PharmaCalc

Acceptance Quality Level (AQL) sampling is an attributes inspection method (defects: pass/fail) that defines the maximum proportion of defects acceptable in a lot. It is based on ANSI/ASQ Z1.4 (Sampling Procedures and Tables for Inspection by Attributes) and ISO 2859-1, the international equivalents. AQL sampling is widely used in pharmaceutical QC for incoming material inspection, packaging inspection, and final product release.

Key concept: Rather than inspect 100% of a lot (100% inspection), AQL sampling provides risk-managed statistical assurance through strategic sampling. The supplier and customer agree on an AQL (e.g., 0.65% for critical defects, 1.0% for major, 2.5–4.0% for minor). The sampling plan defines: lot size → sample size code → sample size (n) and acceptance number (Ac). If ≤ Ac defects found in n samples, lot is ACCEPTED. If > Ac, lot is REJECTED.

Inspection Levels: Normal (default), Tightened (after failures), and Reduced (after consistent acceptances). Tightened increases stringency; Reduced decreases sample size when supplier is reliable.

AQL sampling uses two-stage lookup tables:

Step 1: Sample Size Code Letter — From Table I, based on lot size N and Inspection Level (typically II = Normal):

Step 2: Sampling Plan (n, Ac, Re) — From Table II-A (Normal Inspection), using Code Letter and AQL:

Example: Lot size 2,500, Normal Inspection Level II, AQL = 1.0%:

• From Table I: N=2,500 → Code Letter K
• From Table II-A: Code K, AQL 1.0% → n=125, Ac=3, Re=4
• Inspect 125 units. If ≤3 defects → ACCEPT. If ≥4 defects → REJECT.

Severity Switching (Z1.4 rules — severity, not inspection level, is what switches):

• Normal → Tightened: when 2 of any 5 or fewer consecutive lots are not accepted on original normal inspection
• Tightened → Normal: after 5 consecutive lots accepted on original tightened inspection
• