Regulations and compliance · Practical guide

GMP metadata: which information belongs with the result?

A laboratory result may look plausible. Months later, the dilution, method or sample assignment may be impossible to establish. A worked laboratory calculation shows how to identify required metadata, keep them linked to the record and recover them after export.

Table of contents

At a glance

Which metadata need to be preserved under GMP?

Retain the metadata needed to understand, attribute and reconstruct the GMP activity. Depending on the data process, these can include unit, sample, method, calculation parameters, operator, timing and change history. A results PDF is insufficient if it omits decision-relevant context. Conversely, every technical system log does not automatically belong to every result.

  • Start with the decision the process supports and the data context needed to make it.
  • Also get the assignment between result, parameters and valid method level.
  • Verify subsequent access using an actual dataset and a specific query.

Metadata provides professional context, not a general blanket requirement

This article examines quality decisions in human medicinal products. As of October 7, 2026, Chapter 4 and Annex 11 in the EU GMP directory remain listed in the 2011 versions. PIC/S PI 041-1 provides guidance on appropriate data controls for GMP/GDP. The FDA Data Integrity Guidance complements the view for applicable US Drug CGMP; it is therefore not to be considered a general European regulatory requirement.

The key distinction is: a technical field may exist without being relevant to the decision. Conversely, an inconspicuous field may be essential. The dilution factor alters the concentration; a screen color scheme typically does not. PIC/S explicitly distinguishes between metadata that is relevant to reconstruction and metadata that is not critical. Therefore, selection must be based on a clear, professional justification, not on a maximum number of stored log files.

Sources: [1] [2] [3]

From the result to the specific context matrix

Take a report that actually supports a decision. Do not first ask what can technically be exported; instead, consider what an independent auditor must be able to explain. The following internal matrix illustrates the relationship for a concentration determination. It is neither a prescribed form nor a universal minimum data set for every laboratory.

Assign a source and a permanent reference to each field. The designation "in the LIMS" is too general. Specific examples include sample identifier, method revision, and the exact parameter set. When multiple systems are involved, clearly identify which system provides the decisive information and define the procedure for resolving discrepancies.

Own work aid: Results and required context
Assessment questionInformation neededEvidence of correct assignment
What was assessed? Probe, material, batch, and sub-sample Unique identifier in the order and measurement record
What does the value mean? Unit, reference value, and dilution Stored calculation status
Which instruction applies? Method, Revision and Acceptance Criteria Historically available approved version
How was the result obtained? Input values, parameters, and calculations Linked Measurement and Result Version
Who decided? Process owners, reviewers, timestamp and status Personalized records with result reference

Sources: [4] [2]

An identifier must remain unambiguous after changes

A filename is helpful but often insufficient as a sole reference. Files can be renamed, folders moved, and sample identifiers can be reissued at different locations. Therefore, define identity within the relevant context: for example, laboratory, order, sample, measurement, and analysis version. Verify that your search function opens the correct historical dataset and not merely the most recent version.

During a master data change, the current designation must not inadvertently replace the previous professional context without notice. A modified method number or unit can cause an older representation to appear differently. What is essential is not that all information is present in a single file. What is essential is that the required context remains protected, understandable, and retrievable throughout the relevant retention period.

Sources: [5] [6] [2]

Example: The same raw value leads to a different result when processed with a different factor.

In the following example, three measured values are associated with sample M-72: 0.96, 1.00, and 1.04 mg/L. The pre-established calculation computes the arithmetic mean and multiplies it by the documented dilution factor of 25. The mean value is 1.00 mg/L, resulting in 25.00 mg/L for the original sample. These figures constitute an independent calculation aid.

An export contains three values and the final result, but no factor. Later, the currently standard factor of 20 is inadvertently applied. This results in a value of 20.00 mg/L—a difference of 5.00 mg/L or 20% compared to the original result. Therefore, a suitable numerical value in the report does not necessarily confirm that a subsequent reconstruction is feasible.

Example: Reconstruction with and without parameter reference
Test stepIndependently established expected resultSignificance of the finding
Sum the original values 0.96 + 1.00 + 1.04 = 3.00 mg/L All three input values relate to the measurement.
Calculate the mean value 3.00 / 3 = 1.00 mg/L Calculation rules and quantities are traceable.
Apply historical factor 1.00 × 25 = 25.00 mg/L Stored factor belongs to the result set
Identify a false factor 1.00 × 20 = 20.00 mg/L Missing context alters the statement
Determine deviation (25.00 − 20.00) / 25.00 = 20 % No new product specification derived from this example number

Sources: [4] [3]

Scope of export and separation of complete record

A specific export may serve a limited purpose. An output list for production planning does not need to possess the same level of detail as the complete, retained laboratory records. Therefore, document the intended purpose of the export, what data remains in the source system, and how an authorized reviewer can access supplementary information. Concern arises when the export is later used as the sole and complete evidentiary basis.

