Is USP a numerical particle-count standard?

No — it is a methodology + binary release criterion. The numerical particle-count standard for sub-visible particles is USP (≥ 10 µm and ≥ 25 µm limits per unit). USP covers visible particles, defined operationally by inspection conditions, and demands 'essentially free' with AQL sampling.

Manufacturing · The complete guide

Visual Inspection Particulates Injectables

TL;DR

Every parenteral unit must be inspected for visible particulates and container/closure defects before release. USP <790> sets the binary release criterion (essentially free of visible particulates), USP <1790> sets the inspection methodology, and PDA TR-79 + Annex 1 (2022) define the manual + semi-automatic + fully automated inspection programmes that operationalise it. Inspection regime hinges on lighting (2000-3750 lux at the inspection point), background contrast (matte black + matte white), unit handling (swirl + invert sequence), inspector qualification (per-shift acuity testing + per-defect Knapp test), and rejection categories (critical / major / minor). AQL-driven sampling for routine release, automated inspection systems for high-volume products, and a Knapp test programme that links operator + system performance to particle-detection probability are the operational pillars of a defensible inspection programme.

Reviewed May 25, 2026· By V5 Ultimate compliance team· 2,700 words · ~13 min read

01USP <790> — the binary release criterion

USP <790> requires that each unit of parenteral product be 'essentially free' of visible particulates when inspected under prescribed lighting conditions. The standard does not give a numerical limit. Instead it sets a methodology + an AQL-based sampling plan (typically AQL 0.65 for critical defects, AQL 1.0-1.5 for major). Every unit is 100% inspected during manufacture; a statistical re-inspection or AQL sample is taken from finished batches for release confirmation. 'Essentially free' is operationally defined by the inspection conditions — what is visible to a qualified inspector under standard conditions IS a defect; what is not visible is not.

02Inspection conditions — lighting, background, viewing time

Parameter	USP <1790> requirement	Comment
Illumination at inspection point	2,000 - 3,750 lux (200 - 375 ft-cd)	Measured at the unit, not at the lamp; calibrated lux meter
Background	Matte black + matte white (alternating)	Black background reveals translucent particles; white reveals dark particles
Viewing time per unit	5 - 10 seconds per background = 10-20 seconds total	Insufficient time = false negatives; common operator deviation
Container handling	Swirl, invert, and tilt to suspend particles in liquid + check meniscus + cake	Static inspection misses settled particles
Inspector posture + distance	Comfortable distance (typically 25-35 cm); rest breaks; rotation	Fatigue degrades detection sensitivity
Acuity testing	Per-shift near-vision check; corrected to 20/30 or better	Vision degrades through the day

03Defect categories — critical / major / minor

Category	Examples	AQL
Critical	Foreign particles ≥ 100 µm, broken / cracked container, missing closure, container leak	AQL 0.10 - 0.65 typical
Major	Discolouration, fill volume out of spec, label legibility issue, container surface defect	AQL 1.0 - 1.5 typical
Minor	Cosmetic blemish, minor label imperfection	AQL 2.5 - 4.0 typical

04Inspector qualification + the Knapp test

Initial training — theoretical (USP <790>/<1790> + product defect catalogue) + practical (mentored inspection of known defect sets);
Initial qualification — Knapp test on a defect-seeded test set; per-defect detection probability ≥ pre-defined threshold (often ≥ 70% per defect, ≥ 95% per defect-type panel);
Re-qualification — typically annual + after any extended absence; same Knapp methodology;
Per-shift acuity check — near-vision card or equivalent; documented;
Rotation + breaks — typical 30-60 minute rotation off the line to maintain detection sensitivity;
Detection-probability tracking — per-inspector trend over time; downward drift triggers re-qualification.

05Automated inspection — Eisai / Bosch / Brevetti machines

High-volume products typically run on automated inspection systems (Eisai, Bosch, Brevetti, Stevanato) that use rotating-spindle illumination + multi-camera vision systems + specialised algorithms (moving-particle detection, shadow-detection, fill-level, cosmetic). Automated systems are qualified per the Knapp test methodology: a defect-seeded set is presented to the machine + to a panel of qualified manual inspectors; the machine's per-defect detection probability is compared and required to be statistically non-inferior. Periodic re-qualification + per-shift challenge set + per-batch sampling for confirmation are routine.

