Pharmaceutical QA Skills
Reading Batch Records
Without Missing Critical Details
The skill pharmaceutical employers assess in every interview — and the one most training programmes never actually teach. Work through six practical stages and build the habit that separates thorough QA reviewers from good ones.
You are a QA reviewer at a pharmaceutical manufacturing site. Batch records land on your desk every day. Each one determines whether product reaches patients — or gets stopped.
Your job is not to understand how the product was made. Your job is to verify that every required element is present, complete, and consistent. Those are different tasks. This course trains the second one.
No preparation required. Start reading.
This course is built around doing before knowing — you will be asked to review records before you are given tools to review them. That is deliberate. The struggle is the learning. Work through each stage in order and do not skip ahead.
Interactive Course
Batch Record Review Training
Stage 1 of 6
Stage 1 — Cold read
No preparation. Just read.
A batch record excerpt has landed on your desk. Read it and flag anything that concerns you. There is no guidance yet — that comes next. Trust what you notice.
Batch excerpt — raw material dispensing
Material: Microcrystalline cellulose (MCC)
Required quantity: 12.500 kg
Actual dispensed: 12.500 kg
Lot number: MCC-2024-0312
CoA reviewed: Yes
Dispensed by: R. O'Brien Date: 14-Mar-2024
Verified by: S. Kaur Date: 13-Mar-2024
What concerns you? Write it down before revealing.
Critical — data integrity
The verifier signed the day before the activity took place.
S. Kaur verified on 13-Mar-2024. R. O'Brien dispensed on 14-Mar-2024. A verification recorded before the activity is impossible — it is either backdated or fabricated. This is a data integrity failure and would generate a 483 observation.
Most people miss this on first read. The row looks complete — name, date, signature all present. The issue is not what is there, it is the sequence. In Stage 2 you will get the framework that makes this visible every time.
Stage 2 — The framework
Why you missed it — and how to stop missing it.
Reading to understand asks what happened. Reading to verify asks whether everything is present, complete, and consistent. These are different tasks. Here is the structure that makes verification systematic.
What a batch record must contain
Product identity
Name, batch number, size, date, expiry.
Materials used
Lots, quantities, CoA reference, reconciliation.
Process steps
Equipment, parameters, actual readings, results.
Operator sign-offs
Performer and independent verifier — always different people.
Environmental records
Temp, humidity, pressure. Complete — no N/A for mandatory parameters.
Yield and reconciliation
Theoretical vs actual. Every discrepancy explained.
Deviations noted
Any departure from master record — even minor ones.
QA review and release
Authorised Person sign-off. Nothing ships without this.
The 8-point verification checklist
- 1Product identity matches the master batch record exactly.
- 2All process steps present and completed — no blanks, no skipped steps.
- 3Dates and times are chronologically consistent — nothing verified before it happened.
- 4No self-verification — performer and verifier are always different people.
- 5All in-process results within limits — OOS results have a deviation raised.
- 6Yield arithmetic correct and within range — every discrepancy explained.
- 7All deviations have a complete loop — root cause, impact, disposition, CAPA.
- 8Environmental records complete — no N/A for mandatory parameters.
Go back to Stage 1 now. The date discrepancy is caught by point 3 — dates and times are chronologically consistent. A verification cannot precede the activity it verifies.
In Stages 3A, 3B, and 3C you will practise using this checklist — starting with full support, then progressively less, until you are working from memory.
Stage 3A — Guided practice
One error. Full checklist beside you.
Use the checklist as you read. Work through each point in order. One issue is planted — identify which checklist point it fires and why.
Checklist — use this now
- 1Identity matches master record
- 2All steps present and complete
- 3Dates chronologically consistent
- 4No self-verification
- 5Results within limits
- 6Yield arithmetic correct
- 7Deviation loops complete
- 8Environmental records complete
Excerpt — granulation sign-off
Step 3.1 — dry mixing
Performed by: A. Flynn
Date/time: 08-May-2024 / 07:30
Verified by: A. Flynn
Date/time: 08-May-2024 / 07:32
Step 3.2 — wet granulation
Performed by: A. Flynn
Date/time: 08-May-2024 / 09:15
Verified by: P. Doyle
Date/time: 08-May-2024 / 09:18
Which point fires? What is the issue?
Critical — point 4Self-verification on Step 3.1.A. Flynn performed and verified the same step. Two minutes apart makes genuine independent review impossible. Step 3.1 cannot be accepted without re-verification by a different person.
Stage 3B — Guided practice
Two errors. Checklist fading.
The checklist is still visible but the point numbers are greyed out. You decide which points apply. Two issues are planted — find both.