Intentionally test a dataset with modifications and multiple evaluation versions. A simple, unaltered dataset is insufficient to reveal typical context loss. After transfer, verify not only the number of rows but also unit, field meaning, method reference, and link integrity. For dynamic data, additionally assess which interaction capabilities are required for subsequent subject-matter review.

Sources: [2] [6] [3]

Do not provide missing metadata retrospectively as a historical fact.

If information is missing, distinguish between secured evidence, justified reconstruction, and open uncertainty. A timely documented approach or an unaltered device export may provide a credible foundation. A participant's recollection may assist in an investigation, but it does not automatically substitute for the missing original parameter value. Any addition must remain clearly identifiable as a subsequent addition with its source.

Assess which decisions were based on the results. If only a non-relevant comfort parameter is missing, the consequence differs from that of unknown dilution or sample identity. A complete new test is not always the appropriate response: it may fail to reproduce the condition of a previous sample. The responsible quality function requires the actual evidential basis and its limitations.

Sources: [5] [6] [7]

Practical testing of review access and searchability

A document owner may have knowledge of the storage location, while the reviewer lacks authorization for method formulation. Therefore, verify access rights using the role that will be employed subsequently. Administrative full access during demonstration may obscure such gaps. The reviewer should be able to transition from the result to the required input data, relevant changes, and the associated assessment.

A relevant acceptance question is: Can a competent colleague explain why exactly this value appeared in the report? Have them identify a known parameter and a previous version. Document the search path, version identified, and remaining gaps. The audit depth follows risk assessment and data volume; it is not determined by a universal click count or fixed sample size.

Sources: [6] [7] [2]

Checklist for the next data set

Begin with a critical result process and transfer the approach only after clarification to related procedures. QA and laboratory leadership determine the technically necessary reconstruction. IT investigates data storage, identities, and access. The system owner documents the decisions and open items relevant to the considered system.

Closure is achieved when the key contextual questions have been addressed and the remaining limitations are reasonably evaluated and understood. A completed matrix alone is insufficient. For example, use a change in the export route or a method change as an opportunity to re-evaluate the previous rationale.

  • Clear separation of result, sample, and decision.
  • Unit, calculation parameters, and historical method assigned.
  • Relevant changes and evaluation statuses identifiable.
  • Purpose of export and fully retained record distinguished.
  • Access with the actual reviewer role verified.
  • Missing information together with decision rationale is assessed.

Sources: [4] [6]

Project experience and collaboration

Preserve the context needed to interpret the measurement

With IT and process owners, we identify the contextual information needed to understand a later decision. We trace units, method, time reference, assignments and relevant changes through display, export and archive. A static record may need less context than a dynamic analysis. We identify what the use actually requires and test that it survives transfer or restoration. An inventory of technical fields cannot replace this assessment.

Frequently asked questions about GMP metadata

Are raw data and metadata the same?

No. Raw data represents the first recording of information, whether on paper or electronically. For electronic records, all data upon which quality decisions are based must be designated as raw data (EU GMP Chapter 4, definition of "Records"). Metadata provides the necessary context for understanding, such as unit, timestamp, user or device identification, and change history (FDA Question 1(b); MHRA 6.3). Metadata forms part of the complete original record. Therefore, a measurement signal lacking this context cannot serve as adequate evidence.

Sources: [2] [3]

Must we retain all technical logs on a long-term basis?

No. PIC/S distinguishes between relevant metadata and non-critical information, where complete capture may be exempted upon justified risk assessment (9.1.5.2). A system error log may be relevant if it explains a measurement interruption. Decide based on content, process, and future traceability requirements. The filename or storage volume alone does not suffice to address retention questions.

Sources: [2]

Is the current method version sufficient as a reference?

For a historical dataset, the actually used version must remain clearly identifiable. A reference to a current version is insufficient if calculations or decision rules have been amended. Verify that the historical version is accessible and uniquely attributable. A clear, protected version reference may suffice; an additional copy in each report is not automatically required.

Sources: [4] [5]

Must every export include all metadata?

