Guides · Process data
Annual product quality review: getting the batch data together
Both the US annual product review and the EU product quality review expect a yearly look at batches, and most of the process evidence sits in executed batch records. Collect it as each batch is released, keep every value traceable to its page, and rule out a copying error before investigating a trend.
The US annual product review and the EU product quality review ask the same basic question once a year: is the process still consistent, and does anything need to change? Answering it takes data from several systems, but the process evidence (critical in-process control results, yields, process parameters and the deviations raised during manufacture) is written in the executed batch records. On paper, that means someone copies values from many records into one dataset.
The work goes better when the data are collected batch by batch through the year, each value carries a reference to its page, and every surprising value is checked against the record before anyone investigates the process.
What do the regulations ask of the annual review?
In the US, 21 CFR 211.180(e) requires written records to be kept so that their data can be used for evaluating, “at least annually”, the quality standards of each drug product, to decide whether specifications or manufacturing or control procedures need to change. Written procedures must cover a review of “a representative number of batches, whether approved or rejected”, with their associated records where applicable, and a review of complaints, recalls, returned or salvaged products and investigations under 211.192.
In the EU, EudraLex Volume 4, Chapter 1, section 1.10 calls for “regular periodic or rolling quality reviews” of all authorised medicinal products, export-only products included. Their purpose is to verify the consistency of the process and the appropriateness of specifications, to highlight trends and to identify improvements. They should normally be conducted and documented annually, taking previous reviews into account, and section 1.10 lists twelve minimum contents. Section 1.11 asks the manufacturer, and the marketing authorisation holder where different, to evaluate the results and assess whether corrective and preventive action or revalidation is needed.
For active substances, ICH Q7 section 2.5 asks for regular quality reviews “with the objective of verifying the consistency of the process”, normally annually, with a shorter minimum list.
| Topic | US: 21 CFR 211.180(e) | EU: Chapter 1, 1.10 and 1.11 | APIs: ICH Q7 2.50 and 2.51 |
|---|---|---|---|
| Frequency | At least annually | Periodic or rolling, normally annually | Normally annually |
| Batches | A representative number, approved or rejected | All batches that failed specification, and their investigation | All batches that failed specification |
| Process data | Records associated with the batch, where applicable | Critical in-process controls and finished product results | Critical in-process control and critical API test results |
| Deviations | Investigations under 211.192 | Significant deviations, their investigations and the effectiveness of CAPA | Critical deviations and related investigations |
| Complaints, returns, recalls | Yes | Yes | Yes |
| Stability | Not named in 211.180(e) | Stability monitoring results and adverse trends | Stability monitoring results |
| Changes | Not named in 211.180(e) | Process and analytical method changes; MA variations | Process and analytical method changes |
| Outcome | Need for changes to specifications or procedures | Assess CAPA or revalidation | Assess corrective action or revalidation |
The EU list also covers starting materials and supply chain traceability, post-marketing commitments, the qualification status of equipment and utilities, and contractual arrangements. Many companies write one procedure that satisfies both the US and EU texts. How yours is scoped is a decision for your quality system.
Which data come from the executed batch records, and which from elsewhere?
The batch record is where the manufacturing evidence is written. 21 CFR 211.188 requires it to include in-process and laboratory control results, actual yield and percentage of theoretical yield, the equipment and component batches used, and any investigation under 211.192. EU Chapter 4, section 4.20 lists the same kinds of content for the batch processing record. ICH Q7 section 6.52 adds “actual results recorded for critical process parameters”.
