6 min read
In this guide
When a supplier recalls a component, the question it sends you runs the opposite way from how most kit records are built. A batch record answers “what went into kit lot 2611.” A component recall asks “which kit lots used swab lot 4471, and where did every one of those kits go.” For an at-home testing program, the second half of that question often sits in another company’s system.
The recall question arrives component first
ISO 13485:2016 asks for a record of each batch that provides traceability and states the amount manufactured and the amount approved for distribution, and it asks the manufacturer to define how far traceability must reach under the regulations that apply (clauses 7.5.1 and 7.5.9). Most kit records meet that by listing every component lot against the finished kit lot. That is a top-down record. Reading it from the bottom up, from one component lot to every kit lot that consumed it, is a search. How long the search takes depends on whether the records are structured data or scanned batch records.
Then comes the second half: where the kits went. FDA’s recall policy sets a depth for every recall, from wholesale level to retail level to consumer or user level, and an effectiveness check level, from Level A, which means contacting 100 percent of consignees, down to Level E, no checks at all (21 CFR 7.42). A kit that ships direct to patients makes every home address a consignee.
A real cascade: 9.1 million gowns, 2.9 million packs
In January 2020 Cardinal Health recalled about 9.1 million surgical gowns after learning that a supplier had moved some production to unapproved sites in uncontrolled environments. A week later it announced field actions on about 2.9 million Presource procedure packs, built between September 2018 and January 2020, that contained those gowns. The gown was one component. The packs were the finished product that carried it into hospitals.
The split between the two actions is the useful part. 374,794 packs had their other components separated from the gown by inner sealed packaging, so they could be corrected with an over-label and used once the gown was discarded. 2,518,653 packs had components that were not separated from the gown, and those had to come back. The same recalled component produced a correction in one configuration and a removal in the other. Packaging design, settled long before anyone knew about the supplier, decided which.
The scope of a kit recall is set on the build line and in the packaging spec, months before the recall notice arrives.
Lot mixing sets the size of the recall
An illustrative example. A 10,000-kit lot is built over three days. Swab lot A runs out at kit 6,200 and the line continues with swab lot B. The batch record lists both swab lots against kit lot 2611. Six months later the swab maker recalls lot B.
| How the build was recorded | Kits in scope | What you can tell consignees |
|---|---|---|
| One kit lot, both swab lots listed | 10,000 | Any kit from lot 2611 may contain the affected swab |
| Sub-lots 2611-A (kits 1 to 6,200) and 2611-B (kits 6,201 to 10,000), printed on the label | 3,800 | Only kits labeled 2611-B are affected |
| Sub-lots, plus a ship record linking each kit ID to its sub-lot | 3,800, with a named consignee list | Exactly which addresses received an affected kit |
The first row is not a traceability failure. Both swab lots are traceable. It simply makes the recall 2.6 times larger than the problem, and at consumer-level depth with a Level A check, that is 6,200 extra people contacted about a kit that was never affected.
The fix is cheap when it is decided before the build: when a component lot changes mid-run, close the kit lot or open a sub-lot, and print the sub-lot on the label. Reconstructing the split afterward, from line logs and operator memory, is expensive and rarely convincing.
Who holds the ship-to list
In a bulk program, kits ship by the carton to clinics and laboratories, and the consignee list is a short shipping log. In a direct-to-patient at-home testing program, kits often ship against orders placed in the sponsor’s platform, a telehealth partner or a laboratory ordering portal. The kit manufacturer knows which kit lot left the building in which carton. The name and address may live in a system it does not control, and linking a kit to a person is a deliberate step with its own privacy obligations (see why a kit ID is not a patient identifier until someone links it).
That split is workable as long as someone has written down the join: kit ID or sub-lot to shipment, shipment to consignee, and who runs that query when a recall lands. If nobody owns it, the first recall is where the program finds out.
Time matters. When a device manufacturer starts a correction or removal to reduce a risk to health, it has 10 working days from starting it to report to FDA under 21 CFR 806.10. Who counts as the manufacturer for a given kit, and whether a given action is reportable, are legal determinations; confirm them with regulatory counsel. Having the records to act on is an operations question, and the answer needs to come back in days.
Run the drill from the component end
Mock recalls usually start from a finished kit lot, because that is how the records are filed. Start one from a component lot instead and time each step.
- Pick a component lot at random from receiving records: a swab, a tube, a preservative fill.
- Find every kit lot and sub-lot that consumed it. Note whether you searched structured data or opened batch records by hand.
- For each, find the quantity built, released, shipped and still in inventory.
- Produce the consignee list for the shipped kits, and note who had to be asked for it.
- Check whether the component sits in its own sealed inner pack or loose with others. That decides whether a correction is possible or only a return.
- Record the elapsed time and the step where the drill stalled.
The step that stalls is usually the handoff between two organizations’ records, not the search inside either one. Supplier change notices and component expiration dates fail at the same handoff.
Ask your kit manufacturer for the last mock recall it ran from a component lot, how long it took, and who produced the ship-to list. If the drill started from the finished kit lot, ask for it the other way. The audit support you should expect includes that record.
Sources: 21 CFR 7.42, recall strategy (depth of recall and effectiveness check levels). 21 CFR 806.10, reports of corrections and removals. Cardinal Health, voluntary field actions for select procedure packs containing affected surgical gowns, January 30, 2020. Infection Control Today, Cardinal Health pulls 2.9 million Presource procedural packs. ISO 13485:2016, Medical devices: Quality management systems, clauses 7.5.1 and 7.5.9.