Not every export is intended to serve as a complete record. A limited data export may be acceptable if the purpose and boundaries are clearly defined and the necessary complete context remains available. However, if the export is intended to replace the source system as a retention record, its suitability for this broader purpose must be separately demonstrated.

Sources: [2] [3]

Are Audit Trails also considered metadata?

You may contain decision-relevant metadata, particularly relating to the creation and modification of a record. However, an audit trail does not replace the raw data or the technical justification for a decision. Identify the specific events that explain the particular dataset, and link these to the associated review and the original information.

Sources: [3] [6]

Is a screenshot sufficient as proof of the dilution factor?

A screenshot can complement a test batch. As a long-term parameter reference, it is only suitable when origin, timestamp, and assignment are reliably identifiable. Additionally, verify that it contains all required information and is properly controlled and retained. Native, version-specific logging is generally easier to systematically link to the results.

Sources: [5] [2]

Who determines which metadata are relevant?

This requires combined subject-matter and system knowledge. The technical department explains the decision, QA defines the applicable documentation requirements, and IT outlines the actual data pathways. Document the rationale throughout the data process. A sole selection by the export developer risks overlooking technical details that may appear inconspicuous but are critical for future evaluation.

Sources: [6] [2]

What do we do when a historical parameter is missing?

First, preserve existing original information and verify independent, contemporaneous evidence. Document what can be substantiated and what is merely assumed. Any subsequent reconstruction must remain clearly identifiable as such. Evaluate the affected results and decisions; do not record a suspected value as the original parameter without comment.

Sources: [5] [7]

Must metadata and raw data be stored in the same file?

No, a specific file organization is not generally mandated. However, the relationship must be verifiable and practically reconstructible. In particular, unique identifiers, version references, and concurrent access during the required duration should be assessed. Dispersed files lacking stable associations, even with individual data points, do not constitute a convincing and complete traceability record.

Sources: [2] [6]

How can we effectively assess metadata coverage with minimal effort?

Select a professionally relevant dataset containing a known change and define specific search and calculation queries. Verify these queries after export or reactivation using the intended role. The scope must be justified in relation to the process. Passing a simple example test does not automatically demonstrate coverage of all methods, interfaces, and historical formats.

Sources: [6] [4]

Sources and review date

Source date: 7 October 2026. The technical examples relate to human medicinal product GMP. FDA and MHRA statements are incorporated within their respective areas of applicability. Tables and numerical examples are intended as internal reference tools.

  1. European Commission: EudraLex Volume 4Current Directory, retrieved 07.10.2026; GMP requirements applicable to medicinal products for human use: Chapter 4 January 2011; Annex 11 Revision January 2011; Chapter 1 and Chapter 6
  2. PIC/S: PI 041-1, Data Management and IntegrityPIC/S PI 041-1, 1 July 2021. See section 3.7 for status, sections 7.5–7.8 and 8.1–8.8, sections 9.1.5 and 9.3 and 9.5–9.10, section 12.1 and the glossary.
  3. FDA: Data Integrity and Compliance With Drug CGMP, Questions and AnswersFinal Guidance, December 2018; Section III Q1(b)–(d), Q2, Q4–6, Q7–10, Q12–13, and Q18
  4. European Commission: EU GMP Chapter 6, Quality ControlRevision 2014; effective from 01.10.2014; §6.7 Documentation; §6.9 Out-of-Specification/Out-of-Trend; §§6.15–6.17 Test methods, calculations, and laboratory records
  5. European Commission: EU GMP Chapter 4, DocumentationJanuary 2011; in effect since 30.06.2011; Principle; §§4.1–4.9, in particular timely recording 4.8 and visible correction 4.9; §§4.10–4.12 Storage
  6. European Commission: EU GMP Annex 11, Computerised SystemsRevision January 2011; from 30.06.2011; Principle; §§1,4,5,7,9–13,17; §12.4 Identity, Date and Time; §10 controlled changes
  7. MHRA: GxP Data Integrity Guidance and DefinitionsRevision 1, March 2018; Sections 6.1–6.11 Data/Metadata/Processing; Sections 6.12–6.17 Transactions, Audit Trail, Review, Access, and Retention

Sources checked: .

Daniel Herrmann, Daniel Herrmann Consulting

Daniel Herrmann Consulting

Clarify the required data context specifically

We support you in aligning a specific outcome process with traceable metadata, stable mappings, and verifiable exports. Together with your department, we assess the existing systems and the actual pending decisions.

  • 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 →