Warning Letters and Recalls

Two different kinds of “worse”

Two things land on your desk the same week.

First: Lakeshore’s written response to its Form FDA 483 — the two-observation report from the routine FDA (Food and Drug Administration) inspection — went out weeks ago. FDA’s own further review of that response is still pending. Your Quality Analyst (QA) asks you, over coffee, the question you’ve been turning over yourself: “If FDA isn’t satisfied with our response, what actually happens next? Do we get a Warning Letter?” You don’t know yet, and neither does FDA — but you can reason through what would have to be true for that to happen, and what wouldn’t.

Second, a completely unrelated event: your QA flags that a single production lot of plasma units, already shipped from Lakeshore’s Fairview site and received by one hospital transfusion service, was printed with the wrong plasma subtype on its container label. A reference table inside Lakeshore’s BECS (Blood Establishment Computer Software) — the one that maps an internal processing code to the exact text that prints on a label — had a wrong entry for one code. The lot’s labels read “Plasma Frozen Within 24 Hours After Phlebotomy” (PF24) when the units were actually standard Fresh Frozen Plasma (FFP), a different subtype with meaningfully different clotting-factor levels. No unit has been transfused. The hospital hasn’t issued any of them yet. But the lot has left Lakeshore’s direct control, and it’s sitting on a shelf somewhere with the wrong name on it.

Both problems are “worse” than a routine finding. Neither one is the same kind of worse. This module is how you tell them apart, and how to act on each one correctly.

Why this rule exists — and the ITSM bridge

A 483 that escalates and a product that gets pulled from distribution are two different failure modes FDA’s enforcement system is built to catch. One is about whether a firm is fixing what it finds. The other is about whether a specific, already-distributed product is dangerous enough that leaving it in the field is itself the violation. Confusing the two — treating every serious finding as “recall territory,” or assuming a messy inspection automatically means a recall is coming — gets a Director either complacent about the wrong thing or alarmed about the wrong thing.

You already run something structurally close to both halves of this. Escalation is incident management’s own escalation path: a ticket that isn’t resolved, or where the fix isn’t working, moves up — more senior ownership, more scrutiny, a different review cadence — because the facts earned it, not because someone decided to be dramatic. A Form 483 escalating to a Warning Letter works the same way.

Recall classification is incident severity classification, applied to a product instead of a service outage. You already sort incidents into severity tiers — Sev 1, Sev 2, Sev 3, or whatever your organization calls them — using a structured set of impact factors: users affected, business criticality, data loss, safety. You don’t eyeball a severity label from the smoke in the room. A major incident triggers a different response entirely: executive notification, a bridge call, a post-incident review. FDA’s own health-hazard evaluation and three-tier recall classification is that same discipline, applied to a blood product: fixed factors, walked deliberately, not a severity guess. Where the regulated world asks for more: your own severity rating is your organization’s call, revisable as you learn more. A recall classification is FDA’s own determination — even for a recall your firm starts on its own — and your internal risk assessment is an input to it, never a substitute for it.

The rule in plain English

What would actually move a 483 toward a Warning Letter. A 483 is an opening move, not a verdict — module 1 already established that. After Lakeshore submits its written response, FDA reviews it. If FDA is satisfied the problems are being promptly and adequately addressed, that’s typically where it ends: corrective action, documented, done. If FDA is not satisfied — the response is weak, incomplete, or the underlying violations are serious enough that a written promise to fix things isn’t enough — FDA can issue a Warning Letter as the next, more serious step. A Warning Letter is far more likely to follow an OAI (Official Action Indicated) classification, or a VAI (Voluntary Action Indicated) classification paired with an inadequate response, than it is to follow a VAI classification that gets a solid corrective-action response. Lakeshore’s own Director reasoned, back in module 1, that this 483 is “probably VAI” — real documentation gaps, but nothing repeated, deliberate, or tied to patient harm. If that reasoning holds, and Lakeshore’s written response is genuinely solid, a Warning Letter is not the expected next step here. It’s worth understanding the mechanism anyway, because “probably VAI” is a working hypothesis, not FDA’s actual classification yet.

