The conclusion first: risk decides the route, not a list of test items
The right starting point for a biological evaluation is the device description and material information, followed by chemical characterization to separate what is known from what is not. Animal testing is scheduled only when existing data plus characterization still fail to bring the risk to closure. The overall framework of the GB/T 16886 and ISO 10993 series is risk-management driven, and ISO 10993-1 sets out an approach for selecting evaluation and testing rather than a checklist to be ticked off line by line.
Doing it the other way round — ordering a batch of animal tests off the contact category and gathering material information as you go — is one of the more common sources of rework we see. The reason is plain. Animal testing answers the question of biological response, and once the formulation, the process or the sterilization method changes, whether that response still applies has to be supported by evidence at the chemical level. Without that layer, the tests you already paid for stand alone and support very little.
| Your situation | Suggested route | Why |
|---|---|---|
| Common materials, traceable formulation and process, history of use in similar devices | Start with a literature and data review plus chemical characterization; decide on animal testing from the residual risk | Most of the risk can be closed out at the data level, and what you save is mainly schedule |
| New material or new formulation, or additives and processing aids of unclear origin | Chemical characterization has to come first, to identify the unknowns | If unknowns are not identified, no animal test result can be attributed to anything |
| Nothing more than a supplier change or a masterbatch change | Handle it as a change evaluation, comparing old and new characterization data | With no substantive difference, a full re-run of the test programme is usually not needed |
| Long-term contact or implantable | Run characterization and test preparation in parallel, but let the characterization results drive the test design | A wrong test design costs far more schedule to repeat than characterization does |
| Sterilization process has changed | Look at residuals and degradation products first, then judge whether the earlier evaluation conclusion still holds | Change the sterilization method and you change what can be extracted |
Fix the nature of contact first, before picking any test items
Contact is categorised by site — surface contact, external communicating, implant — and by duration — limited, prolonged, long-term. Those two groupings multiplied together are the starting point for choosing evaluation endpoints. Two errors show up constantly:
- Categorising the whole device instead of categorising by component. Different parts of one piece of equipment can have completely different contact characteristics. Forcing the whole device into one category either buys a pile of tests you do not need, or misses the component that actually required evaluation.
- Ignoring indirect contact. Components in fluid or gas paths that contact a fluid which subsequently enters the body are within the evaluation scope. These parts are frequently bought in, and material information is harder to obtain for them than for parts you make yourself.
When the categorisation is done, put it in a table: component, material, contact site, contact duration, and whether it lies on a patient contact path. That table is the input for every judgement that follows, and it is also what you compare against, line by line, when a change comes along.
What chemical characterization is doing, and what happens when it is done badly
The role of the ISO 10993-18 side of the work is to identify and quantify the substances a material may release under conditions of use, providing the input for toxicological risk assessment. When it goes wrong, it goes wrong in four places:
| Problem | What it looks like in practice | Consequence |
|---|---|---|
| Samples submitted are not in final product state | Unsterilized semi-finished parts arrive, or samples are sent stripped of their actual packaging | The results do not represent what the patient actually contacts, and the report cannot be used in the evaluation |
| Extraction is not representative | Medium, ratio and conditions copied from somebody else's plan | Too mild and you miss things; too aggressive and you detect a pile of substances with no clinical relevance, which then need extra justification |
| Material and process information withheld | Samples supplied without formulation, additive or processing aid information | Detected substances cannot be traced to a source, identification drags on, and some may end up reported as unknowns |
| No blanks or controls designed in | Labware and medium background not considered | Background contamination gets attributed to the device, adding argumentation work for nothing |
Extraction conditions are a self-contained piece of decision logic in their own right, and choosing wrongly can make an entire report unusable. That is not expanded here; before shipping samples, check the preparation requirements on the biocompatibility testing services page.
When animal testing can be avoided
The basis for that call is not whether you feel like running the tests. It is whether three things can be argued at once:
- Whether the material and process are equivalent to something for which adequate data already exist: same formulation, same process, same sterilization method, same contact conditions. Break any one of those and you have to explain the effect of the difference.
- Whether every substance identified by chemical characterization can be identified, and whether the estimated exposure falls within a toxicologically acceptable range.
