Case Study · Multi-Site Quality & the Read-Across Failure

A kids’ gummy called CAPA Crunch™ — three quality systems, and not one of them caught a 40× overdose.

Vanteon didn’t lack quality systems — it had three, across three sites, that couldn’t see each other. A single dropped unit conversion (international units to micrograms, a factor of 40) lived as three orphaned records in three separate systems, connected by none — and shipped a children’s vitamin-D gummy at forty times its label dose.

A constructed teaching case · built from real regulatory patterns, not a real company

CAPA Crunch kids immune gummy stand-up pouch, front and back — The product at the center of the file — a children’s immune-support gummy made across three sites, two working units, and one label.

Why we built this one

A file like this follows a pattern we know well. Not this company — Vanteon, its sites, its lot numbers are invented — but the pattern: a brand acquired into a larger group, three plants each with its own quality system, and a single defect that lives in all three and is owned by none.

This is the opposite of a firm too small to have systems. Vanteon has too many that can’t see each other. A vitamin-D unit error — “400” in international units read as 400 micrograms — passed from one site’s spec into another’s batch sheet, and every gate checked the batch-average, the central paper, or the supplier’s assurance — never the gummy a child eats.

So we built the case, because the lesson is worth more than the recall it usually arrives with.

— Regulatory Options

Read this if it sounds like your operation

You run more than one site, or grew by acquisition, and each plant kept its own QMS, LIMS, or unit conventions.
The same defect could be recorded in two systems and never connected — you have no enterprise CAPA read-across.
Your management review shows all green while sites each count the same event as a first occurrence.
You fortify a children’s product and release it on a composite (pooled) batch-average assay.

If two or more are true, this case is describing your operation, not a stranger’s.

The file, as it arrived

Every record behind the case — each one signed, formatted, and internally consistent. Each one, read on its own, looks like a company doing its job.

A network like this fails between documents and between systems, never inside one. Open the file and read what you’d catch — then see what we found, just below.

Open the full case file

What the records showed, together

No single page is alarming — read apart, each record looks like compliance. Read together, the same evidence resolves into four threads, all turning on one fact: three systems that couldn’t see each other.

Three systems, one defect no one could see.

The defect that named the case was a dropped unit conversion — and the reason it shipped is that nothing in the network was built to connect three sites’ records.

No enterprise CAPA read-across — the same vitamin-D unit defect was recorded at all three sites over 14 months and never connected. · 21 CFR 111.135 / 111.140

One specification, three live “effective” revisions — IU at one site, micrograms at corporate, dual-unit at a third; no authority could say which superseded. · 21 CFR 111.95(b)

Records that don’t reconcile across systems — three QMS/LIMS re-keyed by hand in non-constant units, with no cross-unit verification. · 21 CFR 111.95 / 111.260

A disqualified pectin supplier still in active use at two sites — the disqualification never propagated. · 21 CFR 111.75(a)(2) / 111.73

A children’s gummy at forty times its dose.

The unit error became a real lot: vitamin-D3 entered at 400 micrograms — the international-unit numeral — against a 10-microgram target, and every downstream gate passed it.

A lot formulated at 40× labeled vitamin D3 — ~16,000 IU per serving, an acute pediatric hypervitaminosis-D exposure. · 21 CFR 111.70(e) / 111.75

A true out-of-spec result retested into compliance against the mis-entered limit — invalidated “preparation error,” no assignable cause. · 21 CFR 111.113

A serious adverse event — a child hospitalized with hypercalcemia — filed “non-serious”; the 15-business-day report never sent. · AER Act · 21 U.S.C. 379aa-1

The recall scoped to a single lot, with other sites’ lots under the same listing never assessed against the systemic mechanism. · 21 CFR 7.42

A process that makes overdose units by design.

Even with the unit math right, the gummy itself is the risk: vitamin D3 is a trace charge across a 600,000-gummy batch, and release is judged on a pooled average that’s mute about individual gummies.

