence and translation, and a methodological table covering design, sample size, statistics, follow-up and limitations.

Switching route needs criteria, not instinct. Abandoning comparison for a trial is a major decision. Three criteria are used together.

A switch is advised only where all three hold; moving because review might ask questions pushes the project into the most expensive channel.

4. Practical recommendations

• Run the three comparator questions first, covering principle, performance comparability, and overlap of intended use and setting, before drafting.
• Build the five-dimension list, marking each item consistent, different or not determinable, with a reason and next step for the third.
• Argue every difference in four parts: description, clinical impact, evidence, conclusion.
• Keep two verification tables, one for data legality and one for methodological quality.
• State the purpose of non-clinical evidence precisely: it shows a difference adds no new clinical risk, not clinical performance.
• Schedule corrections from the one-year limit, tracking material, owner, source and days weekly to avoid termination under Article 41.
• Assess any switch against the three criteria before committing to a trial.
• Fix language and format compliance early, completing Chinese translations and rights holder licences under Article 17.

5. Verification checklist

• □ The comparator passed the three questions on principle, performance and use.
• □ Name, certificate number and marketing status are recorded.
• □ The difference list covers five dimensions at judgeable granularity.
• □ Each item is marked consistent, different or not determinable, with reason and next step.
• □ Every difference follows the four-part argument.
• □ Each evidence type is labelled by source in the list.
• □ The legality table confirms sources and any licence for unpublished work.
• □ The methodology table records design, sample size, statistics, follow-up and limitations.
• □ Non-clinical data cover the applicable items in Article 28.
• □ The analysis reaches a conclusion rather than listing references.
• □ The three switch criteria have been assessed and documented.
• □ The response plan follows the one-year limit (Article 41); fees follow the NMPA catalogue.

How GRZAN can help. GRZAN provides medical device and IVD registration, quality system, authorised representative and local testing support for the Chinese market. Our team manages NMPA Class II/III registrations and filing procedures, coordinates typetesting, and supports post-market vigilance obligations. Business enquiries: info@grzanglobal.com