A Warning Letter itself is also not what people sometimes assume. It is not a seizure, not an injunction, not a criminal referral — those are formal legal actions, a different and heavier category entirely. A Warning Letter is informal and advisory: a serious, public escalation that puts a firm on notice, but still one deliberate step short of formal legal action.

The recall, market withdrawal, stock recovery distinction — the one people conflate. Not every time a firm pulls a product off a shelf is it a “recall” in FDA’s sense. Three different things can look similar from across the room:

• A recall is a firm’s removal or correction of a marketed product FDA considers to violate the law, where FDA would initiate legal action (seizure) if the firm didn’t act.
• A market withdrawal is removing or correcting a distributed product over a minor violation that wouldn’t draw legal action, or no violation at all — routine stock rotation, routine equipment adjustment, that kind of thing.
• A stock recovery is removing or correcting a product that was never actually released: it’s still on the firm’s own premises, under the firm’s own control, and no part of the lot ever went out the door.

The test isn’t “how bad does this feel.” It’s two concrete questions: did the product ever leave the firm’s direct control (if not, it’s a stock recovery, full stop), and if it did, is the underlying problem a real violation FDA would act on, or something minor or non-violative (if the latter, it’s a market withdrawal, not a recall). Lakeshore’s plasma lot already left Fairview and reached a hospital — that rules out stock recovery immediately. And mislabeling a blood component’s actual identity isn’t a minor labeling quirk; it’s a real violation that puts the wrong information in a clinician’s hands. That rules out market withdrawal too. This is a recall.

Recall classification — walked, not asserted. Once something is a recall, FDA assigns it one of three classes, based on the relative degree of health hazard, not a gut read:

• Class I — reasonable probability of serious adverse health consequences or death.
• Class II — temporary or medically reversible adverse consequences, or a remote probability of serious ones.
• Class III — not likely to cause adverse health consequences at all.

FDA doesn’t jump straight to a class. An ad hoc committee of FDA scientists works through six specific factors first — the same “don’t skip to the conclusion” discipline this curriculum has used before, walking a structured test instead of asserting a severity label. We’ll walk Lakeshore’s plasma lot through all six later in this module.

Classification is FDA’s call — even when the firm recalls itself. A firm can decide, on its own, to pull a product it believes is violative, and notify FDA. That’s a firm-initiated recall, and it’s the common case. But the firm doesn’t get to also assign its own classification. FDA reviews what the firm submits and tells the firm what class it is. Your QA can and should form an internal risk judgment before that call comes back — exactly the way you’d triage severity before a formal incident review confirms it — but that judgment is an input, not the answer.

One more distinction, stated once, because it matters and this module doesn’t develop it further: a recall and a Biological Product Deviation Report (BPDR) are not alternatives to each other. The same event — distributed blood components affected by a deviation — can trigger both. A recall is about correcting or removing the product itself. A BPDR, under 606.171, is a separate reporting duty to FDA about the deviation that caused the problem. Lakeshore’s plasma-labeling event plausibly triggers both obligations side by side. Filing a BPDR — the form, the office, the window — is module 3’s job, not this one.

Citation decoder

Citation	How to read it aloud	What it requires	In our scenario
FDA, “CDER GCP Inspections and Outcomes” (2022), slide 17	“the Warning Letter slide”	FDA informational source (GCP-specific presentation). Describes a Warning Letter’s general character: public (redacted), informal and advisory.	What the instrument actually is if Lakeshore’s response doesn’t satisfy FDA.
21 CFR 7.3(g), (j), (k)	“Title twenty-one, Part seven, Section three, paragraphs g, j, k”	Regulation. Defines recall, market withdrawal, and stock recovery as three distinct things.	Sorting the plasma-lot event into the right one of the three.
21 CFR 7.3(m)	“Section three, paragraph m”	Regulation. Defines Class I, II, and III recall classification by relative health hazard.	The three possible endpoints of the classification exercise below.
21 CFR 7.40	“Section forty”	Regulation. States recall is voluntary, and an alternative to FDA-initiated court action (seizure).	Why Lakeshore recalling on its own is the normal path, not a failure.
21 CFR 7.41	“Section forty-one”	Regulation. The six-factor health hazard evaluation FDA uses to actually derive a classification.	The test we walk, factor by factor, against the plasma lot.
21 CFR 7.46	“Section forty-six”	Regulation. The nine items FDA wants for a firm-initiated recall, and that FDA — not the firm — assigns the classification.	What Lakeshore’s QA needs ready, and who makes the final call.