Content uniformity of the micro-dosed D3 across deposited gummies was never characterized — uniformity merely assumed from a 15-minute mix. · 21 CFR 111.70(b)/(e)

A composite (pooled) release and stability assay masks unit-to-unit variability — 20 gummies ground to one batch-average, no single-gummy assay. · 21 CFR 111.75

Three sites, three depositors, three overages (+10/+18/+15%), no cross-site equivalence study — one label, three real dose distributions. · 21 CFR 111.27

The “%DV” math is built on adult reference values — a declared 15 mg zinc actually exceeds the pediatric tolerable upper limit for ages 4–8. · 21 CFR 101.36

A launch decided, not determined.

The product’s classification was asserted rather than analyzed — and a stack of premarket exposures inherited the unmade decision.

The conventional-food vs. supplement classification was never performed — a candy-like children’s gummy proceeded on the supplement basis by assumption. · FD&C §201(ff)(2)(B)

Color additives (titanium dioxide, vegetable-juice color) mis-screened as “GRAS” — colors added to food are color additives, which GRAS cannot establish. · FD&C §721 · 21 CFR 73.575

Disease claims across channels — “fewer sick days,” “fight off colds & flu” — pushing an unapproved-drug status, with no pediatric substantiation. · FD&C §201(g) / §403(r)

Lead at/above the Prop 65 daily limit with no warning, and child-resistant packaging never even screened. · CA H&SC §25249.6 · PPPA / 16 CFR 1700

Read apart, every record was defensible — a controlled spec, a passing composite assay, an all-green management review, a converter’s letter of guarantee. Read together, they describe a company that owned three quality systems and could account for none of them: a 40× overdose that lived in all three and was owned by none, a process that makes hot units invisibly, and a launch whose classification was assumed. The defect was closed against one site. The hospitalization was filed “non-serious.” The dashboard stayed green.

None of it needed a new test to see. It needed one owner of the chain from COA to label, a read-across that crosses sites, and a release that looks at the gummy, not the batch-average — the things every other check agreed to skip.

What was at stake

What actually happened

An overdose lot, a hospitalization, and a Form FDA 483

A children’s gummy shipped at forty times its vitamin-D dose, a child was hospitalized, and the recall couldn’t bound the affected lots. The 483 and the recall followed the same fragmented paperwork that had passed.

What catches it first

One owner, one read-across, one unit

A single owner of the COA-to-label chain and an enterprise read-across connect the three orphaned records before the lot ships; a unit-resolved release reads the gummy a child eats, not the pooled average.

If this is your operation

You don’t need us to find out whether your network has the same blind spot. Most of it you can check yourself, this week, with records you already have.

Could the same defect be recorded in two of your systems and never connected?— or do you have an enterprise read-across that crosses every site?
Does one person own the chain from supplier COA to finished label?— or is it diffused across plants and units no one reconciles?
Does your release read individual units of a fortified product?— or only a pooled, composite batch-average?
Did someone determine your product’s classification — food or supplement — with the analysis?— or was it assumed, with GRAS, claims, and Prop 65 inheriting the unmade call?

Nothing here is built around this one brand — each door points somewhere bigger, and no one of them is the “right” one. The case is just where they all meet.

Understand the process · Technology Gummy & pastille depositing How a dosed gummy is actually made — and why a micro-dosed active sets the dose at the deposit, where a pooled batch-average can’t see the hot unit a child eats. See the reference Check the controls · Service Process Validation & Content-Uniformity Review For fortified, micro-dosed products: whether the process is validated to make a uniform dose across deposited units and across sites — not judged on a pooled batch-average. See the service Implement the system · Program Multi-Site cGMP & Quality Governance Program Harmonized specs and units, an enterprise CAPA read-across, single-owner accountability, and unit-level release — a buy-and-go document system you run at your own pace. Coming soon Grade your readiness · Assessment Multi-Site Quality-System Readiness Assessment A rubric-graded review of whether your network can connect the same defect across sites — CAPA read-across, spec and unit control, and single-owner accountability, scored against the standard. Coming soon Learn the concept · Training Why three quality systems can be worse than one The idea this case turns on — that fragmentation lets a defect live in every system and be owned by none — is what our forthcoming multi-site quality course will teach. Coming soon

Three quality systems were never the same thing as one that could see across them.