Regulations and compliance · Practical guide

GMP manual data transfers: how do you check entries across different media?

A correctly transcribed number assigned to the wrong sample can still lead to a wrong decision. For a manual transfer between an instrument, paper form and electronic system, keep the source, calculation, destination and later corrections connected. Targeted error cases test that chain for a human medicines GMP process.

Table of contents

At a glance

What should a manual GMP data transfer demonstrably preserve?

Preserve content and its unambiguous context, including sample, unit, method, time and applicable calculation. Document each change of medium, the person responsible for the handover and the check at the destination. Critical manual entries require an additional accuracy check. Test whether subsequent corrections reach affected downstream records without obscuring the original information.

  • The chain of custody begins at the original record and ends at the decision that is acted upon.
  • The same numbers do not indicate the correct sample, the correct unit, or the correct calculation.
  • A correction is only considered finalized when its impact on subsequent records has been assessed.

1. Map the actual chain of handovers

In human medicines GMP, Chapter 4 addresses electronic, paper and hybrid records. Annex 11 section 6 addresses an additional accuracy check for critical data entered manually. These provisions neither prohibit manual transfers nor require two signatures for every value.

Start with a specific result that supports a business decision. Work backwards: Which LIMS line contains it, which worksheet originated it, and which report or electronic raw dataset is associated? Also include intermediate workspaces and subsequent report copies. A simple list of deployed systems does not reveal these gaps.

This article addresses the cascading effect of multiple transfers. The design of a single input form and its accuracy control is further explored in the linked article on manual data entry.

Sources: [2] [1]

Further reading:

2. Assign each transfer a verifiable profile

A transfer profile describes source, destination, identifiers, processing, and control point. Separate the exact copy from a conversion or derived result. If conversion from milligram to gram occurs, a transformation has already taken place; a simple character-by-character comparison would constitute an inappropriate control.

Define which source holds precedence in case of discrepancies and specify who is authorized to make that decision. A scanned page is not inherently the complete original record merely because it is legible. Dynamic data, context, or additional pages may be missing. The matrix below is a standalone work aid for a multimodal laboratory process.

Working aid: handover profile from the device to the use of results
TransferReceived contextProcessingControl at the target
Device → Work sheet Probe P-17, Measurement M-08, Unit mg, complete printout Copy of the measurement value Assign to P-17/M-08 and verify value together with unit
Worksheet → LIMS P-17, Sheet WB-03 Revision 2, Calculation Rule R-1 Conversion of mg to g Independent calculation and target field format verification
LIMS → Result report P-17, valid result version, evaluation status Selection of the approved result Compare version and status, not just numerical values
Correction → Follow-up records Original and corrected value, reason, affected objectives Controlled Correction Determine follow-up usage and document closure

Sources: [2] [1]

3. A worked example with three errors

The following example does not describe a product test. The sample record M-08 for sample P-17 contains 125.0 mg. The controlled work sheet WB-03, Revision 2, requires the LIMS field to be entered in grams. Prior to testing, the required value is established: 125.0 ÷ 1,000 = 0.1250 g. An results specification or release limit is explicitly not assumed here.

The tester uses a predefined copy of the source record and documents the actual response. The "expected" distinguishes between computational accuracy, correct assignment, and valid document version. A system may correctly store all numerical values yet still permit one of these technical errors.

Fail cases are intentionally executed one at a time. After each case, the initial state is restored or a clearly distinct new test dataset is used. Otherwise, a previous correction may obscure the observation of the next case.

Example: Suggested matrix for P-17 / M-08
CaseInput or transferExpected result set in advance
Normal case 125.0 mg from M-08; 0.1250 g in P-17 Source, conversion, and target sample are consistent
Unit errors 125.0 is recorded as g in P-17. Control detects factor 1,000; no unassessed onward transfer
Assignment error 0.1250 g is assigned to P-71 Identity comparison detects another sample despite correct number entry
version error Correct value entered on a worksheet that is no longer valid Assess the validity of the computational rule and potential downstream impacts; do not accept merely due to identical numbers

Sources: [1] [2]

4. Perform control prior to intended use

A control that occurs only after a decision has been made cannot prevent prior incorrect usage. Therefore, assign the control point to the actual point of use: are the data being used for an interim decision, or only in the final report? A status such as "unverified" is only meaningful if the relevant work steps also account for this status.

During an independent review, the reviewer should have access to the relevant source. Two individuals verifying the same previously incorrectly transcribed intermediate value may only be assessing the consistency between two copies. For each critical transfer, an effective control method must be justified; a validated electronic control may be appropriate depending on the task, as opposed to a further manual copy.

Also assess missing pages, illegible units, or ambiguous measurement assignments. The appropriate expected value should then be a clearly defined exception, not a post-hoc plausibly adjusted numerical value.

Sources: [1] [2]

5. Trace a correction through all affected objects

If the initial value is corrected, a causality chain must be established. Which calculations, results, and reports have contributed to this correction? Has a decision already been made? Document the affected versions and the assessment of their continued usability. The original result must not be silently replaced by a new value.

Distinguish a correction to a record from a change to the procedure. A transfer error may trigger a controlled correction and a deviation evaluation. A incorrect formula in a locked template additionally requires assessment of the template and its scope of use. Not every typing correction constitutes an automatic system change.

For a practical test, modify the key value to 120.0 mg in a copy of the example. The new calculated required amount is 0.1200 g. Verify that the original 0.1250 g remains detectable and that the affected report is properly identified.

Sources: [2] [1]

Further reading:

6. Verify reconstruction without knowledge of the operator

Let a qualified colleague independently trace an example dataset from the intended records, identifying the source, editing rule, control measure, and version of results used. This independent methodological check demonstrates whether the chain of custody relies solely on personal memory.

