Never miss the enforcement signal that hits your SOPs

Data-integrity findings FDA cites to sterile injectable / aseptic drug manufacturers

Scope: FDA warning letters (CDER) and Form 483 inspectional observations issued to sterile/aseptic drug product makers and closely adjacent API/QC-lab operations. The findings cluster into seven recurring categories, ordered roughly by how frequently they appear across the reviewed enforcement documents.

1. Missing, disabled, or unreviewed audit trails (most common)

The single most frequently cited data-integrity theme, with two faces: systems configured without working audit trails, and audit trails that existed but were never reviewed before batch release.

• HPLC operating with no activated audit trail (no record of injection type, date/time, analyst, action, reason) ⁴⁰; audit trail feature not enabled on the chromatography data management system ⁴¹; GC, FTIR, and UV systems with no audit trail capability ⁵².
• GC software lacking active audit trail functions, with raw data deleted after chromatograms were printed ⁶³; Empower message-center audit trails purged without justification or backup ¹³²; analysts manipulated and deleted audit trails themselves ⁶⁰.
• No / inadequate review: procedures existed but no reviews had been performed before inspection ⁴⁷; only a subset planned for review, which FDA said was inadequate because full audit trail data must be reviewed prior to batch release ⁴³; of many equipment systems, only two had ever been reviewed ¹³³; "fail to conduct an instrument audit trail review at any frequency" ¹⁴⁶.

2. Weak computer-system access controls (shared logins, analyst admin rights)

FDA's core objection is loss of attributability: actions cannot be traced to an individual, and users can alter or delete data.

• All analysts sharing a common login ID and password with full administrator rights ⁸²; shared HPLC/XRD login IDs and shared Windows credentials for GC workstations ⁸¹.
• Sterile/aseptic contexts: aseptic filling-line HMI using a shared username/password with no audit trail for recipe changes ²¹; operators/supervisors/managers sharing one User ID for HMI access with rights to edit process time and set-points ⁴; shared general laboratory Windows login for sterility sample-result systems, with ability to add/delete files ²³; a non-reviewing "Analyst" user able to change a pH result at a sterile CMO ⁷.
• Analysts / QA with administrator rights able to delete projects, data, results, and methods ⁹; QA Vice Manager logged in as administrator with unlimited ability to change/delete records ¹³.

3. Deleting, overwriting, aborting, or reprocessing analytical data

• Sample-sequence deletions and re-analysis on another HPLC; firm acknowledged deleting chromatograms ⁴²; audit trail showing 1,441 deleted results, 3,643 manual integrations, 194 altered sample sets ⁵⁰; analyst set the PC clock back, deleted the first four injections, and reported only the favorable result ⁵⁴; 2,404 injections found in an "Experimental" folder with deleted raw data ⁶².
• Manual reintegration turning a failing stability result into a passing one used to extend shelf life ⁸⁹; failing GC injection excluded without documentation, only the passing injection reported ⁹²; deletable QC data with no backup ⁹⁵.

4. Unofficial / "trial" injections and testing into compliance (retest-until-pass)

A hallmark sterile-injectable QC finding: actual samples run as "trial," "test," "prep," or "experimental" injections before the "official" run, then failing data discarded.

• Multiple single "trial" injections without justification, using actual samples to disguise testing into compliance ⁶⁷; OOS sample reinjected ~14 hours later in a separate series omitted from the OOS investigation, injections labeled "Experimental" ⁶⁸; trial stability injections acquired in a "Test" folder before official testing with raw data deleted ⁷².
• OOS potency lot re-injected, still OOS, then reprepared and retested until all four results passed, released with no investigation (outsourcing facility handling injectables) ¹⁵¹; trial result OOS while official results invalidated after a single passing retest ¹⁵⁷¹⁶⁶.

5. Falsification, fabrication, and backdating (most serious; often sterility/EM records)

• Sterility tests never performed, records fabricated to show two-analyst signoff and "no growth" for samples that did not exist; no personnel, viable-air, or surface monitoring samples collected for the dates shown ¹²⁷.
• Non-contemporaneous approval ("backdating") of environmental monitoring records ¹⁰⁸; admitted backdating of a "Microbiological Examination of Finished Product" document ¹¹⁴; QC employee backdated multiple forms and signed with another person's initials ¹¹⁹.
• OOS results substituted with passing results via cut-and-paste chromatograms, vial substitution, and changed sample weights ¹²⁰; sterility test data sheet torn up and rewritten ¹⁴.

6. Failure to record contemporaneously; incomplete / uncontrolled records

• Pre-signed blank preventive-maintenance documents and backdated reconciliation records ³⁰; two garbage bags of torn analytical records and staff using personal diaries to record results ³⁰; manipulated manual visual-inspection records for injectable batches so defect counts stayed within limits ³².
• Aseptic entry/exit logs not accurate/complete/contemporaneous, with names signed for days not present ¹⁸⁵; production raw data recorded on uncontrolled yellow paper and later inserted into the batch record ¹⁸⁴; torn uncontrolled GMP paper recovered from scrap — original records, raw data, metadata ¹⁷⁷¹⁷⁸; scrapyard torn pieces of sterility printouts, EM printouts, filter-integrity printouts, and BMR pages ¹⁹⁰.

7. Environmental monitoring, media fill, and microbiology data integrity (sterility-specific)

The most sterility-specific category, and the one where data-integrity lapses most directly threaten sterility assurance.

• Operators manipulated EM and nonviable-particle data: plates not consistently contacting surfaces, equipment wiped before sampling, ISO 5 Grade A particle data reported as ISO 7 Grade B ¹⁶⁷.
• EM samples falsely reported as collected (plates incubated without ever being exposed); EM records showing no excursions while 12 plates required action; the firm confirmed the data were fraudulent ¹⁶⁸; media-fill record falsified (operator signed "checked by" for 63 units without being present) ¹⁶⁸.
• Team leader instructed staff to discard bioburden plates with growth, manipulated camera timestamps to backdate records, and left an endotoxin failure uninvestigated; microbiology logbook pages removed and replaced with fabricated pages ¹⁷².

How to read this

The pattern FDA describes is consistent: absent or unreviewed audit trails and shared/over-privileged system access remove the controls that would normally prevent or detect manipulation, and the manipulation then shows up as deleted/aborted injections, "trial" runs, testing into compliance, and — most seriously — fabricated or backdated sterility, media-fill, and environmental-monitoring records. In the sterile injectable setting, the microbiology/EM findings carry the greatest weight because they bear directly on sterility assurance.