Risk-Based Laboratory Calibration: Start with the Measurement

An effective calibration programme starts with the decisions supported by measurements, not the intervention calendar. Identify what you measure, over which range and against which requirements; then define calibration, intermediate checks, review of results and responsibilities. The interval is one part of the programme. It cannot compensate for calibration at irrelevant points, inadequate uncertainty or a certificate that nobody evaluates. A balance, a pipette and a probe may support different decisions and need different controls, even within the same laboratory.

This guide proposes an original strategy for pharmaceutical QC, a matrix and a simulated case. It assigns no universal frequencies and does not replace approved procedures, metrological competence or requirements applicable to an individual measurement.

1. Connect the measurement to the laboratory decision

For each activity, describe the quantity, actual working range, conditions and decision that depends on the result. A mass used to prepare a standard, a volume transferred in a dilution and a temperature associated with a sample are more than three numbers to record. They can change the interpretation of an analytical result or the assessment of storage conditions.

GuideGxP recommends working back from the method or material requirement to the allowable measurement contribution. Do not automatically copy the instrument’s commercial specification as the fitness criterion. Ask whether the range, resolution, uncertainty and conditions of use support the intended decision. Where the connection is unclear, record the open issue and involve the method owner, not just the person booking calibration.

The EU GMP framework calls for suitable equipment and relevant metrological controls. Chapter 3 addresses calibration and checks of measuring instruments at defined intervals, using appropriate methods and records. Chapter 6 places equipment and resources within the quality-control system. [3] [4] The matrix proposed here organises this reasoning; it is not a form prescribed by those chapters.

2. Build an inventory of measurement chains

The inventory must identify the configuration producing the data. Record its identifier, function, location, owner, working range and relevant components. A probe connected to a reader or acquisition system may require consideration of the entire chain. A certificate for the sensor alone does not automatically demonstrate the performance of the installed combination.

Connect sensors, channels, accessories, software processing the value and the references used. If components are assessed separately, explain how results are combined and which interfaces still need verification. If calibrated together, identify the combination covered by the certificate. Changing a reader or channel must not become invisible within the programme.

Include reference standards and instruments used for intermediate checks. A check offers little reassurance if its reference is unsuitable or uncontrolled. Distinguish equipment that is available, restricted to particular uses, out of service or awaiting review. A label can make status visible, but it must point to reliable, current information rather than replace it.

3. Distinguish calibration, verification, adjustment and qualification

In the VIM, calibration relates reference values, indications and their uncertainties, and uses that relationship to obtain measurement results. Verification evaluates evidence that specified requirements are met. Adjustment changes a system so it provides the intended indications; it is not another word for calibration. [1] Clarify in the contract and records which activities will take place and in what sequence.

An intermediate check monitors defined aspects between calibrations. It needs a purpose, reference, conditions, criterion and action following an abnormal outcome. Passing it does not demonstrate functions that the check never challenges. A check at one point, for example, does not automatically describe the whole working range.

Maintenance concerns upkeep or restoration of equipment; qualification assembles evidence against intended use in its context. Annex 15 provides the qualification and lifecycle framework. [5] Calibration, qualification and method validation support each other, but one activity does not automatically encompass the others. A plan calling all of them “calibration” makes it difficult to understand what was actually demonstrated.

4. Define the calibration service before purchasing it

The request should describe identification and configuration, quantities, relevant range and points, conditions, required uncertainty and expected documentation. Choose points to support use: covering a nominal range does not guarantee representation of the laboratory’s decision-relevant regions. If the available service does not meet the need, expose the gap before the intervention.

Define how data before and after any adjustment will be handled. Initial conditions help assess previous use; subsequent results describe the state after intervention. A report containing only the final state may leave the most urgent question unsupported: how did the instrument perform when generating results already used? NIST also highlights the value of data before and after intervention when reviewing intervals. [7]

Assess supplier competence, method, references and measurement capability against the request. ISO/IEC 17025 concerns laboratory competence, impartiality and consistent operation and is used for accreditation. It is not equivalent to GMP and does not make accreditation mandatory for every QC laboratory in every situation. [6] Where accreditation is required or chosen as evidence, check the actual service scope, not just the presence of a symbol.

Metrological traceability concerns the result: it requires a documented link to the reference through calibrations and their uncertainties. It differs from tracking a serial number and does not, by itself, guarantee fitness for use. [1] Request evidence relevant to the quantity and service required, avoiding generic “traceable” labels without context.

5. Choose an initial interval and plan its review

ILAC-G24/OIML D 10:2022 proposes risk-based initial selection and data-supported review. Relevant factors include uncertainty needs, use, environment, stability, transport and intermediate checks. No single method is best for every instrument. [2] NIST specifies no general fixed interval; applicable external requirements must also be considered. [7]

To make the choice operational, GuideGxP proposes retaining the initial rationale and its assumptions. Infrequent use is insufficient justification if equipment can change during storage; frequent use alone does not justify a calendar formula either. Describe what is known and what still needs observation, avoiding scores that produce a date without a verifiable rationale.

Review must distinguish instrumental drift, test variability and changes in conditions. Compare like-for-like data, retain intervention information and assess whether checks detect relevant failures. An extension needs adequate evidence and approval under the local system; an approaching deadline is not shifted merely to solve an organisational problem.