The actual text

The Warning Letter’s general character, from an FDA presentation on CDER’s (Center for Drug Evaluation and Research) GCP (Good Clinical Practice) inspections — FDA informational source, not guidance or regulation:

Available to the public (redacted). Informal and advisory. An opportunity to improve compliance. Follow-up inspection [may occur].

The same scope caveat module 1 already established, restated briefly: this source is specifically about GCP inspections of clinical drug trials — a different inspection program than the CGMP (current good manufacturing practice) inspection Lakeshore just had. None of that presentation’s clinical-trial-specific content transfers here. What does transfer is the Warning Letter’s own general character on this slide — public, redacted, informal and advisory, an opportunity to improve, often followed by a follow-up inspection — because that’s how a Warning Letter works FDA-wide, not something specific to clinical trials.

21 CFR 7.3 — the three-way definition, regulation:

Recall means a firm’s removal or correction of a marketed product that the Food and Drug Administration considers to be in violation of the laws it administers and against which the agency would initiate legal action, e.g., seizure. Recall does not include a market withdrawal or a stock recovery.

Market withdrawal means a firm’s removal or correction of a distributed product which involves a minor violation that would not be subject to legal action by the Food and Drug Administration or which involves no violation, e.g., normal stock rotation practices, routine equipment adjustments and repairs, etc.

Stock recovery means a firm’s removal or correction of a product that has not been marketed or that has not left the direct control of the firm, i.e., the product is located on premises owned by, or under the control of, the firm and no portion of the lot has been released for sale or use.

21 CFR 7.3(m) — classification, regulation:

Class I is a situation in which there is a reasonable probability that the use of, or exposure to, a violative product will cause serious adverse health consequences or death.

Class II is a situation in which use of, or exposure to, a violative product may cause temporary or medically reversible adverse health consequences or where the probability of serious adverse health consequences is remote.

Class III is a situation in which use of, or exposure to, a violative product is not likely to cause adverse health consequences.

21 CFR 7.40 — recall policy, regulation:

Recall is a voluntary action that takes place because manufacturers and distributors carry out their responsibility to protect the public health and well-being from products that present a risk of injury or gross deception or are otherwise defective. … These sections also recognize that recall is an alternative to a Food and Drug Administration-initiated court action for removing or correcting violative, distributed products…

A request by the Food and Drug Administration that a firm recall a product is reserved for urgent situations…

21 CFR 7.41 — the six factors, regulation:

An evaluation of the health hazard presented by a product being recalled or considered for recall will be conducted by an ad hoc committee of Food and Drug Administration scientists and will take into account, but need not be limited to, the following factors:

1. Whether any disease or injuries have already occurred from the use of the product.

2. Whether any existing conditions could contribute to a clinical situation that could expose humans or animals to a health hazard. …

3. Assessment of hazard to various segments of the population … who are expected to be exposed to the product being considered, with particular attention paid to the hazard to those individuals who may be at greatest risk.

4. Assessment of the degree of seriousness of the health hazard to which the populations at risk would be exposed.

5. Assessment of the likelihood of occurrence of the hazard.

6. Assessment of the consequences (immediate or long-range) of occurrence of the hazard.

21 CFR 7.46 — firm-initiated recall, regulation:

1. … the firm will be asked to provide the Food and Drug Administration the following information: (1) Identity of the product involved. (2) Reason for the removal or correction and the date and circumstances under which the product deficiency or possible deficiency was discovered. (3) Evaluation of the risk associated with the deficiency or possible deficiency.

