The Line That Decides the Class: How a Wound Dressing's Claim — Not Its Chemistry — Sets Its Regulatory Class in the US, EU and China

1. Two liquid bandages, two regulatory universes

Takeaway: the smallest possible proof of the thesis. One CFR section, two product codes, opposite routes to market — and the only difference is what the label claims.

Title 21 of the US Code of Federal Regulations, section 880.5090, defines a liquid bandage as "a sterile device that is a liquid, semiliquid, or powder and liquid combination used to cover an opening in the skin or as a dressing for burns." It is a Class I device. The exemption from premarket notification applies when the device functions solely as a topical skin protectant ⁴.

FDA implements that single regulation as two product codes:

• NEC — "Bandage, Liquid, Skin Protectant": Class I, 510(k) exempt.
• KMF — "Bandage, Liquid": Class I, 510(k) required ⁶.

Same regulation. Same shelf in a pharmacy. Same polymer, in many cases the same factory. One of them needs a premarket submission and one does not, and the fork is entirely a function of how the intended use is written.

The consequences show up immediately in the public record. KMF has accumulated 222 clearance decisions since 1977, the most recent in September 2024. NEC has three, the last in December 1997 ⁷. A naive reading says the skin-protectant category died in the nineties. The registration file says the opposite: 144 establishments are currently registered to list NEC products, against 79 for KMF ⁸. The exempt category is nearly twice the size of the reviewed one, and it stopped producing paperwork almost thirty years ago.

This is the pattern to hold in mind for the rest of this report. In wound care, an empty clearance record usually means "exempt", not "absent".

2. The ladder: what actually moves a dressing up a class

Takeaway: US wound-dressing classification is a ladder, and every rung is triggered by a claim or an added ingredient — never by the base material.

Read the exempt regulations closely and the logic is explicit. An occlusive wound dressing under 21 CFR 878.4020 is "a nonresorbable, sterile or non-sterile device intended to cover a wound, to provide or support a moist wound environment, and to allow the exchange of gases such as oxygen and water vapor through the device." It is Class I and exempt from premarket notification, subject to the limitations in § 878.9 — and the regulation expressly excludes occlusive dressings that contain added drugs, biologics, or materials derived from animal sources ¹.

The hydrophilic wound dressing at 878.4018 is built the same way: "intended to cover a wound and to absorb exudate", made of nonresorbable hydrophilic material — cotton, cotton derivatives, alginates, dextran, rayon — Class I, exempt, with the identical carve-out for added drugs, biologics and animal-derived materials ².

Silicone sheeting at 878.4025, "intended for use in the management of closed hyperproliferative (hypertrophic and keloid) scars", is likewise Class I and exempt ³.

So three of the largest sub-categories in wound and skin care — the absorbent dressing, the moisture-retentive dressing, and the scar-management sheet — require no premarket submission in the United States at all. What breaks the exemption is not a change of polymer. It is:

1. Adding a drug. The product moves to FRO, "Dressing, Wound, Drug" — unclassified, 510(k) required. This is the single busiest file in the category: 705 clearance decisions, a third of the 2,101 wound & skin clearance decisions on record, and still active with decisions through June 2026 ⁷.
2. Adding animal-derived material. KGN — unclassified, 510(k) required, 189 decisions ⁷.
3. Adding a permanently bound antimicrobial agent. SHA, 21 CFR 878.4013 — Class II ⁶.
4. Claiming the dressing is "interactive". MGR, "Dressing, Wound and Burn, Interactive" — Class III, premarket approval ⁶.

And the exemption is conditional even when nothing is added. Section 878.9 withdraws it where a device has a different intended use from legally marketed devices of that generic type — including a move from professional to consumer use — or where it operates by a different fundamental scientific technology ⁵. A dressing sold to hospitals for a decade does not carry its exemption automatically into a retail consumer launch.

There is a visible signature of all this in the clearance record. Several major codes simply stop:

Source: FDA 510(k) database — VEMERIX analysis, July 2026 ⁷.