Also identify events needing evaluation before the due date: impact, transport, repairs, replacements, abnormal results or a new use. Not every event demands the same action, but each must be assessed against the affected functions. The programme must explain who decides whether to restrict use, verify, recalibrate or reassess qualification.

6. Original matrix for programme governance

The following matrix connects measurements and controls without assigning universal intervals or limits. Criteria must be defined and approved locally. The review column identifies information that may change a decision, not automatic permission to extend an interval.

Add identifiers, versions, due dates, status and links to records in the operational register. Where several people contribute, distinguish execution, technical evaluation and authorisation. Do not assign everything generically to “QA” if the judgement requires metrological competence that must be specifically identified.

7. From the certificate to a fitness decision

On receipt, compare the service requested with the service delivered. Check identity, configuration, range, points, conditions, results, uncertainty and references; assess limitations, adjustments and missing information. Filing a certificate without review does not conclude the use decision. Interpret any conformity statement using the criteria and decision rule actually applied.

If a result exceeds an approved limit or casts doubt on fitness, preserve data and the initial state, control instrument use and initiate the defined abnormality process. The investigation must identify potentially affected measurements, relevant period and consequences for decisions, using the available history. Do not assume that every previous result is wrong or that adjustment cancels the impact.

Return to use requires a documented decision with clear restrictions where applicable. Acceptable calibration, functional checks, configuration and qualification status may contribute in different ways. Make any remaining limitation visible to users. In the workflow recommended by GuideGxP, “certificate received” and “authorised for use” are distinct states.

8. Simulated case: balance, pipette and probe in one QC laboratory

A laboratory manages all three instruments with a single annual reminder and no documented rationale. This is a simulated case: it assigns no real results to the instruments and does not imply that a particular duration is always wrong. The problem is that the programme does not connect its calendar with actual measurements.

For the balance, the owner reconstructs the preparations and mass range used. They discover that the requested service has not been compared with the quantities most demanding for the method. Before choosing the interval, they define requirements, calibration coverage and useful intermediate checks. They do not invent a universal ratio between resolution and minimum mass.

For the pipette, they document the body, tips, volumes and liquids used. Calibration under the service’s stated conditions remains one part of the evidence; applicability to use conditions needs evaluation. Operator competence and technique suitability are considered separately from metrological status. A pipetting anomaly is not automatically attributed solely to the pipette.

For temperature, they identify sensor, reader and channel. The previous programme retained only the probe certificate. The group establishes which combination was covered and what further evidence is needed. Sensor calibration alone demonstrates neither uniformity throughout the refrigerator nor correct operation of alarm notification.

The review produces a programme differentiated by purpose, coverage and controls, with responsibilities and historical information to collect. Only then does the laboratory justify intervals and establish when to review them. The case assigns no new dates because the data needed to justify them are missing.

9. Manage deadlines, suppliers and changes

Allow time for intervention, transport, report receipt, review and gap management. The visit date does not necessarily coincide with the instrument becoming available for work. Identify qualified alternatives or other organisational measures where activity continuity requires them, without automatically authorising use beyond the due date.

During periodic programme review, consider missed deadlines, incomplete certificates, abnormal checks, unavailability, changed use and supplier performance. An “interventions completed” indicator does not show whether results were evaluated. Connect programme data with decisions and actions, avoiding counts that hide exceptions.

Common mistakes include confusing calibration with adjustment, accepting only final data, forgetting check references, extending intervals for convenience and treating ISO/IEC 17025 as synonymous with GMP. Another mistake is importing NIST documents headed “Good Measurement Practice” without context: that abbreviation does not automatically mean pharmaceutical Good Manufacturing Practices.

Operational conclusion

The strategy is sound when the laboratory can explain the requirement, coverage, control, evidence and decision for every relevant measurement. Start with use, preserve history before and after interventions and review the programme as information or risks change. The calendar should follow from this reasoning and remain connected to it.

Continue in the Calibration & Laboratory Metrology hub. To distinguish selection of a dispensing system from its metrological control, see Air Displacement or Positive Displacement Pipettes: How to Choose.

Sources and applicability

Metrological sources checked on 30 September 2026; EU GMP texts checked on 29 September 2026 within the human-medicines framework. VIM3 is the published vocabulary used; VIM4 working documents are not treated as the current edition. The official ISO/IEC 17025 catalogue, status and scope were consulted, not the complete standard. ILAC/OIML and NIST provide guidance to apply in the appropriate context; the matrix, case and workflow organisation are original GuideGxP recommendations.

1. JCGM/BIPM — International Vocabulary of Metrology — JCGM 200:2012 (VIM3). Third edition, corrected 2012; calibration, verification and adjustment definitions.
2. ILAC/OIML — Guidelines for the determination of recalibration intervals of measuring equipment. ILAC-G24:2022 / OIML D 10:2022.
3. European Commission — EU GMP Chapter 3 — Premises and Equipment. Revision effective 1 March 2015.
4. European Commission — EU GMP Chapter6 Quality Control. Revision effective 1 October 2014.
5. European Commission — EU GMP Annex 15 — Qualification and Validation. 2015 revision, effective 1 October 2015.
6. ISO/IEC — ISO/IEC 17025:2017 — General requirements for the competence of testing and calibration laboratories. 2017 edition, confirmed in 2023; public scope and status.
7. NIST — Recommended Calibration Interval. Page updated 29 May 2026.