06Particulate investigation + source classification

Every rejected unit category-tracked + trended by SPC; rising trend triggers investigation;
Particulate-source identification — FTIR / SEM-EDX / Raman to characterise particle type (glass, stopper rubber, gowning fibre, API agglomerate, environmental contamination);
Cross-reference to upstream process — glass from vial manufacture / wash, stopper fragment from coring, fibre from gowning, API agglomerate from formulation;
CAPA — process correction, equipment intervention, supplier action (e.g. tighter glass-vial CoA), gowning review;
Inspection effectiveness review — was the trend missed at line? Was Knapp-based threshold appropriate?

07Common failure modes

Lighting below 2,000 lux at the inspection point — measured at lamp not at unit; common audit-finding.
Inspectors inspecting > 10 seconds per background but counting once — fatigue then degrades subsequent units.
No swirl / no invert — static inspection misses settled and trapped-in-meniscus particles.
Knapp test seeded with only easy-to-see particles — qualification artificially passes; commercial detection low.
Per-shift acuity check skipped — declining inspector vision goes undetected.
Rotation skipped on rush batches — fatigued operators on critical product.
Automated-system threshold tightened to reduce false rejects, with no Knapp re-qualification — true defects also missed.
Defect catalogue not updated as new commercial defect types emerge — operator sees a new defect and does not know how to classify.
AQL sample drawn from one position only — non-representative; positional defect concentration missed.
Investigation closure 'attributed to particulate from environment' without source-identification — recurrence not prevented.

08How V5 Ultimate runs visual inspection

Per-product defect catalogue: photographic reference + category (critical/major/minor) + AQL + change-control;
Inspector register: training + Knapp-test results + per-defect detection probability + per-shift acuity check + rotation log;
Inspection-station configuration: light meter result + background colour + viewing time SOP + auto-block on lighting deviation;
Per-unit defect capture: barcode-scan + tablet-or-button defect category capture; auto-tally per AQL plan;
Automated-system integration: per-unit machine result + Knapp-test history + per-batch challenge-set log;
AQL-release decision: per-batch tally vs AQL plan; auto-pass / auto-fail with per-defect-category breakdown;
Particulate-source investigation: link to upstream batch + supplier + environmental + gowning context; CAPA wiring;
Trend dashboard: per-product per-defect-type SPC; per-inspector + per-line + per-shift performance;
Inspection pack: per-batch inspection record + inspector qualification + acuity check + AQL decision — exports as one PDF.

Frequently asked questions

Q.Is USP <790> a numerical particle-count standard?+

No — it is a methodology + binary release criterion. The numerical particle-count standard for sub-visible particles is USP <788> (≥ 10 µm and ≥ 25 µm limits per unit). USP <790> covers visible particles, defined operationally by inspection conditions, and demands 'essentially free' with AQL sampling.

Q.What's the Knapp test?+

A statistical inspection-effectiveness test developed by Knapp + Kushner (1980). The same defect-seeded set is inspected multiple times (typically 10×) by multiple inspectors; per-particle detection probability is calculated. Used for inspector qualification, automated-system qualification, and inspection-condition optimisation. Industry-standard for the past 40+ years.

Q.Can we use a single matte-grey background instead of black + white?+

Not USP-compliant — <1790> explicitly requires alternating black + white because particle types differ in visibility against each. Some sites use grey for an additional inspection pass, but the black + white pair is the regulatory baseline.

Q.What about coloured or opaque products?+

Inspection sensitivity drops significantly. For opaque products (suspensions, lipid emulsions, lyophilised cake before reconstitution), visual inspection is limited to container/closure + cake appearance + fill. Particulate detection in suspension is augmented by post-reconstitution inspection or process-controls upstream.

Q.Are automated systems required for new products?+

Not required, but increasingly the default for high-volume parenterals. Annex 1 (2022) §8.30 + FDA expectation favours automated systems for repetitive inspection at scale; manual inspection retained for low-volume, specialty, or AQL re-inspection.

Q.What's the relationship between visible + sub-visible particulate testing?+

Complementary. <790> + <1790> govern visible particulates (≥ ~50 µm, depending on conditions). <788> governs sub-visible (≥ 10 µm + ≥ 25 µm cumulative limits per unit). Both apply to all parenterals; control strategy addresses both ranges and identifies particulate source across the full size distribution.