Discarded Source Data in Media Fills: Lessons from Sun Pharma's FDA 483
Original inspection data discarded. No second-person verification. No mechanism to confirm what was transcribed matched what was observed. The Sun Pharma 483 reveals a documentation architecture where source data is treated as disposable.
On June 13, 2025, the FDA issued a 483 observation to Sun Pharmaceutical Industries Ltd. at their Halol facility in Gujarat, India. The finding targeted a fundamental gap in batch record integrity: “Batch production and control records do not include complete information relating to the production and control of each batch.”
The specifics centred on media fill operations — the aseptic process simulations that validate a facility’s ability to produce sterile products without contamination. During visual inspection of media fill units, operators recorded initial results on labels. Those labels were then transcribed into the batch record. And then the labels were discarded. The original source data — the direct observation record — ceased to exist the moment it was copied.
The facility’s SOP (SOP032025, V. 2.0, effective 26-Feb-2024) did not require retention of these labels. It provided no alternative mechanism for verifying the accuracy of the transcribed data. And there was no documented second-person verification process to ensure what was written on the label matched what was entered into the batch record. The transcription was a one-way, single-person, unverifiable transfer of critical quality data.
When source data is discarded after transcription and no verification mechanism exists, the batch record doesn’t contain original data — it contains an unverifiable copy. That distinction matters to the FDA, and it should matter to every quality leader running manual documentation workflows.
What the FDA Found
The observation wasn't about missing batch records. It was about a documentation workflow that systematically destroyed the evidence needed to prove those records were accurate.
21 CFR 211.188 requires that batch production and control records include complete information relating to the production and control of each batch. “Complete” is the operative word. A batch record that contains transcribed data with no way to verify accuracy against the original observation is not complete — it is an assertion without evidence.
The ALCOA+ framework makes this explicit. The “O” stands for Original — data must be the first capture, or a verified true copy. At Sun Pharma’s Halol facility, neither condition was met. The original data (inspection labels) was destroyed. The copy (batch record entry) was unverified. The SOP that governed this process not only permitted this outcome — it practically guaranteed it by requiring neither retention nor verification.
Media fills are among the most scrutinised operations in pharmaceutical manufacturing. They exist to demonstrate aseptic process control. When the documentation of media fill results depends on a transcription workflow that discards source data and provides no verification, the integrity of the entire simulation is called into question — not because the results were wrong, but because there is no way to demonstrate they were right.
0 labels
Source Data Retained
Original inspection labels used to record visual inspection results during media fills were discarded after transcription into batch records — eliminating the only source data.
0 steps
Second-Person Verification
No documented second-person verification process existed to ensure accurate transfer of inspection results from labels to the batch record. Transcription was single-operator, single-step.
1 SOP
Codified the Gap
SOP032025 (V. 2.0, eff. 26-Feb-2024) did not require retention of source labels or provide any alternative mechanism for verifying the accuracy of transcribed data.
Why This Keeps Happening
Sun Pharma's Halol facility isn't unique. The transcription-and-discard pattern is embedded in how most pharmaceutical facilities handle intermediate documentation.
The root cause isn’t a careless operator or a poorly written SOP. It’s a documentation architecture that treats intermediate data as temporary — something that exists only long enough to be moved somewhere else. When that architecture meets the FDA’s expectation of verifiable, original, complete records, the gap is structural and unavoidable.
Transcription-based workflows are inherently lossy.
Every manual transcription step is a point where data can be altered, omitted, or misread. When an operator copies results from a label to a batch record, the transfer depends entirely on human accuracy under production conditions. No system confirms the copy matches the source. And in this case, the source was then destroyed — eliminating any possibility of retrospective verification.
SOPs don't require source data retention.
Many pharmaceutical SOPs treat intermediate documentation — labels, worksheets, sticky notes, logbook drafts — as temporary artifacts. The assumption is that once data reaches the batch record, the intermediate document has served its purpose. But ALCOA+ requires either the original or a verified true copy. An SOP that permits destruction of source data without verification is an SOP that permits non-compliance.
No second-person verification is built into the process.
Second-person verification of transcription is a basic data integrity control. But in manual workflows, adding a verification step means adding another person, another signature, and another scheduling dependency — all for a step that 'should' be straightforward. So it gets omitted, deferred, or treated as optional. At Halol, it simply didn't exist.
