01
Define intended use
State whether the set supports training, qualification, manual inspection, SAVI, AVI recipe work, comparability, investigation, or another approved study. One set should not be assumed fit for every purpose.
Evidence hub · Test-set governance
Visual Inspection Test-Set Lifecycle Management
A practical framework for intended use, representativeness, bracketing, maintenance, requalification, replacement, and retirement of manual, SAVI, and AVI challenge sets.
Short answer
A test set is a controlled measurement asset, not a permanent box of samples
As a practical, risk-based framework, suitability depends on a defined intended use and continuing evidence that the units still represent the product, container, defect population, inspection method, and decision being challenged. The sources cited here do not prescribe one universal test-set expiry, total size, bracketing rule, or requalification interval. FDA's nonbinding 2021 draft recommends that defective units typically not exceed 10–20% of the set, that the quantity provide adequate confidence, and that the approved protocol specify requalification methods and frequency.
YB practical lifecycle model · Site-defined
Five controls that keep a test set defensible
02
Establish representativeness
As relevant to the intended study, consider container, formulation, fill, defect material, size or severity, position, mobility, detectability, and defect-free units against the actual product and process risks being challenged.
03
Characterize and approve
A site-defined record may include identification, unique unit control, characterization method, reference status, acceptance criteria, handling limitations, storage, and the approved baseline before routine use.
04
Monitor during use
Consider controls such as issuance, masking, randomization, exposure history, damage checks, reconciliation, cleaning where appropriate, periodic verification, event history, and observed attribute drift according to intended use and risk.
05
Requalify, replace, or retire
Under the site's approved lifecycle procedure, use defined triggers and evidence to determine whether a unit, subset, or full set remains suitable, needs recharacterization, must be replaced, or should be retired from the intended use.
When can one set bracket several products?
Bracketing is an evidence question. An illustrative, risk-based rationale may compare formulation appearance and viscosity, container geometry and material, closure and label presentation, fill volume, defect mobility and position, inspection sequence, equipment, and worst-case conditions. Similar product names or container sizes do not by themselves establish a comparable inspection challenge.
A justified bracket may reduce duplicate assets, but the approved rationale should state what is covered, what is excluded, which attributes are worst case, and what future changes reopen the assessment.
Periodic review is not enough by itself
A site may supplement calendar-based verification with event-driven review. A unit can remain inside a nominal review interval and still become unsuitable after damage, handling, repeated exposure, formulation change, supplier change, equipment work, or a newly observed process defect.
The program should define what is inspected or remeasured, who can change status, how replacement units are linked to the prior version, and whether the change affects training, qualification, or validation records.
Change control · Illustrative site-defined triggers
Events a site may define for documented impact assessment
Product formulation, concentration, viscosity, color, clarity, or particle behavior changes
Container, closure, component, supplier, fill volume, label, overwrap, or presentation changes
Inspection station, lighting, handling sequence, line, site, or contract manufacturer changes
SAVI or AVI hardware, software, recipe, motion, image-processing, or maintenance changes
New process defects, complaint particles, investigation findings, or stability-created attributes emerge
Reference units are damaged, lost, contaminated, relabeled, repaired, remasked, or show attribute drift
Performance trending indicates that the set no longer separates expected detection levels or supports its intended decision
A trigger does not automatically mean that a full set must be rebuilt. It means the effect on intended use and representativeness should be assessed, documented, and approved. Outcomes may include no impact, targeted verification, partial replacement, recharacterization, study repetition, or retirement.
Method-specific design
Training, qualification, SAVI, and AVI do not create identical test-set needs
Manual inspection
For applicable EU GMP and WHO programs, manual inspectors are qualified at least annually using normally worn corrective lenses and appropriate defect-library and worst-case samples. Handling, sequence, product, container, and defect-population details remain study-specific.
SAVI
As industry-informed, site-specific practice, the set may address the combined equipment-and-human method, including presentation, motion, timing, handling, and changes in how the operator observes or rejects units.