None of these categories died. KGX alone has 977 registered establishments and 4,685 device records currently on the US market ⁸⁹. They went exempt, and the paperwork stopped.

3. Three regulators, one product

Takeaway: the sentence that keeps a dressing exempt in the United States is close to the sentence that escalates it in Europe — and China moved the line again in 2022. There is no single "wound dressing class"; there is a class per market, per claim.

The European Union: Rule 4, and the rule above Rule 4

Before Rule 4 there is a sentence that decides everything. EU MDR 2017/745, Annex VIII, Chapter II, implementing rule 3.1: "Application of the classification rules shall be governed by the intended purpose of the devices." Rule 3.5 adds that where several rules or sub-rules could apply, "the strictest rule and sub-rule resulting in the higher classification shall apply" ¹².

That is this report's thesis, written by the regulator. Classification follows the intended purpose — the claim — and where the claim is ambiguous, the ambiguity resolves upward.

Rule 4 then reads in full:

"All non-invasive devices which come into contact with injured skin or mucous membrane are classified as:

• class I if they are intended to be used as a mechanical barrier, for compression or for absorption of exudates;
• class IIb if they are intended to be used principally for injuries to skin which have breached the dermis or mucous membrane and can only heal by secondary intent;
• class IIa if they are principally intended to manage the micro-environment of injured skin or mucous membrane; and
• class IIa in all other cases." ¹²

"Injured skin or mucous membrane" is defined in the same Annex as "an area of skin or a mucous membrane presenting a pathological change or change following disease or a wound" — broad enough that essentially every product in this report is inside Rule 4 ¹².

Now read that third bullet next to the American definition of an exempt occlusive dressing — a device "intended to cover a wound, to provide or support a moist wound environment" ¹. The US regulation grants its exemption to a device described by the very phrase that, in Europe, is the definition of Class IIa: managing the micro-environment.

This is the most expensive sentence in wound-care marketing. A manufacturer that writes "maintains an optimal moist healing environment" on its packaging has said nothing that costs it anything in the United States, and has, on the face of Rule 4, argued itself into a notified-body route in Europe. Where the product is indicated for wounds that have breached the dermis and can only heal by secondary intent — deep and chronic wounds — Rule 4 puts it in Class IIb, two full steps above the US position.

China: sterility, depth, absorbability

China's National Medical Products Administration classifies dressings by a different triad. Per NMPA's guidance on medical dressings: a dressing that claims to be sterile is at minimum a Class II device; a dressing that contacts wounds where the deep dermis or below is damaged, or that is used on chronic wounds, or that can be wholly or partly absorbed by the body, is Class III. Domestic Class II devices register with the provincial authority; Class III devices and imported Class II devices register nationally with NMPA ¹⁵.

Two further Chinese rules matter for anyone selling into or out of that market.

Naming is regulated. A device name must follow the general naming rules and may not contain wording such as "beauty" (美容) or "health care" (保健), nor exaggerate the scope of use. NMPA has stated flatly that the "device-registration facial mask" (械字号面膜) does not exist as a category and that a medical device may not be named 面膜 at all ¹⁵. For a manufacturer whose wound and skin line touches cross-border consumer e-commerce, that is not a technicality — it governs what can legally appear on the box and in the listing.

The Class I route for liquid dressings was closed. On 31 December 2021 NMPA published Announcement 2021 No.158, a revised Class I Medical Device Product Catalogue, effective 1 January 2022 ¹³. The official revision note is explicit: item (七) records the deletion of the third entry under the 2017 catalogue's "14-10-08 liquid and paste dressings" — that is, non-sterile liquid and paste dressings that act as a physical barrier by forming a protective layer, used for minor wounds, abrasions and cuts, are no longer in the Class I catalogue. The same note restricts the physical-cooling category, removing "medical cold compress patch / headband / eye mask / cooling gel" from its example names, and adds a prohibited-ingredient annex for named Class I categories: no traditional-Chinese-medicine, chemical-drug, biological, disinfectant or antibacterial ingredients, no plant extracts with pharmacological, immunological or metabolic action, and nothing absorbable by the body ¹⁴.