Paper-to-paper transfer has no digital intermediate.
When data moves from one paper document to another paper document, there is no audit trail of the transfer itself. No timestamp showing when the transcription occurred. No record of who performed it. No mechanism to flag discrepancies. The transfer is invisible to the quality system — a black box between observation and record.
The question isn’t whether your operators transcribe data accurately. It’s whether your documentation architecture makes inaccurate transcription detectable — or whether it destroys the evidence needed to find out.
Manual Transcription vs Digital Capture
The difference between a documentation workflow that creates data integrity risk and one that eliminates it is not about diligence — it's about where and how data is first recorded.
In each comparison below, the digital approach doesn’t just reduce effort — it removes the transcription step entirely, ensuring the batch record contains original data rather than an unverifiable copy of discarded source material.
Data Capture at Source
Operator records inspection results on a temporary label during media fill visual inspection. Results are later transcribed into the batch record by the same operator. The label — the only original record of the observation — is then discarded per SOP. The batch record contains secondhand data with no link to the original.
Result: Unverifiable copy in the batch record
Operator records inspection results directly into the electronic batch record at the point of observation. The entry is timestamped, attributed to the operator, and stored as the original record. No transcription step exists. No intermediate document is created or destroyed. The batch record is the source.
Result: Original data captured directly
Verification Workflow
No second-person verification process is documented. A single operator transcribes results from label to batch record without independent confirmation. Errors in transcription are undetectable because the source label no longer exists. Verification would require a second person to be present during transcription — a logistical burden most facilities don't implement.
Result: Single-point-of-failure documentation
Electronic workflows enforce review and approval steps as part of the data entry process. A second person can verify entries against real-time data without needing to be physically present during the original observation. The system maintains a complete audit trail of every entry, review, and approval — with timestamps and electronic signatures.
Result: Built-in verification by design
Source Data Retention
Intermediate documentation (labels, worksheets) is discarded after transcription. The SOP permits this because the batch record is considered the 'final' document. But under ALCOA+, a transcribed copy of discarded original data is neither Original nor verifiably Accurate. The batch record looks complete but cannot be validated against its source.
Result: Source data permanently lost
All data is retained in its original electronic form with a complete audit trail. There is no 'intermediate' document to discard because the first entry is the final record. Version history, timestamps, and attribution are preserved automatically. Any modification is tracked, reason-coded, and available for review indefinitely.
Result: Permanent, auditable data retention
What Modern Batch Records Must Do
Preventing source-data destruction isn't about better SOPs or additional training. It's about eliminating the transcription step that creates the vulnerability in the first place.
The capabilities below directly address every gap identified in the Sun Pharma observation — not by adding more documentation steps to an already strained workflow, but by redesigning how inspection data enters the batch record.
Single Point of Entry
Inspection results are entered directly into the electronic batch record at the point of observation. No labels, no worksheets, no intermediate documents. The first record is the final record — eliminating the transcription step and the source-data destruction that follows it.
Enforced Verification
Electronic workflows require second-person verification before batch record entries are finalised. Review steps are built into the process, not bolted on as optional SOP provisions. The system will not advance until verification is complete — making unverified transcription architecturally impossible.
Immutable Audit Trail
Every entry, modification, and review action is captured with timestamps, operator identity, and electronic signatures in a tamper-proof audit trail. Source data is never overwritten or discarded — it is preserved in its original form, meeting ALCOA+ requirements for Originality and Accuracy without relying on human discipline.
20→1 days
Batch Review Time
Valent BioSciences reduced batch review from 20 days to 1 day after implementing electronic batch records with automated data capture — because reviewers no longer need to chase down source documents.
2,700 hrs
Saved Annually
Annual hours saved across Valent BioSciences' operations by eliminating manual transcription, paper-based verification, and the reconciliation workflows that manual documentation demands.
60%
Fewer Manual Entries
Reduction in manual data entries at Valent BioSciences — each one a transcription step eliminated, a source-data destruction risk removed, and a potential 483 observation prevented.
From Gap to Prevention
Three phases to eliminate the transcription-and-discard pattern that created Sun Pharma's FDA 483 risk.
The goal isn’t to add retention requirements to existing SOPs or introduce manual verification checkpoints. It’s to remove the transcription step entirely — making source-data destruction impossible because there is no intermediate source document to destroy.