| Data for the review | Usually found in | Regulatory hook for the record |
|---|---|---|
| Critical in-process control results (weights, hardness, pH, fill volume) | Executed batch record; some lab IPCs in LIMS | 211.188(b)(5); EU 4.20(f) |
| Yields at stages and at the end, against theoretical | Executed batch record | 211.188(b)(7); limits from the master record, 211.186(b)(7); EU 4.20(g) |
| Process parameters (temperatures, times, speeds, pressures) | Executed batch record, attached printouts, a historian on automated lines | ICH Q7 6.52; EU 4.18(d) for the instructions |
| Deviations raised during manufacture | Referenced in the record, detailed in the deviation system | 211.188(b)(12) (investigations under 211.192); EU 4.20(h) |
| Starting material batches and equipment used | Executed batch record | 211.188(b)(2) and (3); EU 4.20(d) and (e) |
| Finished product test results | LIMS, certificates of analysis | Laboratory records |
| Stability results | Stability programme, LIMS | Stability records |
| Complaints, returns, recalls | Complaint and quality management systems | Complaint and recall records |
| Changes, variations, qualification status, contracts | Change control, regulatory affairs, engineering, quality agreements | Their own records |
The lower rows usually live in electronic systems that export. On a paper process the upper rows do not, and that is where most of the effort and most of the risk to the data sit.
How do teams assemble the data from paper records?
Practice varies, but the teams that find the review easiest tend to follow the same few steps.
- Fix the data list before the period starts. For each value: its name, process step, unit, number of decimals, printed limits and where it appears in the master record, by version. See NOR, PAR and unlabelled ranges for recording which limit is which.
- Collect each batch when it is released. By then the record is complete and has been through the review required by 21 CFR 211.192, and it is still at hand.
- Copy values as written. Keep the unit and the decimals, do not round, do not recalculate. Where an entry was corrected on the page, take the corrected value and note that it was corrected. Keep “not done” and “N/A” apart from a blank.
- Include rejected batches. 211.180(e)(1) says approved or rejected. Mark each batch’s disposition in the dataset.
- Record where each value came from: batch number, record and master record version, page, step.
- Verify the transcription. EU Annex 11, section 6 expects an additional check on the accuracy of critical data entered manually into a computerised system, by a second operator or by validated electronic means. Whether every value or a sample is checked is your procedure’s choice. See verifying transcribed data.
- Watch for master record changes. A new version can move a step, change a limit or add a parameter. Keep the version beside each batch so that trends are not drawn across a change without saying so.
How should the data be trended?
Neither 211.180(e) nor Chapter 1 prescribes a statistical method. FDA’s process validation guidance, which ties its stage 3 programme to 211.180(e), recommends that data be “statistically trended and reviewed by trained personnel”, and that a statistician or a person trained in statistical process control develop the methods. It also asks procedures to guard against overreacting to single events as well as against missing real variability, and to look at variation within batches as well as between them. EMA’s GMP questions and answers note that the trending in a product quality review can include results from the previous period.
A few basics hold in most procedures:
- Plot each value in manufacturing order, one point per batch.
- Specification, NOR and PAR limits say whether a batch met its requirements. Trend limits describe the process itself and usually come from its own data, for example control chart limits from a stable baseline period. Your procedure defines them.
- With few batches a year, carry earlier periods forward rather than drawing conclusions from a handful of points.
- Mark changes on the chart (master record versions, equipment, sites, raw material suppliers) and compare before and after rather than pooling.
- Use capability indices only when the process is stable and there are enough data to support them.
From paper batch records to CPV data covers setting up the same dataset for continued process verification.
What should you do with an out-of-trend value?
Define out-of-trend in your procedure before you need it. FDA’s OOS guidance is written for test results, but notes that much of it may be useful for examining results that are out of trend. The laboratory supervisor’s first assessment includes examining the raw data and verifying the calculations, before the investigation looks wider.
For a dataset copied from batch records, the equivalent first step is to go back to the page.
- Compare the dataset with the ink. Look for transposed digits, a moved decimal point, a wrong unit, a value from the wrong row, step or batch, or an original entry taken instead of its correction. In a fictional example, a stage yield written as 98.6 is typed as 89.6, and the chart shows a sudden drop that no process explains.
- If the copy is wrong, correct the dataset and document it: what changed, why, who and when, with a reference to the source page. Do not change a value without the source in front of you.
- If the record confirms the value, treat it as a process signal. Check whether it was outside its limits and whether that was raised as a discrepancy at batch record review, as 211.192 requires for unexplained discrepancies. If it was within limits, follow your out-of-trend procedure, and look at neighbouring batches and other parameters of the same step.
