NMPA Registration for IVDs: How Classification Shapes Submission Requirements

For an in vitro diagnostic entering the Chinese market, risk classification determines the regulatory route and the evidence needed to support it. Class I products follow a filing route; Class II and Class III products require registration. Classification also affects how analytical performance, clinical evidence, product specifications, and manufacturing controls must be assembled into a defensible submission.

A technical evaluator should establish the proposed intended use before judging registration readiness. The same analytical technology can support different regulatory conclusions depending on the analyte, specimen, target population, and clinical decision it informs. A dossier built around the wrong classification assumption can contain extensive testing yet still miss the evidence required for the claimed use.

Classification starts with the clinical claim

China’s IVD classification framework distinguishes products by risk, with detailed classification rules and a classification catalogue guiding the assignment. For a specific reagent, the relevant catalogue entry and its scope are more useful than a broad description such as “molecular diagnostic” or “clinical chemistry test.” Technology alone does not establish the class.

Start with a precise intended-use statement: what the product detects or measures, which specimens it accepts, who will be tested, and how the result will be used. A test supporting a consequential treatment decision raises different questions from an assay providing supplementary information. Screening, diagnosis, monitoring, and treatment selection should not be treated as interchangeable claims.

Consider an assay described simply as detecting a pathogen. Its classification assessment requires more detail: which pathogen, what clinical purpose, and whether the result supports diagnosis or another use, such as donor screening. The clinical consequence of an incorrect result helps explain the risk assignment, but evaluators should still establish the applicable catalogue entry rather than infer classification from perceived severity alone.

Novel analytes, unusual specimen types, and combined claims deserve particular scrutiny. A catalogue entry that appears relevant may have a narrower scope than the proposed product. Where the classification cannot be established confidently through the applicable rules and catalogue, the classification determination process should precede major evidence-generation commitments.

The assessment must also identify what is being submitted. An IVD reagent, its associated instrument, and software may have different regulatory treatment. A reagent’s class should not automatically be assigned to an analyzer or software component. Compatibility among them must nevertheless be reflected in the evidence supporting the diagnostic system.

What changes between filing and registration

Class I filing and Class II or III registration are different administrative routes. Filing does not provide the same premarket technical review as registration, but it still requires the filer to support the product’s safety, performance, and conformity with applicable requirements. A lower risk class reduces certain procedural demands; it does not make unsupported performance claims acceptable.

• Class I: Filing documentation supports the product description, intended use, technical requirements, performance, manufacturing arrangements, and other applicable filing requirements.
• Class II: Registration requires a technical dossier demonstrating safety and effectiveness for the proposed use, including applicable analytical, clinical, and quality management evidence.
• Class III: Registration also requires evidence matched to the proposed use, with particular attention to the consequences of incorrect results and the product-specific requirements associated with higher risk applications.

The responsible authority also depends on manufacturing origin. Domestic Class I products generally follow filing with the relevant municipal authority, while domestic Class II registration is handled at provincial level. NMPA handles domestic Class III registration and imported Class II and III registration, as well as imported Class I filing. For an overseas manufacturer, this distinction affects the application route, representation arrangements, and required supporting documents.

Class II and Class III should not be understood as two fixed document packages differing only in size. Product-specific guidance, the intended use, clinical evaluation route, and technical characteristics determine the depth of the submission. A complex Class II assay can require substantial validation, while an established Class III product may have a well-defined evidence pathway. Class sets the route; the particular claim determines much of the work within it.

Foreign authorization can contribute useful supporting material, but CE marking or another market authorization does not establish China registration readiness. Evaluators need to compare the authorized overseas configuration and claims with the proposed Chinese submission, then assess whether the underlying evidence satisfies the applicable Chinese requirements.

Analytical evidence must match the submitted product

Across risk classes, analytical performance must support the claimed result. The practical difference is how that evidence is documented, reviewed, and connected to the applicable submission requirements. For Class II and III registration, evaluators should expect a coherent relationship among the research reports, product technical requirements, testing evidence, instructions for use, and risk management documentation.

Begin with the measurement claim. A quantitative assay may require evidence addressing precision, measuring interval, detection capability, interference, and agreement or traceability, as applicable. A qualitative assay needs evidence supporting detection limits, decision thresholds, reproducibility around the cutoff, and relevant cross-reactivity. The exact study set depends on assay design and intended use; a generic validation checklist cannot establish completeness.

Acceptance criteria need a clinical rationale. An impressive detection limit is insufficient if performance around the clinical decision point is unstable. Likewise, a high overall agreement figure can conceal weaknesses in low-concentration specimens or a subgroup central to the claim. Technical reviewers should examine the distribution of results and the study conditions behind headline metrics.

Specimen claims are a frequent source of gaps. If labeling includes several specimen types, the evidence must justify each one through appropriate studies or a defensible bridging approach. Collection devices, transport conditions, storage duration, and sample preparation can affect results. Evidence obtained with one specimen matrix should not silently support a broader label.