Phase 1: Map every transcription point.
Audit your current documentation workflows for every instance where data is recorded on an intermediate document and later transcribed into the batch record. Media fill labels are one example, but the pattern exists across weighing slips, environmental monitoring worksheets, in-process check forms, and cleaning verification records. Each transcription point is a potential 483. Document which source materials are retained, which are discarded, and where second-person verification exists — or doesn't.
Phase 2: Eliminate transcription through digital capture.
Deploy electronic batch records that capture data at the point of observation. Configure workflows so that inspection results, test outcomes, and process parameters are entered directly into the system — with timestamps, operator attribution, and electronic signatures applied automatically. Process interlocks should prevent advancement until all required data points are captured and verified, making incomplete or unverified entries impossible rather than merely non-compliant.
Phase 3: Validate and demonstrate control.
Validate the electronic system per 21 CFR Part 11 requirements. Establish metrics that prove the transcription gap has been closed: zero intermediate documents discarded, 100% second-person verification completion, and full audit trail coverage for every batch record entry. Within 90 days, the data should demonstrate that the documentation architecture — not operator behaviour — is what ensures data integrity.
Sun Pharma’s FDA 483 wasn’t caused by operators who failed to save labels. It was caused by a documentation architecture that told them not to — and provided no alternative for verifying that what reached the batch record matched what was actually observed. The fix isn’t a revised SOP. It’s a system where source data and the batch record are the same thing.
The pharmaceutical industry’s data integrity failures are rarely about falsification. They are about documentation architectures that create unverifiable records through entirely sanctioned workflows. An operator follows an approved SOP, discards a label as instructed, enters data into a batch record as trained — and the result is a record that cannot be validated against its source. Not because anyone did anything wrong, but because the system was designed to produce exactly this outcome.
The Sun Pharma observation at Halol is a precise illustration of this pattern. Media fill visual inspection — one of the most critical aseptic process validations — depended on a transcription workflow that destroyed original data and provided no verification mechanism. The SOP permitted it. The process normalised it. And the FDA cited it, because 21 CFR 211.188 and ALCOA+ principles do not make exceptions for data loss that happens by design.
Modern electronic batch records eliminate this class of risk entirely. When inspection results are captured directly into the system at the point of observation, there is no transcription step, no intermediate document to discard, and no verification gap to close. The batch record contains original data — attributable, timestamped, and immutable — because the architecture makes that the only possible outcome. The operators do less documentation work. The records contain more verifiable data. And when the FDA arrives, the evidence of data integrity isn’t reconstructed from memory of discarded labels — it is there, complete and traceable, captured at the moment the observation was made.
Related Articles
Three Facilities, Three FDA Actions, Five Architectural Gaps: How AI Agents Address Cipla's Regulatory Exposure
Between 2023 and 2026, three Cipla facilities — Pithampur, Raigad, and Pharmathen Greece — received FDA enforcement actions documenting the same five systemic failures: complaint investigation, CAPA effectiveness, electronic data review, contamination control, and QC oversight. LeucineOS AI agents map directly to each gap.
35% OOS Invalidations, Zero Scientific Justification: Lessons from Aurobindo Pharma's FDA 483
A February 2026 FDA 483 at Aurobindo Pharma's Unit-III found 35% of OOS invalidations in the QC Chemistry lab — with 57% blamed on analyst error and 18% on equipment, none supported by adequate scientific justification. Batches shipped to the US after unresolved Grade A maintenance interventions.
Equipment Swapped, Cleaning Not Revalidated, OOS Dissolved Away: Lessons from Dr. Reddy's FDA 483
A December 2025 FDA 483 at Dr. Reddy's FTO-SEZ facility in Srikakulam found cleaning validation not performed after equipment replacement, OOS dissolution results invalidated despite contradictory evidence, and process qualification gaps — all traceable to a single uncontrolled equipment change eighteen months earlier.
Newsletter
Stay ahead in the Industry
Regulatory updates, pharma quality insights, and AI in manufacturing — written for quality leaders, not marketers.
Please use your official work email. Personal email addresses (Gmail, Yahoo, etc.) will not receive the newsletter. No spam. Unsubscribe anytime.
Ready to see what an AI-native quality platform looks like? Leucine unifies quality management, regulatory compliance, and production operations into one intelligent system.
