Non-Clinical Study Data for China Medical Device Registration

1. Key takeaways

Article 28 of the Medical Device Registration and Filing Measures (SAMR Order No. 47, published 26 August 2021, effective 1 October 2021) sets the scope of non-clinical research as nine study types: physicochemical performance, electrical safety, radiation safety, software, biological characteristics, safety of biological source materials, sterilisation and disinfection processes, animal studies, and stability.

• The nine are a topic list, not a mandatory checklist: whether a study is needed follows from design requirements and risk management, so Article 28 is risk-driven.
• Article 29 sets two pillars, method validation and statistical analysis; a comprehensive study remains unconvincing if the method is unvalidated or the sample size unjustified.
• The representativeness principle in Article 31 applies to every study: samples must represent the safety and effectiveness of the product applied for.
• Non-clinical work, technical requirements, test reports and clinical evaluation form the evidence set: the first three address conformity with specifications, the last benefit-risk in humans.
• GRZAN’s assessment: failures cluster in three places — studies mismatched to product risks, missing method validation, samples differing in origin from the applied product — all diagnosable early.

2. Legal basis

Administrative regulation. The Regulation (State Council Order No. 739, published 9 February 2021, effective 1 June 2021) provides:

• Article 5: supervision follows risk management, whole-process control, scientific oversight and social co-governance.
• Article 14: seven categories, including risk analysis, technical requirements, test report and clinical evaluation; the report may be a self-test or from a qualified institution.
• Article 24: two situations exempt from clinical evaluation. Article 25: two routes for clinical evaluation.

Departmental rule. Order No. 47 specifies:

• Article 19: applicable mandatory standards met; inapplicable ones explained with materials.
• Article 26: the technical requirements state objectively determinable functional and safety specifications and test methods.
• Article 28: non-clinical research completed before applying, covering the nine types above.
• Article 29: method validation required; where statistics are involved, the statistical method, sample size basis and acceptance criteria must be specified.
• Article 30: testing against the requirements must pass before trials, registration or filing.
• Article 31: test samples must represent the product applied for.
• Article 32: the report may be a self-test or from a qualified institution.
• Articles 33 and 34: clinical evaluation as the principle, with two exemptions.
• Article 41: deficiencies notified once in full, answered once within one year, otherwise review terminates with a refusal.
• Article 121: IVDs follow the separate IVD measures.

Normative documents. Announcement No. 121 places non-clinical materials in the catalogue; No. 126 governs self-testing, with the guide of 16 September 2025; the QMS inspection guide (notice No. 50 of 2022) covers on-site inspection; No. 63 of 2025 clarifies review requirements.

Practical estimate: depth per study type, sample selection and accelerated versus real-time ageing combinations have no clause-level standard; current guidelines, review points and mandatory standards govern. Map study items to regulatory sources before running studies.

3. Expert insights

A topic list, not a checklist. Article 28 anchors the decision in design requirements and risk management. The table frames the trigger test.

A decision not to study needs a reason traceable to a risk management output; a blank is an absence of record, not an exemption.

Method validation is the entry ticket. Article 29 puts validation first, targeting a habit: adopting a standard method without confirming it suits the product. Validation answers four questions. Does the method suit the matrix, range and interference profile? Do precision, accuracy and detection limit meet the decision needs? Was a standard method deviated from, and why? Does a non-standard method have a complete confirmation record? Disputes rarely concern whether a study was done but whether the numbers can be trusted.

Statistics means prior specification. Article 29 requires the statistical method, the basis for sample size and the acceptance criteria, all protocol-stage decisions. The method covers superiority, non-inferiority or equivalence testing, parametric or non-parametric choice, and missing data and outliers. The sample size basis means the effect size, significance level and power used, not a habit of three groups. The criteria define what passes and where the threshold comes from. An explanation written afterwards is weaker than a calculation fixed beforehand, as in active device performance verification, algorithm confirmation and animal efficacy evaluation.

Representativeness cuts across all studies. Article 31 addresses test articles, but the principle applies everywhere. Three breaks recur: engineering prototypes whose materials, processes or suppliers differ from production; one model in the unit without typical-model justification; samples whose sterilisation batch or ageing state differs from the marketed state. A sample traceability matrix, one row per report recording identifier, batch, differences from the marketed state and typicality justification, answers a representativeness question in one response.

Division of labour. Mixing evidence types causes duplicated spend or missing evidence.

Non-clinical research is the middle layer, supporting the specifications upward and giving clinical evaluation a mechanistic and safety foundation downward; without it, a fully passing test report can still attract further study requests.

Correction costs differ sharply. Under Article 41 all deficiencies are notified once and answered within a year, but repair ranges from weeks for a missing protocol or statistical report, through months for missing method validation, to quarters for remade samples and repeated studies, and half a year or more for an added study type including ethics and execution. The inception assessment is therefore the only low-cost chance to correct course.

4. Practical recommendations

• Complete the nine-type applicability assessment at inception, recording omissions with reasons in the risk file.
• Write the protocol first, covering purpose, method, sample size basis and criteria, and review it internally.
• Keep method validation standalone, covering applicability, precision, accuracy, detection limit and deviations.
• Fix the statistical design in advance and retain the records.
• Maintain a sample traceability matrix per report, including batch and differences from production.
• Verify the currency of every cited standard (Article 19).
• For self-testing, prepare capability and system documents under Announcement No. 126.
• Archive protocol, raw records, statistical output and report to support QMS inspection.

5. Verification checklist

• □ Studies completed before applying (Article 28)?
• □ Applicability judged for each type, omissions traceable to the risk file?
• □ Each study protocol covers purpose, method, sample size and criteria?
• □ Method validation done, deviations and non-standard methods recorded (Article 29)?
• □ Where statistics apply, method, sample size basis and criteria specified?
• □ Samples from a production-equivalent state, with a traceability matrix (Article 31)?
• □ Typical model justified where the unit has several models?
• □ Name, models and material grades consistent across documents?
• □ Cited standards current; inapplicable mandatory standards explained (Article 19)?
• □ Self-testing compliant with Announcement No. 126 and the guide?
• □ Conclusions support the specifications set?
• □ Testing completed with a pass and report obtained (Articles 30 and 32)?
• □ Protocol, raw records, statistical output and report archived?
• □ Correction answered once within one year (Article 41)?

Regulatory information verified as at 29 October 2026. Legal texts follow the versions published by NMPA, SAMR and gov.cn and must be rechecked after amendment or repeal. Time limits, fees and catalogue scope reflect the official position on that date; content marked practical estimate has no legal effect.

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 prepares NMPA-compliant dossiers, coordinates typetesting and clinical evaluation documents, and supports post-market vigilance obligations. Business enquiries: info@grzanglobal.com