A laboratory record can show that a sample passed without proving the sample was ever collected. Two recent FDA warning letters show how far apart those two things can drift, and what the agency expects a quality control lab to do about it.
On October 14, 2026, FDA sent a warning letter to a manufacturer of over-the-counter ophthalmic drugs. The letter described environmental and personnel-monitoring plates that went to the laboratory unexposed, as though sampling had taken place. It also reported air samples collected at the wrong locations but recorded as correct, and cleaning recorded as complete when management confirmed it had not been done. FDA called the findings "systemic data falsification."
A September 29 letter to a US manufacturer of injectable drugs described failing cephalosporin environmental samples discarded before further analysis. It also described a sterility failure invalidated without adequate scientific justification.
Both letters come down to one question for QC teams: can the lab reconstruct what actually happened, including when a result was unfavorable? When records are unreliable, the damage spreads to environmental trending, investigations, product-impact assessments and the evidence behind batch release.
The ophthalmic manufacturer: a recorded result with no activity behind it
The first letter followed an April 2026 inspection of a facility that makes OTC ophthalmic products by aseptic processing. FDA cited incomplete laboratory records under 21 CFR 211.194(a) and cleaning and maintenance deficiencies under 21 CFR 211.67(a).
According to the letter, the facility repeatedly sent unexposed surface-monitoring plates to the laboratory without doing the required sampling. Personnel-monitoring plates also went in unexposed while the records said sampling had occurred. Active air samples were taken at the wrong locations and documented as taken at the required ones.
The gap showed when FDA compared the history with readings taken during the inspection. The facility had reported zero colony-forming units across environmental monitoring (EM) and personnel monitoring (PM), apart from one deviation after a water leak. Plates read during April 10–14 showed microbial growth on numerous samples, including about 25 ISO 5 excursions.
The company's response then leaned on that history. FDA noted that the response cited historical viable EM data and the absence of adverse trends in its product-impact assessment, despite acknowledging falsification. The agency's reply: "Falsified EM data cannot be used for trending analysis."
A clean trend is only as good as the observations under it. Before historical results support a decision about control or product impact, the organization has to establish that the activities behind them took place.
The injectables manufacturer: discarded samples and contradictory sterility records
The second letter followed a March–April 2026 inspection of a US facility that makes injectable drug products. Under 21 CFR 211.42(c), covering separation and monitoring for cephalosporin cross-contamination, FDA stated that the facility "mistakenly discarded multiple failing cephalosporin environmental samples before further analysis could be conducted."
These samples monitored beta-lactam containment. That is a different hazard and a different test from the microbial plates in the ophthalmic case, even though both sit under environmental monitoring.
FDA asked for full investigations and data on failing cephalosporin samples discarded since August 2024, including location, date, recovery amount and any further testing that had been planned. It also asked for an assessment of documentation systems and remediation so records are attributable, legible, complete, original, accurate and contemporaneous.
A second finding, under 21 CFR 211.192, concerned a 2025 sterility failure on a cephalosporin product. Signed and dated test records first indicated that container integrity had been verified. After turbidity appeared, the investigation stated that integrity had not been verified and classified the event as a "no-test." FDA said the company invalidated the result without adequate scientific justification, did not conduct a Phase II investigation and released the lot in July 2025.
FDA also pointed to the lack of contemporaneous documentation for any deviation that could explain the contradiction. Laboratory error was possible. The company still had to show evidence for it before invalidating the result.
What should QC teams check in their own workflows?
Start with four areas the letters put under pressure: how activities are recorded, how samples are reconciled, how failures are investigated, and whether investigation activity tracks quality signals. The remediation FDA requested is specific to each facility, so treat it as a guide to where to look rather than a new set of requirements.
Record the activity when and where it happens
Both letters ask for contemporaneous, attributable records. For sampling, check whether the record establishes who sampled, the actual location, the sample identity and the time of each step. A record with every required field filled in can still describe sampling that never happened, as the ophthalmic manufacturer's did. FDA's requests there also go past documentation to independent verification of sampling custody and cleaning execution.
Reconcile every sample through its lifecycle
FDA asked the ophthalmic manufacturer for uninterrupted custody and full reconciliation of microbiological samples. That covers unique labels, signatures from everyone who handles a sample, dates and times for collection and each later step, digital time-stamped photographs of all plates, documented readings and routine QA oversight.
The test for your own lab: does every issued plate or collected sample have an explainable outcome? A sample that was not collected, was damaged, went missing or was discarded should stay visible in the record with its status and disposition. How you get there belongs in your approved sample-handling procedure.
Investigate failures before invalidating them
The injectables sterility finding shows why the first observation has to be kept alongside the evidence for any later conclusion. FDA asked for a broader assessment of investigation procedures, scientific rigor, scope, root-cause evaluation, staff competencies and quality unit oversight across all phases.
For each invalidation, ask whether it rests on demonstrated evidence or on an explanation chosen after the result came back unfavorable. Procedures should also separate event types. An analytical out-of-specification result, a sterility failure and an EM excursion each need their own kind of investigation.
Compare quality signals with investigation activity
FDA noted that the injectables manufacturer's leak rate rose every year from 2023 to 2026 while the number of nonconformance reports opened fell. It challenged both the scope of the retrospective review and the adequacy of corrective and preventive action.
That suggests a simple cross-check. Plot defect, monitoring and complaint trends against the investigations they generate. If adverse signals rise while investigations fall, find out whether reporting and escalation practices explain it before reading it as improvement.
Supplier COAs still need verification
An October 1 warning letter to a cosmetics company that also makes OTC drug products adds a related example from incoming-material testing. Citing 21 CFR 211.84(d)(1) and (d)(2), FDA described inadequate identity and impurity testing. It also described reliance on supplier certificates of analysis without checking, at appropriate intervals, that the suppliers' results were reliable.
Supplier COAs can still be used. FDA asked for at least one specific identity test on each incoming component lot. Where supplier results replace testing for strength, quality and purity, it asked for initial validation and periodic revalidation of those results. The findings concern OTC drug manufacturing and don't extend drug CGMP requirements to cosmetics in general.
What this means for QC teams
Run one test on your own evidence. If a failing sample disappeared today, could you say what was lost, who handled it, why it was discarded and which quality decisions it touched? If any of those answers depends on someone's memory, start there.
Digital records can help with part of this. Sample identifiers, timestamps, user attribution, custody records, disposition histories and audit trails make missing activities, unexplained changes and discarded results visible to reviewers. They can't show that an operator sampled the right surface or finished a cleaning task. That still takes the independent verification, investigation rigor and quality unit authority FDA asked for. For more on how structured records support integrity across the lab, see strengthening data integrity in enterprise R&D.
How Uncountable approaches QC lab records
Uncountable's LIMS for QC captures sample IDs, custody steps, timestamps and dispositions as structured records with a full audit trail, so a missing or discarded sample stays visible at review. Environmental monitoring workflows, including plate issue and reconciliation, run on the same records. For the wider compliance picture, see our GxP compliance page.
.png)
.png)