Put the three regulators side by side and the asymmetry is the whole story.

For a distributor, the practical consequence is that a supplier's regulatory status is not portable. "We are FDA-listed" may mean the supplier typed its details into an exempt-code registration. "We hold NMPA Class II" is a real, examined registration, but it says nothing about Rule 4. The only question that travels across all three systems is: what exactly does the label claim, and in which market am I about to make that claim?

4. Where the supply actually is

Takeaway: the wound-care supply base is far more Chinese, and far more concentrated on the exempt shelf, than the clearance record implies. This is not an inference — it is what the registration file says.

We built the picture from FDA's Establishment Registration & Device Listing file across the same 31 product codes. It contains 2,061 unique establishments.

By country of the registered establishment ⁸:

Now split the same file by whether the product code requires a 510(k):

Sources: FDA 510(k) database, FDA Registration & Listing, FDA Product Classification — VEMERIX analysis, July 2026 ⁷⁸⁶. An establishment that lists codes on both sides of the line appears in both rows, so the establishment column is not additive; the clearance column is, and sums to the 2,101 decisions on record across these codes.

Three observations follow.

First, the exempt shelf is three times more Chinese than the reviewed shelf. The mechanism is straightforward and has nothing to do with the quality of any individual factory: where a market can be entered by registration alone, the cost and time of entry collapse, and a large export manufacturing base serves those sub-categories first. The sub-categories that still require a submission — drug-bearing, biologic-bearing, antimicrobial — carry a regulatory workload that concentrates in fewer hands everywhere, not only in China.

Second, the concentration is code-perfect. Rank the codes by number of Chinese establishments and the top seven are all exempt codes: adhesive tape and bandage (KGX, 347), elastic bandage (FQM, 275), hydrophilic dressing (NAC, 201), occlusive dressing (NAD, 183), hydrogel dressing (NAE, 114), compression dressing (MHW, 86) and silicone scar management (MDA, 84). The first 510(k)-required code to appear is FRO, drug-bearing dressings, with 50 ⁸.

That MDA line deserves a moment. Silicone scar sheeting — one of the largest consumer-facing categories in this entire report — has 191 registered establishments worldwide, 84 of them in China, and requires no US premarket submission whatsoever ⁸³. A shopper comparing scar sheets is comparing products that FDA has never reviewed and, under the current regulation, never will.

Third, the clearance record does register the change — just late and small. Chinese applicants hold 56 clearances across these 31 codes, 42 of them in FRO, the drug-bearing code ⁷. Their share of decisions in the 2016–2026 window is 15.4%, against 59.9% for US applicants and 7.6% for the UK, from effectively nothing two decades ago ⁷. Chinese manufacturers are climbing the ladder — but on the exempt rungs they never had to file at all, so the clearance record understates the position by design.

One more asymmetry is worth naming, because it shapes what a distributor is actually buying. The US patent estate in wound dressings sits with the incumbent durable-device and large-dressing companies: across 2,647 US patent applications matching "wound dressing" filed since 2005, the most frequent named applicants are KCI Licensing, Smith & Nephew, Solventum, Mölnlycke, 3M and Coloplast, and of the applications that name an applicant at all, roughly 2% carry a China-indicating one ²⁰. We treat that as a structural observation only — two filing years are missing from the public export we used and 37% of rows carry no applicant name, so it will not support a trend line. But the direction matters: the exempt shelf is not where the patents are. It is where manufacturing capability and registration discipline are, which is precisely why the diligence questions in section 8 are about the quality system rather than the intellectual property.

Finally, what is actually on the shelf. FDA's device-identification database returns 19,245 records touching these codes, of which 17,696 — 92.0% — are flagged single-use ⁹. That is the same durable-to-disposable signature we documented across four other VEMERIX device lines, arriving here as a starting condition rather than a transition.