Checklist — reference only
- 1Identity matches master record
- 2All steps present and complete
- 3Dates chronologically consistent
- 4No self-verification
- 5Results within limits
- 6Yield arithmetic correct
- 7Deviation loops complete
- 8Environmental records complete
Excerpt — environmental monitoring
Cleanroom: Grade B
Temp start: 21.2C — pass
Temp end: 21.8C — pass
Humidity start: 48% — pass
Humidity end: 47% — pass
Diff. pressure: N/A
Reviewed by: C. Nolan
Review date: 07-Jun-2024
Activity date: 08-Jun-2024
Two issues. Which points? What are they?
Major — point 8Differential pressure marked N/A in a Grade B cleanroom.Pressure differential is a mandatory GMP control in classified areas. It cannot be not applicable. Either it was not checked or not recorded.
Critical — point 3Review recorded the day before the activity.C. Nolan reviewed on 07-Jun-2024 but the activity took place on 08-Jun-2024. A review cannot precede the activity it reviews. Data integrity failure.
Stage 3C — Guided practice
Three errors. No checklist.
The checklist is gone. Work from memory. This is the bridge to Stage 4 — you are on your own, but still in a controlled environment. Take your time.
Batch excerpt — tablet compression, batch 2024-TC-0047
Product identification
Product name:Metformin HCl 500mg Tablets
Batch number:2024-TC-0047
Batch size (theoretical):200,000 tablets
Yield and reconciliation
Tablets produced (actual):197,340
Rejected during compression:210
Yield (%):98.77%
Acceptable yield range:98.0–100.5%
Operator sign-offs
Compression performed by:T. Walsh — 22-Apr-2024
Compression verified by:M. Brennan — 22-Apr-2024
QA review by:C. Nolan — 22-Apr-2024
Deviation summary
Deviation reference:DEV-2024-0312
Root cause:Punch wear — replaced at 12:45
Impact assessment:Tablets 10:00–12:30 quarantined pending review
CAPA raised:No
Batch disposition:Released
List all issues. Include severity and checklist point from memory.
Critical — point 7Batch released with unresolved quarantine.Tablets from 10:00–12:30 were quarantined but the batch is marked Released with no documented disposition. The deviation loop is incomplete. A batch cannot be released with an open quarantine.
Major — point 7No CAPA raised for a process failure.Punch wear causing an OOS weight result is a process failure. GMP requires a CAPA to prevent recurrence. Marking No without documented justification is insufficient.
Minor — point 6Yield arithmetic does not reconcile.197,340 + 210 = 197,550. Against 200,000 theoretical = 98.775%, which rounds to 98.78% — not 98.77% as recorded. Every number in a batch record must be accurate.
Stage 4 — Independent application
No scaffolding. Transfer the skill.
Full record. New product. No checklist. No hints. Work systematically, as if this batch is waiting for a release decision.
Batch excerpt — Atorvastatin 20mg granulation, batch 2024-GR-0091
Product identification
Product name:Atorvastatin 20mg Tablets
Batch number:2024-GR-0091
Batch size (theoretical):150,000 tablets
Manufacturing date:08-May-2024
Environmental monitoring
Cleanroom classification:Grade C
Temperature at start:22.1C — pass (18–25C)
Temperature at end:22.6C — pass (18–25C)
Humidity at start:51% — pass (30–65%)
Humidity at end:52% — pass (30–65%)
Differential pressure:N/A
Granulation process
Step 3.1 — dry mixing:Completed
Performed by:A. Flynn — 08-May-2024 — 07:30
Verified by:A. Flynn — 08-May-2024 — 07:32
Step 3.2 — wet granulation:Completed
Performed by:A. Flynn — 08-May-2024 — 09:15
Verified by:P. Doyle — 08-May-2024 — 09:18
In-process checks
Moisture content target:2.0–4.0%
Moisture check 1:2.8% — pass
Moisture check 2:3.1% — pass
Particle size D90:412 microns — pass (<500 microns)
Yield and reconciliation
Granules produced (actual):147,100 tablets equiv.
Yield (%):98.07%
Acceptable yield range:97.0–101.0%
QA review
Reviewed by:M. Costello — 08-May-2024
Batch disposition:Approved for compression
Your findings — work through the full record systematically:
Critical — point 4Self-verification on Step 3.1 dry mixing.A. Flynn both performed and verified Step 3.1. A two-minute gap makes genuine independent review impossible. This step cannot be accepted without re-verification by a different person.
Major — point 8Differential pressure marked N/A in a classified cleanroom.Grade C is a classified area. Differential pressure is a mandatory GMP control — it cannot be not applicable. The environmental monitoring record is incomplete.
Minor — point 6Reconciliation cannot be independently verified.No rejection figure is recorded. Without it the yield calculation cannot be checked. The record is incomplete even though the stated percentage falls within range.
Three questions to carry into every review
"Does this make sense?"Read each entry as a story. If a time is too fast, a result too perfect, a name appears twice — stop. Instinct is a signal, not a distraction.
"Is this complete?"Absence is as important as error. A missing entry is not nothing — it is an unverified step.
"What is the patient risk?"Every finding assessed against one question: could this have affected product quality, safety, or efficacy? That determines severity and urgency.