Where the Public Record Stops, and Who Answers After That

A Korean licence answers some questions so completely that nothing further is needed, and others not at all. The difficulty for a buyer is that the two kinds of question look alike from outside, and the register gives no signal when you have crossed from one to the other. It returns a clean row either way. Ninety articles into reading these records, the boundary is worth drawing explicitly.

What a Korean drug licence settles outright

Start with the strong side, because it is genuinely strong and gets undersold by people who have never opened the register. A drug filing is a public record with named fields, and where a field exists the answer it gives is complete.

The declared active ingredient is indexed in English, which means a product can be found by what it contains rather than by what it is called. The licence holder is named, which resolves the most common identification problem in this trade. The grant date is recorded, which places a product in time. The supply classification states whether the product is prescription-only or over the counter. A classification code places it in a therapeutic category. And the current status field distinguishes valid filings from ended ones, naming the kind of ending as a word rather than a flag: 취하 for a withdrawal taken by the holder, 유효기간만료 for a licence that reached the end of its term, 폐업 where the company ceased trading.

Those six fields answer more than most buyers expect. Whether a product exists as a licensed article in Korea, who is responsible for it, what it declares as its active, and whether that licence is still live are all settled by a lookup that takes under a minute. When a supplier's account of a product contradicts any of those six, the contradiction is decisive rather than debatable.

It is worth noticing what kind of questions those six fields are. Every one of them is a question about permission: was this allowed, by whom, when, and under what conditions of sale. That is what a licensing register exists to record, and it records it well. The questions it leaves alone are questions about substance, handling and outcome, which belong to other kinds of document produced by other parties for other purposes. The register is not incomplete. It is complete for its own subject.

The device register, which is where hyaluronic acid fillers and most biostimulators sit, is built differently and answers a different set. It carries a class, a licence number whose structure encodes whether the product was made domestically or imported, and a product type name. What it does not carry is an ingredient index. Read the two record structures side by side and the asymmetry is plain: a drug record carries twenty-one fields, a device record eleven, and none of the eleven names a material. Every one of the 609 filings in the category fillers sit under carries the same single item name, so neither material nor brand is a searchable field there at all. That is why a filler cannot be found by searching for its material, a point we set out in why your HA filler is not in the drug register.

Nine things the registers do not record

The list below is not speculative. Each entry is a question we went looking for an answer to while writing about a specific product, and did not find.

Read the third column and a pattern emerges. Most of the gaps are not voids. They are questions addressed to the wrong party. The register is a record of permissions granted, and it answers permission questions completely. Composition detail, handling history and characterization are manufacturer and supply chain questions, and they have answers held by people rather than by a database.

The second column is worth reading on its own too, because the reasons divide into two kinds. Some entries are absent because the field does not exist in that register, which is a design fact and will not change with a better query. Others are absent because the information belongs to a private document that was never public in the first place. The first kind is answered by asking a different register or a different party. The second kind is answered only by someone choosing to share it, which makes it a relationship question rather than a research question.

Only one row has nobody in the third column, and it is narrower than it first looks. The drug register does name the kind of ending. Across 302 filings for a single active, read on 11 September 2026, thirty-six had ended and the field splits them three ways: 유효기간만료 twenty-one, 취하 fourteen, 폐업 one. A term that ran out and a filing pulled back are not the same event, and the English word withdrawn hides the difference as effectively as a single count does. What is not recoverable is the motive behind a voluntary withdrawal, which is why the honest practice is to quote the field with the date it was read and stop, rather than to supply a plausible cause that nobody will be able to check later. The device register is thinner again here: it marks an ending with a numeric code and no word at all.

Why a clean lookup can still leave the question open

The register gives no indication when it is out of its depth. Search for a filler by its material and you get an empty result set, formatted identically to the empty result set you would get for a product that does not exist. Search a blended vitamin injectable by one of its components and you get thousands of unrelated filings, which is also not an error. The mechanism is working as designed in both cases and the answer is unusable in both.

This is what makes the boundary worth memorising rather than rediscovering. A buyer who knows that the device register has no ingredient index will not read a null result as evidence of anything. A buyer who does not know it will read the same null result as a red flag and either walk away from a legitimate product or start asking a supplier to explain something that has no explanation.

The index also has smaller traps of its own, including matching behaviour that varies with capitalisation, which we have covered elsewhere and which is worth knowing but is not the structural point here. The structural point is that an index answers only the questions it was built to index.

There is a version of this that applies to us as much as to any buyer. When we write about a product and a field turns out to be empty, the discipline is to record which register we asked, on what date, and with what query, so that the emptiness stays attributable rather than becoming a general claim that a product is undocumented.

Three products where the boundary decided the reading

Abstract boundaries are easy to agree with and hard to apply. Three recent readings show the line falling in a different place each time.