4. Total amount of such products produced and/or the timespan of the production. (5) Total amount of such products estimated to be in distribution channels. (6) Distribution information, including the number of direct accounts and, where necessary, the identity of the direct accounts. (7) A copy of the firm’s recall communication if any has issued, or a proposed communication if none has issued. (8) Proposed strategy for conducting the recall.
5. Name and telephone number of the firm official who should be contacted concerning the recall.

2. The Food and Drug Administration will review the information submitted, advise the firm of the assigned recall classification, recommend any appropriate changes in the firm’s strategy for the recall, and advise the firm that its recall will be placed in the weekly FDA Enforcement Report.

Walking the plasma lot through the six factors

Lakeshore’s QA has already established this is a recall, not a market withdrawal or stock recovery: the lot left Fairview’s direct control, reached a hospital, and mislabeling a component’s actual identity is a real violation. Now the classification has to be reasoned, not guessed.

1. Have any disease or injuries already occurred? No. The mislabeling was caught during routine label reconciliation before any unit was issued to a patient. This factor alone doesn’t settle the classification — a hazard can be real even with zero occurrences so far, which is exactly why the next five factors exist.
2. Could existing conditions expose someone to a hazard? Yes. If the mislabeled units aren’t retrieved, a clinician could select what’s labeled FFP for a patient who specifically needs FFP’s higher clotting-factor levels — and unknowingly transfuse PF24 instead, or vice versa.
3. Who’s at greatest risk? Most plasma transfusions wouldn’t be meaningfully affected either way. The real risk sits with a narrow group: patients where the specific subtype matters clinically — for example, a patient needing plasma exchange for a factor deficiency where PF24’s lower clotting-factor levels would under-treat the condition. That’s a small population segment, not the whole set of plasma recipients.
4. How serious is the hazard? Not acutely dangerous the way an ABO-incompatible unit would be. The worst realistic outcome is under-treatment of a clotting-factor condition — something a clinician would likely catch through ordinary monitoring (continued bleeding, lab values not responding as expected) and correct with additional product. Temporary and medically reversible, not life-threatening on its own.
5. How likely is it to actually happen? Low. One lot, caught before any unit reached a patient, with the hospital notified before any unit was issued. The specific clinically-sensitive use case is itself a minority of total plasma orders.
6. What are the consequences if it does happen anyway? A patient not receiving the clotting-factor support their treatment plan assumed, requiring reassessment and likely additional product — manageable with monitoring, not catastrophic.

Put together: no occurrence yet, a real but narrow exposure risk, seriousness that tops out at reversible rather than life-threatening, and a low likelihood of actually reaching a patient unretrieved. That’s Class II territory — “temporary or medically reversible adverse health consequences… or where the probability of serious adverse health consequences is remote” — not Class I, because nothing here plausibly causes death or a severe, irreversible injury, and not Class III, because the risk is real enough that leaving the lot in circulation unaddressed would be genuinely unsafe, not merely cosmetic. This is Lakeshore QA’s own reasoned recommendation going into FDA — not the final word. FDA still reviews the submission and assigns the actual class.

What FDA’s own review process actually looks for

When Lakeshore notifies FDA of this firm-initiated recall, FDA isn’t looking for a firm that already feels bad about it. It’s looking for the 7.46(a) list, answered specifically: what the product is, how and when the deficiency was found, a documented risk evaluation (your QA’s six-factor write-up, not a one-line guess), how much product was made, how much is actually out in distribution, which accounts received it, a copy of the recall notification letter sent to the hospital, a proposed strategy, and a named contact. A bad answer sounds like “we think it’s probably fine” with no written evaluation behind it, or distribution counts that are a guess instead of a BECS-pulled number. A good answer is a document trail, not a recollection.

For the Warning Letter thread, FDA’s review looks at the written 483 response itself: does it actually address each observation, propose a credible corrective action, and commit to a real timeline — or does it minimize, argue, or leave gaps. A weak response is exactly what turns a “probably VAI” into something FDA decides needs a stronger message.