A folder containing all files may still be unusable: file names may be ambiguous, print pages may be missing, or a formula version may not be linked. Ensure necessary connections are maintained so that they remain accessible throughout the intended retention period. A new document folder is not necessarily required; often, a clear reference within the existing dataset is sufficient.

Re-evaluate the media change whenever device prints, units, form versioning, or LIMS fields are modified. A previously valid control may lose its basis due to a minor change in presentation.

Sources: [2] [1]

7. Preparation for QA, Laboratory, and IT

The laboratory manager understands the technical origin of the value, IT is aware of the storage and transfer pathways, and QA is familiar with the relevant usage and documentation requirements. Clarify these contributions prior to testing. Effort increases significantly, particularly in cases of unknown intermediate storage areas, variations, and unclear correction procedures—not solely due to the number of input fields.

  • Select a representative data pathway with original source and all intermediate media.
  • Determine required identifiers, units, and calculation rules in advance.
  • Retain valid form and system versions as well as the initial state.
  • Identify, assign responsibility for, and define actions to be taken in response to uncertainties.
  • Normal case, identity errors, transformation errors, and subsequent corrective actions to be prepared.
  • Review the use of results and ensure retrievable storage of associated records.

Sources: [2] [1]

8. Avoid counterfeit security by not creating additional copies

Additional copies and signatures do not automatically improve the chain. For each additional copy, inquire about the specific error it is intended to detect and the new risk of confusion it introduces. A clear transfer reference may be more useful than an additional full reprint.

Equally, "staff are trained" alone is insufficient as evidence. Let the planned controls operate against specific error cases. Document the observed outcomes and open limitations. If a transfer pathway is regularly only manageable through extensive root cause investigations, evaluate a targeted technical or organizational improvement.

Sources: [1] [2]

Project experience and collaboration

Test the manual transfer against common errors

With the process team, we trace a value from its original record to subsequent use. We check transposed digits, units, decimal separators, mapping and completeness. A targeted second-person check may suit one process. A suitable validated electronic control may be preferable in another. We try the selected control against prepared failure cases and agree how deviations will be handled. A second signature field alone does not demonstrate that the control works.

Frequently asked questions about Manual data transfer

Is manual retyping inherently unacceptable under GMP?

No. The key factors are the specific data pathway, criticality, and the effectiveness of controls. For critical manual entries, Annex 11 recommends an additional accuracy verification. A manual pathway may be acceptable if its source, assignment, and processing remain controlled and traceable.

Sources: [1] [2]

How does this validation differ from a single input control?

The input validation checks, for example, the value in the LIMS field. The handover chain additionally links device, work sheet, calculation, result version, and subsequent usage. This enables visibility into errors arising between properly functioning individual checks, such as a misassigned sample on the intermediate form.

Sources: [2] [1]

Is the comparison of numerical values sufficient?

Only if this actually covers the entire relevant information. In the example of this article, sample, measurement, unit, and form revision are also crucial. Identical numbers with different units or incorrect sample assignment do not constitute a correct transfer.

Sources: [1] [2]

Must every transfer be signed off by a second person?

No. For critical manual entries, Annex 11 Section 6 requires an additional accuracy check; this may be performed by a second operator or through validated electronic means. Therefore, a signature from a second party is not the sole solution for every transfer. The selected control must be capable of reliably detecting actual entry errors.

Sources: [1]

What do we do when the original copy is missing?

Clarify which relevant record actually exists and whether it is fully accessible. An identical value from a subsequent copy does not automatically close a traceability gap. Assess the impact on previously used results and document the subsequent decision.

Sources: [2]

Is the conversion of mg to g considered a significant additional inspection?

When unit conversion is relevant, the correct target value should be clearly defined. In the example provided in this article, 125.0 mg is divided by 1,000, yielding 0.1250 g. Verify the unit, decimal representation, and rounding rule; a second copy of the same incorrect formula does not constitute an independent calculation check.

Sources: [1] [2]

May we correct erroneous subsequent records?

Yes, controlled corrections are possible. The original information and the context of the change must remain traceable in accordance with the current documentation process. Identify the affected downstream records and any previously made decisions, rather than merely correcting the last target field.

Sources: [2] [1]

Which documents should a small and medium-sized enterprise provide first?

A complete real-world workflow, key original records, valid forms, calculation rules, and result usage guidelines serve as a solid starting point. Additionally, responsible personnel from laboratory, IT, and QA functions should be involved. Missing data pathways should first be made visible before a large number of additional test cases are generated.

Sources: [2] [1]

Sources and review date

Source date: 7 October 2026. Human medicinal products – GMP. The tables and figures are internal work aids and do not represent completed product testing. Requirements from other GxP areas are not to be equated on a general basis.

  1. European Commission: EU GMP Annex 11 – Computerised SystemsRevision January 2011, applicable since 30.06.2011; §§ 1, 4, 6–7, 9–13, 16–17. Human Medicines GMP Guideline; no draft version.
  2. European Commission: EU GMP Chapter 4 – DocumentationRevision January 2011; Principle, § 4.1 and §§ 4.7–4.12: Documentation relationships, timely documentation, corrections, and retention.

Sources checked: .

Daniel Herrmann, Daniel Herrmann Consulting

Daniel Herrmann Consulting

Check the transfer chain to concrete records

We support you in identifying critical media changes, designing appropriate controls, and preparing meaningful deviation cases. The starting point is your actual data flows and the decisions based on them.

  • Computer system validation →

    Bring your GxP systems into validated operation, from laboratory software, MES and eQMS to ERP. We define the scope, plan risk-based tests using GAMP 5 and hand the evidence to QA for release.

Send a project enquiry Customer testimonials →