- Whether the remaining uncertainty can be covered by other evidence: published literature, prior use data, historical data from products on the same platform.
If all three hold, the evaluation can be closed out at the data level. If one does not hold, the question is which piece is missing. Missing substance identification means more characterization; only missing biological response justifies a test. Conflating those two is why a great many evaluation plans get sent back for additional work as soon as they are submitted.
When animal testing cannot be avoided
- Implants and products with prolonged or long-term contact, where evidence of local and systemic response usually cannot be replaced entirely by chemistry.
- Materials with no comparable history of use, or in-house composite systems.
- Characterization results containing substances that cannot be identified and whose exposure is not negligible.
- Unusual product forms, such as resorbable, porous or actively coated products, where release behaviour changes over time.
Even once a test is confirmed as necessary, the test design still has to be driven by the characterization results. If the route of administration or contact, the dose levels and the observation endpoints do not line up with how the product is actually used, the finished study will not support the conclusion. That "ran it and it bought us nothing" outcome is not rare.
What getting the order wrong actually costs
A recurring scene: to hit a registration milestone, a manufacturer sends a batch of animal studies out early, and only when the results come back does anyone notice that the sterilization method changed in the meantime, or that the supplier switched masterbatch. At that point there are two options — repeat the work, or produce an equivalence justification. Whether the second one stands depends on whether comparable chemistry data were captured at the time. Usually they were not.
Repeating animal work is not only a fee. It carries an ethical cost and a schedule that cannot be compressed. By comparison, pinning down the material list, the suppliers and the sterilization method together at design freeze costs almost nothing. Fixing it at design stage costs nothing. Fixing it after the study has been run means rescheduling.
What the material list should look like
| Information required | Notes | Effect if missing |
|---|---|---|
| Material grade and supplier for each component | Including base resin, masterbatch and all additives | No way to judge whether existing data can be carried over |
| Processing methods | Injection moulding, extrusion, welding, bonding, coating and so on | Substances introduced during processing get missed wholesale |
| Surface treatment and residues | Mould release agents, cleaning agents, lubricants | A very frequent source of detected substances |
| Sterilization method and process control points | Stating the process type and how it is controlled is enough | Residual and degradation product evaluation has no input |
| Packaging format and shelf life design | Whether the packaging can migrate | Post-ageing evaluation cannot be carried out at all |
Sterilization and shelf life are tightly coupled to biological evaluation; preparation requirements for each are on sterilization validation services and shelf life and ageing services. In our experience, when the material list is done thoroughly the evaluation route more or less surfaces by itself; when it is vague, every subsequent step has to be backfilled with testing, and the cost is in a different bracket entirely.
A running order you can follow
- Produce the device description and a component-level contact categorisation table, and state whether the evaluation subject is the whole device or individual components.
- Collect material, process, sterilization and packaging information into a single material list.
- Run a data review and strike out everything already covered by existing data.
- Schedule chemical characterization, using samples in final product state.
- Take the characterization results into a toxicological risk assessment and see whether the risk closes.
- Design tests only for what did not close, and write into the test plan why those endpoints were chosen.
The merit of this order is that each step narrows the scope rather than opening work up in parallel. Running everything at once looks faster; what it really does is spread rework risk evenly across every item on the list.
A note on report acceptance
Whether an evaluation conclusion is accepted depends on whether the item evaluated matches the product being filed, whether the data chain is complete, and whether every change left a traceable comparison record behind. It should be noted that an accreditation mark only demonstrates that the laboratory holds the relevant technical competence within its accredited scope; it is not a commitment regarding market access in the target market, and the applicable acceptance position remains subject to review by the receiving authority. Where a standard is cited, the currently effective version of the standard text governs. General preparation points before shipping samples are on sample submission requirements.
If you would rather skip a round of rework
If you are stuck on which step to take first, send over the component list, the material information and the sterilization method. We can map the route for you on a risk-driven basis: what can be closed on paper, what has to be measured, and what should wait until the process is settled. Getting that straight in one pass beats finishing a programme and then finding the direction was wrong. Call +86 132 4819 8029, or simply request a quote.