Ruling out a copying error first matters in both directions. A copying error investigated as a process trend wastes the investigation and can leave a wrong conclusion in the review. A real trend dismissed as “probably a typo” without checking the page is worse. Note too that in the EU, Annex 16, point 1.7.16 expects out-of-trend investigations that pertain to a batch to be completed to a sufficient level before the Qualified Person certifies it. A trend that surfaces only at year end has missed that point.
Rolling review or year-end campaign?
Section 1.10 names both “periodic or rolling” reviews. EMA’s questions and answers say the review is expected annually, that review timeframes can be adjusted to manufacturing and campaign duration with justification, and that the timeframe criteria belong in an SOP. Even when nothing was manufactured in the period, the quality and regulatory review still takes place. The US text sets a floor of at least annually.
| Aspect | Rolling collection | Year-end campaign |
|---|---|---|
| When data are captured | At each batch release | In the weeks before the report is due |
| Records | At hand, recently reviewed | Retrieved from the archive |
| Trends seen | During the year, while batches can still be investigated | After the period has ended |
| Workload | Spread out | Concentrated, often on a few people |
Many sites mix the two: collect at release, look at the charts quarterly and write the review once a year. Chapter 1.11 also allows reviews to be grouped by product type, such as solid dosage forms, where scientifically justified. Your quality system decides which approach is justified for each product.
How do you keep the dataset traceable to the records?
The review’s conclusions are only as good as the link between each number and the page it came from. A checklist:
- Each value carries its batch number, record identity and master record version, page and step, the value as written and its unit.
- Each value shows who transcribed it, who verified it and when.
- The source page is retained as the original or a true copy, which 211.180(d) permits.
- Every change to the dataset after verification is recorded with its reason. See ALCOA+ on paper for the principles behind this.
- The dataset version that supported the review’s conclusions is fixed and kept with the review.
- If the dataset lives in a spreadsheet or another computerised system, Annex 11 applies: it covers all computerised systems used as part of GMP regulated activities. How far you validate a spreadsheet is a decision for your quality system.
- The paper side needs the same care: EMA’s questions and answers note that data integrity applies equally to paper and electronic systems.
Sources
- 21 CFR 211.180 General requirements (records and reports), accessed 18 November 2026.
- 21 CFR 211.186 Master production and control records, accessed 18 November 2026.
- 21 CFR 211.188 Batch production and control records, accessed 18 November 2026.
- 21 CFR 211.192 Production record review, accessed 18 November 2026.
- EudraLex Volume 4, Part I, Chapter 1: Pharmaceutical Quality System (2013), accessed 18 November 2026.
- EudraLex Volume 4, Part I, Chapter 4: Documentation (2011), accessed 18 November 2026.
- EudraLex Volume 4, Annex 11: Computerised Systems (2011), accessed 18 November 2026.
- EudraLex Volume 4, Annex 16: Certification by a Qualified Person and Batch Release (2015), accessed 18 November 2026.
- ICH Q7 Good Manufacturing Practice Guide for Active Pharmaceutical Ingredients, accessed 18 November 2026.
- EMA, Guidance on good manufacturing practice and good distribution practice: questions and answers, accessed 18 November 2026.
- FDA, Process Validation: General Principles and Practices (January 2011), accessed 18 November 2026.
- FDA, Investigating Out-of-Specification (OOS) Test Results for Pharmaceutical Production, Revision 1 (May 2022), accessed 18 November 2026.
General information, not regulatory advice. Check the current text of each regulation; your quality system decides how it applies to you.
Related guides
- From paper batch records to CPV data, without breaking data integrity
How to move process parameters from paper executed batch records into a CPV, APR/PQR or investigation dataset that holds up to data integrity review.
- NOR, PAR and unlabelled ranges on batch records
How a normal operating range differs from a proven acceptable range, how each is printed on a batch record, and what to do with a range that has no label.
- Verifying transcribed GMP data: second-person checks that hold up
What regulators expect when GMP values are copied from paper into a spreadsheet, LIMS or CPV database, and how to design a check that stands up.