5. "Registered" is not one thing

Takeaway: an FDA registration tells you that an entity filed, not what that entity does. Roughly three in ten Chinese listing rows in this category describe an export operation rather than a factory — and the US rows describe a different industry altogether.

Every establishment in the registration file declares what it does with the device. Split those declarations by country and the two halves of the register turn out to be doing almost opposite jobs ⁸:

The American side of this category is largely a finishing and branding layer: sterilizing, repacking, relabelling, and developing specifications for product manufactured elsewhere. Over half of US listing rows describe one of those functions rather than manufacturing. That is consistent with the device-record ranking, where distribution and private-label houses — Medline with 2,029 records, DeRoyal 603, Cardinal Health 472 and 355, McKesson 289 — sit at or above most manufacturers ⁹.

The Chinese side is largely making and shipping. But note the second and third rows: about 30% of Chinese listing rows carry the "export to the US, no other operation" role, either alone or combined with manufacturing. In plain terms, a meaningful share of the Chinese entries a buyer finds in FDA's database are trading and export entities, not the plant where the product is made.

The register cannot answer that question for you. An establishment registration is a self-declaration; on the exempt route there is no clearance to check it against; and the entity that appears in the search result may be one step, or three, removed from the line that actually extrudes the film. Ask for the manufacturing FEI, the physical plant address, and which role the registration declares. All three are answerable in a sentence by a manufacturer and awkward for a trader.

One technical caveat we hold throughout: these are counts of registered establishments, not of companies. A corporate group can hold several registrations, so the country shares describe registered sites, not corporate market share.

6. What the record does and doesn't warn you about

Takeaway: the recall file is genuinely useful and says the risk lives in the process and the pack. The adverse-event file is genuinely misleading unless you decompose it, and comparing suppliers on raw report counts rewards whoever reports least.

Recalls: the failure is in the process, not the polymer

FDA's device-recall database returns 370 recall events across the cluster. Ranked by root cause ¹⁰:

Under investigation by the firm (58), process control (38), package design or selection (38), nonconforming material or component (37), material or component contamination (37), other (31), packaging (21), packaging process control (19), employee error (15), environmental control (15), labeling design (11), and — worth its own line — no marketing application (10).

Almost nothing in that list is about the dressing's design. It is about whether the plant held its process, whether the pack maintained the barrier, and whether the incoming material was what it was supposed to be. Recent entries make the point concrete: in May 2026 DeRoyal Industries recalled a sterile wound dressing that "lacks sterility assurance"; in March 2026 Integra LifeSciences recalled an animal-derived dressing over an "increased rate of out-of-specification endotoxin results"; in April 2026 Oculus Technologies de México recalled a drug-bearing product because of corrosion in the internal metallic spring of the sprayer component ¹⁰. These are large, established manufacturers. The failure mode is structural to the category, not particular to any tier of supplier.

By product code, the recall burden concentrates on dressing kits (MCY, 98 events) and drug-bearing dressings (FRO, 81) — the two places where more components, more suppliers and more claims meet in one pack ¹⁰.

The buyer's conclusion writes itself. In a category where most of the product needs no premarket review, the quality system is the product. Sterilisation validation, incoming-material control, packaging validation and environmental monitoring are not back-office paperwork here; they are the entire mechanism by which this category fails.

Adverse events: how to read the number without being misled

Now the trap. Pull MAUDE for the same cluster and you get 31,826 reports, against 9,217 for the powered NPWT pump codes ¹¹. Taken at face value, the passive consumable looks three and a half times more dangerous than the electrically powered pump.

Decompose it and the number dissolves:

• 47% of all dressing reports sit on a single product code — NAD, occlusive wound dressing, with 14,952 reports. That is the Class I exempt code.
• 86% of those NAD reports — 12,848 of 14,952 — come from one manufacturing entity's filings.
• And the patient-problem coding on NAD is dominated by non-events: "no clinical signs" 7,215, "symptoms or conditions" 7,181, "no patient involvement" 6,143, against 61 coded as "injury" ¹¹.