AVI
Applicable EU GMP and WHO texts call for a validated method, detection of known quality/safety defects at least equal to manual inspection, and representative-defect challenges. Repeatability, presentation control, and defect-level analysis are equipment- and protocol-specific design choices.
Suggested lifecycle record (site-defined)
- Approved intended use and protocol linkage
- Master inventory, unique unit identifiers, version, and status
- Defect family, material or attribute, severity/detectability, location, and characterization evidence
- Representative-product or surrogate rationale and known limitations
- Storage, handling, access, exposure, reconciliation, and maintenance history
- Periodic and event-driven verification results
- Change assessment, requalification decision, replacement linkage, and retirement disposition
Common questions
Test-set lifecycle FAQ
Do visual-inspection test sets have one universal expiry period?+
The cited sources do not establish one universal period for every unit, defect type, product, container, storage condition, and intended use. The site should define risk-based verification and retirement criteria, including event-driven triggers, and preserve evidence that the attributes used for decisions remain suitable.
Can one test set cover several products or presentations?+
Potentially, but only with a documented bracketing or representativeness rationale. Differences in formulation appearance, viscosity, container geometry, closure, fill, defect behavior, inspection method, and worst-case conditions may defeat simple portability assumptions.
Does a training set automatically qualify for AVI validation?+
A training set is not automatically sufficient. It may be used only if it is shown and approved as fit for the AVI validation purpose, including the machine-relevant attributes, presentation control, defect distribution, and study design needed for the intended work.
What should trigger test-set requalification?+
The site's change-control procedure should define triggers. They may include product, component, supplier, line, site, method, recipe, maintenance, defect-population, investigation, or reference-unit changes. The response can range from a documented no-impact assessment to partial or full requalification.
References and source status
The labels below matter: regulation, compendial information, draft guidance, WHO guidance, and industry consensus do not have the same legal or procedural status.
EU GMP Annex 1 controls applicable EU sterile manufacture; FDA's December 2021 document is a nonbinding draft; WHO is guidance whose enforceability depends on local adoption; USP <1790> is official informational guidance and not compendially required; BioPhorum is nonbinding industry practice. The remaining lifecycle design choices on this page are site-defined and require justification under the applicable pharmaceutical quality system and risk-management process.
[1] Applicable EU GMP Annex 1 expectation
European Commission, EU GMP Annex 1 (2022), sections 8.30–8.33Addresses individual inspection, defect libraries and classification, manual-inspector qualification, automated-method validation, representative defects, and trending/investigation for applicable EU sterile medicinal-product manufacture.
[2] FDA draft guidance — nonbinding
FDA, Inspection of Injectable Products for Visible Particulates (Draft Guidance, 2021)Presents FDA's draft, holistic lifecycle approach to visible particulates in injectable products. It excludes subvisible particles and other physical-defect categories from that draft's stated scope.
[3] WHO guidance
WHO TRS 1044, Annex 2: GMP for Sterile Pharmaceutical Products (2022)Provides WHO guidance for individual visual inspection, controlled conditions, inspector qualification, and validated automated methods. Enforceability depends on applicable local adoption and requirements.
[4] USP <1790> — official informational guidance; not compendially required
USP General Chapter <1790>, Visual Inspection of InjectionsProvides a lifecycle framework for inspection capability, training and qualification, defect standards, and test-set considerations. Access may require a USP subscription.
[5] BioPhorum industry-practice paper (nonbinding)
BioPhorum, Creation, Maintenance and Use of Test Sets for Visual Inspection Programs (May 14, 2026)Industry guidance addressing test-set creation, bracketing, surrogate products, variability, integrity, maintenance, and use across inspection modalities.
Apply the framework
Need a lifecycle review for an existing or planned challenge set?
YB Inspection can scope product- and presentation-specific defect kits, review intended use and representativeness, define replacement documentation, and support manual, SAVI, or AVI study planning. Final criteria and validation decisions remain under the customer's quality system.