A hyaluronate injectable sold under one brand turned out to correspond to three valid Korean filings held by a single company, granted across thirty-one years, and the most recent of them names a cross-linked gel rather than the free hyaluronate of the other two. The register answered every question about each filing and could not answer the one that mattered to a buyer holding a carton: which of the three this box was made under. That correspondence is a manufacturer question, and knowing it is a manufacturer question is what turns a dead end into a request.

An exosome product declares a particle count, and the register has no field for particle counts in either book. This is not an omission by the regulator. Particle characterization belongs to the manufacturer's analytical documentation, and asking a licence to contain it is a category error, in the same way that asking a passport to contain a medical history would be. A buyer who understands the boundary asks for the characterization package. One who does not concludes that the product is poorly documented because the register is silent.

A hyaluronic acid filler cannot be found by searching either register for hyaluronic acid. The drug register does not hold it because fillers are devices, and the device register holds it but has no ingredient index to search. Two true facts combine into a null result that looks exactly like absence. This one is worth committing to memory because it applies to an entire product category rather than to one product, and because the null result is so easy to over-read.

In all three cases the register behaved correctly and the reading required knowing where its remit ended. That is the whole practical content of this article.

Who to ask once the record runs out

The practical value of the map above is that it converts dead ends into addressed questions. Four parties hold what the registers do not.

The manufacturer holds composition detail beyond the declared active, characterization data where the product type calls for it, and the correspondence between an export trade name and a domestic filing. These are ordinary documentation requests within a legitimate supply chain, and a distributor with a direct relationship can obtain them. A distributor who treats them as unreasonable is telling you about the length of their chain.

The distributor holds handling and provenance: how a consignment moved, under what conditions, and with what batch identifiers. No register will ever hold this, so it is not a gap to be closed by more searching. It is a question that has to be asked of a party, and the quality of the answer is itself information.

The foreign regulator holds approval status in its own jurisdiction. A Korean licence is not evidence about the United States or the European Union in either direction, and the correct move is to query the relevant database rather than to infer. This cuts both ways: the absence of a foreign approval is not a defect in a Korean product, and the presence of a Korean licence is not a claim about anywhere else.

The literature holds what is known about a material, as distinct from a product. Studies address molecules and preparations rather than brands, which is why a question about how a category behaves is often answerable while the same question about a specific carton is not.

There is an order to these that saves time. Ask the register first, because it is free and immediate and settles the largest single block of questions. Ask the distributor second, because handling and provenance questions cannot be answered by anyone else and the answer often arrives in the same message as the quote. Ask the manufacturer third, through the distributor, because those requests take longer and are worth batching. Consult the literature last and separately, since it informs a category decision rather than a consignment decision.

Reversing that order is the common inefficiency. A buyer who opens with a documentation request to a manufacturer for something the register would have settled in a minute has spent a week of goodwill on a public fact.

What remains after those four is small and mostly consists of the motives behind voluntary withdrawals. Being able to name that residue precisely is more useful than pretending it is not there.

How KSTATION works this boundary

Our own verification runs the register side first because it is fast, public and decisive where it applies. Name and licence holder, declared active, status, and for devices the licence number and class. A product that fails any of those does not proceed, and that check costs a minute.

What the register cannot settle, we settle through the chain rather than by inference. Sourcing through official Korean channels means the manufacturer documentation is reachable, which is the difference between a supply chain that can answer the second column of that table and one that can only repeat what is on the carton. Cold-chain handling and batch traceability sit on our side of the line because they are records we create rather than records we look up.

The same boundary shapes how we write. When a register query comes back empty, the article says which register was asked, on what date, and with what term, because an empty result is a fact about a query rather than a fact about a product. When a figure exists only in manufacturer material, it is attributed to manufacturer material rather than presented as a register finding. Those two habits are why our register-based pieces can be checked by a reader with the same public tools, and they are the same habits that make a sourcing note durable six months later.

Being explicit about the boundary is not a hedge. A distributor who claims that a public licence proves handling, or that a Korean filing establishes foreign approval, is claiming something the record does not support, and those claims are easy to make and hard to retract. We would rather name the limit and show what covers it. Products across the full K-derma range are read this way before they are listed, whether that means a drug filing for Hyaron Prefilled, a toxin filing for WONDERTOX, or a specification-led reading for a blended booster such as CURENEX Glow where several of the table's rows apply at once.

For the regulatory background to how these registers are structured, inside Korea's MFDS regulation covers the framework, and prescription-only and what it means when you import covers the supply classification field specifically.

Sources and references

• Ministry of Food and Drug Safety (MFDS), Korean drug and medical device product licence registers, field structure and query behaviour as observed across KSTATION editorial register work to 3 September 2026, with record schemas and the ending-status field re-read on 11 September 2026.
• KSTATION inside Korea's MFDS regulation, on the regulatory framework behind these records.
• KSTATION editorial policy, on how register findings and product claims are separated in our writing.