“No sterility failures, no adverse trends, no complaints” — the reassurance your monitoring cannot give
483 response filed 27 March 2026 · Citations 21 CFR 211.192, 211.22(d), 211.113(b), 211.58
A contract sterile injectables site in Cockeysville, Maryland was inspected for two weeks in early 2026. Investigators followed a line of mould recoveries in the Grade A filling area back to March 2024 and found the investigations closed, the root causes open, and product released. The firm answered inside the window with the strongest evidence it had: nothing had gone wrong since.
FDA did not dispute the record. It disputed what the record could prove. A product impact conclusion built on the absence of a signal is worth only as much as the system that would have raised one.
Earlier issues asked how far a corrective action must reach. This one asks what your reassuring data is evidence of.
What FDA found
Pharmaceutics International, Inc., a Jabil company, fills sterile injectables under contract. The first finding is a sterility failure closed without an answer. The firm recovered a fungus, Ustilago spermophora, rejected the batch — the right disposition — and recorded the root cause as “undetermined.” FDA’s objection is that undetermined is where the investigation stopped, not where it arrived: it “did not thoroughly examine all possible root causes,” among them personnel flow, HVAC and disinfection.
Underneath it runs a trail. Fungal recoveries in the aseptic suites have “significantly increased since 2023” — and each was investigated alone. A May 2025 recovery was attributed to a power failure, though a surface sample taken after power was restored returned a second species. FDA’s objection is not the individual conclusions but that the firm “closed investigations without implementing robust CAPA” and never evaluated the trend they formed.
- 20 Mar 2024
- Mould recovered below the vial machine after filling; batch released.
- 16 May – 4 Jun 2025
- Two further recoveries — the second the same species, on the same line.
- 23 Feb – 6 Mar 2026
- FDA inspects. Sterility failure root cause recorded “undetermined.”
- 27 Aug 2026
- Warning letter. Every response inadequate; consultant recommended.
What the firm said
This is the response most firms write, and on its face it is the responsible one. The firm committed to strengthening investigations, a retrospective review, a revised contamination control strategy and routine trend analysis. It went back through its monitoring data to 2024 and reported no adverse trends. Then it answered the question every quality unit is asked about released product: look at what happened next. No sterility failures. No adverse trends. No complaints.
The same reasoning appears inside one of the investigations, in the firm’s words:
“historically there has never been a mold hit isolated on the grade A space on the [redacted] filling Line” Firm’s investigation record, as quoted by FDA
It reads as diligence — someone checked the history first. FDA supplies the next sentence: the same species, same line, fifteen months earlier.
Why FDA rejected it
Absence of a signal is not a finding
FDA does not say the downstream data was wrong. It says the firm reasoned from it instead of doing the work — concluding no product impact while the contamination source stayed unidentified.
“Your response is inadequate because you indicate that product impact conclusions in investigations were appropriate based on a lack of sterility failures, adverse environmental monitoring trends, and complaints, but fail to sufficiently address contamination sources.” FDA, Warning Letter 320-26-120
The clean dataset and the method that produced it
The EM review since 2024 showed no adverse trends. FDA answers in a line: “the quality of the data was compromised by insufficient technique and locations.” Investigators watched swabbing that skipped part of a surface and never covered the maximum accessible area. A programme that cannot reach the contamination cannot report it. Its silence is not a result.
An assertion offered without its basis
Peeling paint and plastic debris were found in classified areas; the firm’s evaluation called them “non-viable particulates.” FDA’s objection is not that this is false — it is that “you do not explain the basis for the assertion that paint debris could not harbor microorganisms,” and never asks whether bubbling paint fed the mould.
A record that could not have supported the conclusion
A February 2026 batch record documented twelve aseptic interventions where records indicated roughly two hundred occurred. Incomplete records, FDA notes, “fundamentally compromise your ability to reliably conduct batch record review, to adequately investigate deviations and batch failures.” A no-impact conclusion drawn from it is not conservative. It is unsupported.
The pattern worth carrying forward
Every 483 response arrives at product impact, under time pressure, for batches that shipped months ago. The honest answer is very often that nothing bad has happened since — and the temptation is to let that stand.
FDA reads it the other way round. No complaint, no failure, no adverse trend is not an observation about the product; it is an observation about your detection systems. Negative evidence is only as strong as the instrument that produced it — and here those instruments were the monitoring programme, the investigation process and the batch record, each of which the same letter finds deficient. Three silences, from three systems under citation. That is not corroboration. It is the finding, restated.
- Is this built on what we found, or on what we did not find? Identified source, identified route, bounded exposure is a finding. No complaints, no failures, no trends is an absence. Lead with the first; offer the second as corroboration, and say which is which.
- Could the system we are citing have detected it? If your sampling locations, technique, limits or trending rules are part of the observation, their reassuring output cannot close it. Fix the instrument before you quote its reading.
- Is the basis attached? “Non-viable particulate,” “not product contact,” “historically never seen” — each is a testable claim. Attach the data, study or record search behind it — and check the history before asserting there isn’t any.
A defensible response would have kept the two claims apart. First the positive case: the source identified, the route into the Grade A zone, the population it could have reached, and what corrected monitoring now shows. Only then the downstream data — support for a conclusion already reached on evidence, never the conclusion itself.
“Nothing went wrong” is not a statement about your product. It is a statement about what you could see.
Veritas Quality Consultants works with pharmaceutical, medical device, and biologics manufacturers on Form 483 responses, root cause analysis, and CAPA development — built to FDA expectations inside the 15-day window.
Request a consultation