MAUDE is a passive, largely voluntary surveillance system. A report is not a finding of causation, and volume tracks who files, and how thoroughly, far more than what is dangerous. A buyer who ranks suppliers by raw MAUDE counts will systematically prefer the companies with the least developed complaint-handling — the exact opposite of the intended outcome.

The honest reading is narrower and more useful: the exempt shelf is not a zero-event shelf; the events that do get reported are overwhelmingly non-harm; and neither observation can be converted into a supplier ranking. Use MAUDE to find specific recurring failure modes on a specific product family. Do not use it as a scoreboard.

7. What you may claim, and what the evidence supports

Takeaway: regulatory permission and clinical proof are different lines, and in this category they run in opposite directions. The claim that needs no FDA submission at all rests on the weakest evidence; the claim with the strongest primary experiment behind it is also exempt.

Moist healing: a 64-year-old primary experiment

The occlusive dressing category exists because of one paper. In 1962 George Winter published "Formation of the scab and the rate of epithelization of superficial wounds in the skin of the young domestic pig" in Nature. He made partial-thickness wounds on pigs, left half open to the air and covered half with a polymer film, and measured the migration of new epithelium. The covered wounds re-epithelialised at roughly twice the rate of the wounds allowed to scab ¹⁶. That result overturned the prevailing "let it dry" doctrine and produced the entire moist-wound-healing category — including the CFR definition, written three decades later, that grants those dressings their exemption.

It is worth noticing what kind of evidence that is: a controlled primary experiment with a clear mechanism, replicated and extended for six decades. Very little else in this category has anything like it.

Silicone for scars: a large market on low-certainty evidence

Silicone sheeting and gel occupy an enormous retail shelf and, as section 4 showed, 191 registered establishments and no US premarket review. The clinical evidence is considerably thinner than the shelf implies.

The current Cochrane review of silicone gel sheeting for treating hypertrophic scars (CD013357, Jiang and colleagues, published September 2021) identified 13 randomised controlled trials with 468 participants, of whom 425 completed. Its authors' conclusion: "there is currently limited rigorous RCT evidence available about the clinical effectiveness of SGS in the treatment of hypertrophic scars." Silicone gel sheeting may slightly improve scar appearance compared with onion extract and may reduce pain compared with pressure garments, but the review rated the certainty of evidence low to very low across all comparisons, citing few and small studies with poorly reported results ¹⁷. The earlier Cochrane review covering both prevention and treatment (CD003826, O'Brien and Jones, 2013) reached the same posture: trials showed improvements in scar thickness and colour, but were of poor quality and highly susceptible to bias ¹⁸.

None of this means silicone does not work, and it is not an argument against stocking the category. It is an argument about what a distributor or manufacturer should put in writing. "Used in the management of hypertrophic and keloid scars" is the regulatory language, is accurate, and is exactly what the US regulation contemplates ³. Claims of a specific percentage improvement, a healing timeline, or superiority over another modality are not supported by the current systematic-review evidence — and, under § 878.9, an intended-use claim that departs from the generic type is also the mechanism by which a product loses its exemption ⁵.

The demand under all of it

One number to keep the commercial context honest. In the United States, debridement services billed to Medicare fee-for-service rose from 2,137,048 in 2013 to 3,708,207 in 2024 — a 73.5% increase — across roughly 776,000 beneficiaries in 2024 ¹⁹. Debridement is brand-agnostic and precedes or accompanies advanced wound management, so it is a clean proxy for how many serious wounds the system is actually treating. This is a Medicare fee-for-service window, a directional read on US demand rather than a world total, but the direction is unambiguous: the wound population that consumes dressings is growing faster than the population itself.