Common traps

“We’ve already decided internally this is a Class II, so we can tell the hospital that and move on.” No. 7.46(b) is explicit: FDA reviews what the firm submits and advises the firm of the assigned recall classification — the firm’s own risk evaluation is real, required input, but it is not the classification. Lakeshore’s six-factor reasoning above is a strong, defensible recommendation. It is still a recommendation until FDA responds.

“Any time we pull something back, that’s a recall.” Also no. If the plasma lot had never left Fairview — still sitting in the building, never released — this would be a stock recovery, not a recall, by 7.3(k)’s own terms. If the mislabeling were something genuinely minor, like a typo in a lot number that didn’t affect product identity or safety, that could be a market withdrawal instead. The word “recall” gets used loosely in ordinary conversation for all three. FDA’s own definitions don’t let you use it loosely in an actual filing.

What this means for your job

• Don’t wait for FDA’s classification to start your own six-factor write-up. Your QA’s documented reasoning before FDA responds is what 7.46(a) actually asks for, and it’s the difference between a credible firm and one guessing out loud.
• Teach your QA the three-way sort as a reflex, before the health-hazard factors come up. “Did it ever leave our control, and is the problem a real violation” settles recall vs. market withdrawal vs. stock recovery before anyone argues about severity.
• Treat a VAI-leaning 483 with a strong written response as the de-escalating path it is — but don’t assume “probably VAI” guarantees no Warning Letter. The response itself still has to hold up to FDA’s review.
• When a deviation involves distributed product, ask both questions, not just one: does this need a recall, and does it independently trigger a BPDR? They’re not substitutes.
• Never let “we think this is minor” substitute for the six-factor write-up, even when your gut says Class III. FDA’s committee works the factors; so should you, on paper, first.

Check your understanding

TipQ1: Lakeshore finds a mislabeling error on a box of unused sample tubes still sitting in Fairview’s own supply room — never sent to any collection site, never used on any donor. Is this a recall?

No. By 7.3(k), a stock recovery is a firm’s removal or correction of a product that has not been marketed or has not left the direct control of the firm — located on premises the firm owns or controls, with no portion of the lot released for sale or use. The supply room box never left Fairview’s control and was never released. It doesn’t matter how serious the underlying error is; a product that never left the firm’s own hands by definition isn’t a recall or a market withdrawal.

TipQ2: Your QA says, “We’ve worked through the six factors ourselves and we’re confident this is a Class III — so there’s no need to wait on FDA before telling the hospital what classification this is.” Right or wrong?

Wrong, and this is the point 7.46(b) exists to make unambiguous. Even for a firm-initiated recall — the normal case, where Lakeshore identifies the problem and acts on its own — FDA reviews the submitted information and assigns the recall classification. Your QA’s six-factor reasoning is genuinely valuable and exactly what FDA wants submitted. It’s an input FDA relies on, not a substitute for FDA’s own determination.

TipQ3: Given the Director’s “probably VAI” judgment from module 1, is a Warning Letter the likely next step for Lakeshore’s 483?

No — and this is not a certainty, just the honest reading of where things stand. A Warning Letter is far more associated with an OAI (Official Action Indicated) classification, or a VAI classification paired with a weak or inadequate written response, than with a VAI finding that gets a solid corrective-action response. If “probably VAI” holds and Lakeshore’s response genuinely addresses both observations, a Warning Letter isn’t the expected path here. But FDA hasn’t actually classified the inspection yet — this is reasoned expectation, not a guarantee, and the Director’s own module-1 judgment was explicitly a working hypothesis, not a verdict.

Where this goes next

Lakeshore now has two live threads: a 483 response working its way through FDA’s review, and a recall recommendation of Class II heading into FDA’s own classification decision. Both involve distributed product and a deviation behind it. filing-a-biological-product-deviation-report picks up exactly there — not whether the plasma-labeling event is reportable (an earlier course already built that determination logic), but how to actually file the report: the right form, to the right office, inside the right window.