Product configuration also matters. Validation should represent the submitted reagent formulation, supported instruments, software version, calibration approach, and operating conditions. Where several configurations are claimed, the dossier needs a reasoned strategy for demonstrating compatibility. Data from a development prototype may require bridging before it can support the commercial configuration.

Registration testing deserves separate attention. Chinese requirements provide routes for qualifying applicant self-testing under specified conditions; external testing is not the only possible approach in every case. However, an internal validation report does not automatically satisfy registration testing requirements. The applicable route, laboratory capability, methods, records, and coverage against the product technical requirements must be assessed explicitly.

For technical evaluators, product technical requirements are a practical anchor. They define relevant performance indicators and test methods for the product’s regulatory control. A submission becomes vulnerable when these requirements, the validation reports, and the instructions for use describe different performance boundaries.

Clinical evaluation depends on the claim and the applicable pathway

Higher risk often increases the clinical questions a submission must answer, but classification alone does not decide whether a new clinical trial is required. The applicable clinical evaluation provisions, exemption catalogue, product characteristics, and claimed use must be considered together. “Class II needs no trial” and “every Class III product needs a new local trial” are both unreliable planning assumptions.

For products covered by an applicable clinical trial exemption, exemption concerns the trial requirement. It does not remove the need for clinical evaluation or the evidence supporting safety and effectiveness. The applicant must establish that the product falls within the relevant scope and provide the evaluation material required for that route. A superficially similar catalogue description is not sufficient if the intended use or technical characteristics differ materially.

Where clinical studies are necessary, the study must answer the diagnostic question in the proposed label. Diagnostic sensitivity and specificity, agreement with a comparator, or other performance measures need to be selected according to the intended use. The comparator should be justified, particularly when no single reference method fully establishes disease status.

Sample selection can materially affect the conclusion. A study dominated by clear positive and negative specimens may underrepresent borderline results, relevant disease stages, interfering conditions, or patients receiving treatment. Evaluators should examine whether the study population reflects the clinical setting and whether subgroup claims have adequate support.

Overseas clinical data may contribute to a Chinese submission where applicable requirements permit their use. Their acceptability depends on relevance and quality, including study design, population characteristics, clinical practice, specimen handling, and the submitted product’s consistency with the tested version. Existing data may support some claims while leaving others unresolved; a foreign study report should not be treated as automatically transferable.

A companion diagnostic illustrates the importance of claim specificity. Demonstrating reliable detection of a biomarker addresses analytical performance. Supporting use of that result to select patients for a particular treatment requires evidence connecting the test, biomarker definition, and therapeutic decision. A broad biomarker study cannot automatically justify every treatment-selection claim.

Quality documentation must explain reproducible performance

A strong dossier connects validated performance to routine production. Manufacturing controls, critical raw materials, lot consistency, stability, and change control determine whether the marketed product can reproduce the results presented in the submission. These questions become particularly consequential when small changes can shift a cutoff or affect a clinically significant result.

A quality management certificate can support an assessment, but it does not replace the quality documentation or verification required for the Chinese regulatory pathway. Evaluators should look for evidence that the quality system covers the submitted product, relevant manufacturing activities, and outsourced processes. The precise documentation and inspection requirements depend on the application and applicable rules.

For reagents, stability evidence should support the claimed shelf life and relevant handling conditions. Unopened storage, transport, open-vial use, onboard stability, and prepared-component stability may present different questions. A favorable accelerated study does not by itself justify every storage and use claim.

Lot coverage also needs a rationale. Review whether the selected lots represent the manufacturing process and whether release criteria control characteristics that affect diagnostic performance. Weak linkage between critical materials, acceptance testing, and assay behavior can undermine confidence even when the initial validation results are strong.

Turn classification into an evidence-gap decision

A useful readiness review begins with a claim-to-evidence map. Place each intended-use element beside its classification basis, applicable requirement, supporting report, and limitation. This exposes the difference between having many documents and having evidence that supports the exact product proposed for registration.

Five questions provide a focused starting point:

• Does the classification rationale address the full intended use and the relevant catalogue scope?
• Do analytical studies cover every claimed specimen, configuration, and clinically significant performance boundary?
• Is the clinical evaluation route justified, with any exemption or use of overseas data supported on its own terms?
• Do the technical requirements, labeling, and study reports describe a consistent product and claim?
• Can manufacturing and quality records explain how validated performance will be maintained across commercial lots?

Prioritize gaps according to their effect on the regulatory route. An unresolved classification or treatment-selection claim can alter the evidence program. A missing configuration bridge can require additional testing. An inconsistency between a report and labeling may be resolved through document correction, provided the underlying evidence supports the final claim.

The next useful decision is therefore specific: establish classification before committing to a study plan, narrow an unsupported claim, generate missing evidence, or align the dossier around an adequately validated configuration. Registration readiness emerges when the intended use, submission pathway, and technical evidence support the same diagnostic product.