We deliberately do not open this report with a market-size figure, and here is why. Published 2024 estimates of the global advanced-wound-care market from commercial report vendors range from $11.25 billion to $15.6 billion for the same year — a spread of roughly 39% ²¹. When the available estimates of a market disagree by two-fifths, the number is not evidence; it is a vendor's model. Everything quantitative in this report is instead counted from primary registers, and every count is reproducible from the sources listed at the end.

8. A sourcing scorecard by claim tier

Takeaway: the higher the claim, the less the FDA register can reassure you — so the diligence has to shift from the database to the quality system. Here is the checklist, rung by rung.

Two habits are worth adding to the list.

Ask what changed, and when. Because so much of this category is exempt, product changes do not generate a public trail. A supplier can reformulate an adhesive, change a sterilisation contractor or move a line to a new building without any of it appearing in a database you can search. The change-control log is the only place that history exists, and a manufacturer running a real quality system can produce it.

Match the pack to the market before the first order, not after. Every failure mode in this report is cheap to fix at specification stage and expensive to fix in a warehouse: the EU wording that moves the class, the Chinese name that cannot be registered, the antimicrobial additive that voids a US exemption.

9. Where VEMERIX sits

Takeaway: VEMERIX manufactures four products that sit at different rungs of the ladder described above, and its wound & skin line is registered on the Chinese side of it.

VEMERIX is the brand of Weihai Medison Medical Equipment Co., Ltd., a Korean-invested manufacturer founded in 2003 in Weihai, Shandong, operating a six-line minimally-invasive surgery platform. The Wound & Skin line is the subject matter of this report ²²:

Where a cell is blank, it means the class or registration number is not published on the product page. We have left it blank rather than infer one, which is the same standard section 8 asks a buyer to hold a supplier to.

Two things are worth stating plainly, because this report has spent eight sections arguing that vague regulatory language is the problem.

First, these are Chinese registrations and nothing more. They are examined Class II registrations on the ladder described in section 3 — which, as that section showed, is a higher floor than the US exempt route for the same product types. They are not CE marks and they are not FDA clearances, and nothing in this line is offered as either.

Second, the products are described here exactly as they are registered and specified — a physical barrier, a silicone-based gel, a cyanoacrylate adhesive, a cooling hydrogel — with no efficacy claim beyond that. That is the discipline section 7 asks of everyone else in the category, and it applies to us first.

10. Conclusion: four things to take away

One. In wound and skin care the claim is the regulation. Not the polymer, not the process, not the price. Three of the largest sub-categories in the field — absorbent dressings, moisture-retentive dressings, silicone scar sheeting — are Class I and 510(k)-exempt in the United States, and what breaks that exemption is an added drug, an added biologic, an animal-derived material, a bound antimicrobial, or a change of intended use ¹²³⁵.

Two. Because of that, the public record systematically under-describes the industry. There are 1,776 registered establishments on the exempt codes against 603 on the reviewed ones, and China's share is 29.8% versus 9.8% ⁸. The clearance database is not a map of this category; the registration file is closer, and even it declares roles rather than capabilities.

Three. The regulatory answer does not travel. The sentence that keeps a dressing exempt in the US is close to the sentence that makes it Class IIa in Europe, and China closed the Class I route for non-sterile liquid and paste dressings on 1 January 2022 ¹²¹⁴. A supplier's status in one market is evidence of very little in the next one.

Four. Where the databases go quiet, the quality system is the only remaining evidence. The recall record says the category fails on sterility, packaging, incoming material and process control — not on design ¹⁰ — and the adverse-event record cannot be used as a supplier scoreboard without producing the wrong answer ¹¹. So the diligence that actually protects a distributor is documentary: the manufacturing FEI and plant address, the sterilisation validation, the change-control log, and the exact intended-use wording that will appear on the box in the market where it will be sold.

If you are evaluating a wound and skin line for a market with its own version of this ladder, the useful conversation is the specification one: what the label will claim, what class that claim creates in your market, and what the quality file has to look like to support it. That is a conversation we are set up to have before the first purchase order, not after it.