QC record review: what 21 CFR 211.194 asks for, and where lab records fail the second read
Every QC result is signed twice. The analyst signs that the test was run. A second person signs that the record is accurate, complete and compliant. That second signature is a regulatory statement under 21 CFR 211.194(a)(8). In most labs it is made in a few minutes, against a worksheet the reviewer has seen a hundred times. That is where records that look finished get through.
What the record has to contain
| 211.194(a) | Required in the record | What the reviewer checks it against |
|---|---|---|
| (1) | The sample: source, quantity, lot number, date taken, date received | The sampling record and the batch record |
| (2) | The method used for each test | The current approved method and its version |
| (3) | The weight or measure of sample used | The amount the method specifies |
| (4) | All data from each test, including graphs, charts and spectra | The instrument's own record of what was run |
| (5) | All calculations, with units and conversion factors | The formula in the method |
| (6) | The results and how they compare with the specification | The current approved specification |
| (7) | Initials or signature of the person who ran each test, and the date | Training and the instrument log |
| (8) | Initials or signature of a second person who reviewed the original records | Accuracy, completeness and compliance with established standards |
SOURCE: 21 CFR 211.194(a). PARAPHRASED. SEE SOURCES BELOW.
Paragraphs (b) through (e) add records of method modifications, the testing of reference standards and reagents, instrument calibration and stability testing. For GLP studies, 21 CFR 58.130(e) sets the parallel rule: data recorded directly, promptly and legibly, dated and signed, with any change leaving the original entry readable and giving the reason.
What the second signature means
The reviewer is not confirming that the worksheet was filled in. The reviewer is confirming three separate things: the record is accurate, it is complete, and it complies with established standards. Each one needs a different comparison.
- Accurate means the worksheet agrees with the raw data. FDA's data integrity guidance recommends that audit trails capturing changes to critical data be reviewed with each record, before final approval.
- Complete means every run is accounted for, including the ones that were not reported.
- Compliant means the test was done the way the approved method says, on the current version, within its system suitability limits.
Where completed records stop matching
| The worksheet says | The approved document says | What it becomes |
|---|---|---|
| Sample weight recorded | Method sets a range the weight falls outside | A deviation from the test procedure that 211.160(a) requires to be recorded and justified |
| Template is version 3 | Method was revised and version 4 is effective | A test run to a superseded procedure |
| Result calculated and passing | Method's formula uses a different dilution factor | A reported result that the raw data does not support |
| System suitability marked "pass" | Method lists five criteria and the record shows three | An unverified run |
| One injection sequence reported | Instrument record shows an earlier sequence the same day | Unreported data, and a question under 211.192 |
| Reference standard lot recorded | Standard's record shows it had expired | Invalid results for every sample in the run |
ILLUSTRATIVE EXAMPLES. NOT DRAWN FROM ANY ONE COMPANY.
None of these is visible from the worksheet alone. Each one appears only when the completed record is read next to the template, the method and the specification it was supposed to follow.
Why a careful reviewer still misses them
- The template is trusted. Reviewers check entries against the form. Few check the form against the method it was built from.
- Volume. A release lab can produce hundreds of records a week. Review time per record is measured in minutes.
- Familiarity. The same analyst, the same assay, the same numbers. A transposed factor looks like every other day.
- The documents live apart. The method is in the document system, the data is in the instrument software, the worksheet is in LIMS or on paper.
A review that holds up
- Verify the template against the current method each time the method is revised, and record that you did.
- Review to the method, not to the form. Every limit and every factor on the worksheet should trace to a numbered step.
- Reconcile runs. The number of sequences in the instrument record should equal the number in the worksheet, or the difference should be explained.
- Treat an out-of-specification result under the OOS procedure from the first minute. FDA's OOS guidance expects a laboratory investigation before any retest.
- Record what the reviewer checked, not only that a review happened.
Where Clinplex fits
Clinplex is the AI regulatory intelligence layer that augments your QMS and LIMS. It reads completed lab worksheets and reports against your approved templates, methods and specifications, read-only, before they reach QA. Each finding cites the document and the requirement it was read against. A qualified reviewer makes the call and signs. How it works is on the platform page.
Sources
- 21 CFR Part 211. Sections 211.160, 211.192 and 211.194. ecfr.gov
- 21 CFR Part 58. Section 58.130(e), conduct of a nonclinical laboratory study. ecfr.gov
- FDA, Data Integrity and Compliance With Drug CGMP: Questions and Answers. Guidance for industry, December 2018. fda.gov
- FDA, Investigating Out-of-Specification (OOS) Test Results for Pharmaceutical Production. Guidance for industry, revised May 2022. fda.gov
Frequently asked questions
What does 21 CFR 211.194 require in a laboratory record?
Complete data from every test: the sample, the method, the amount used, all raw data including charts and spectra, all calculations, the results against specification, the analyst's signature and date, and a second person's signature showing the original records were reviewed.
Who can perform the second-person review?
The regulation requires a second person and does not name a job title. The reviewer has to be qualified to judge accuracy, completeness and compliance for that test, and independent of the person who ran it.
Does the reviewer have to look at audit trails?
FDA's 2018 data integrity guidance recommends that audit trails capturing changes to critical data be reviewed with each record and before final approval of the record.
Can software replace the second-person review?
No. Software can read the record against the method and the specification and point to what does not match. The review and the signature under 211.194(a)(8) belong to a qualified person.
Read the record against the method it came from.
Send three records. We return the gaps with the clause attached in 48 business hours, under a mutual NDA.
Run an Inspection Stress Test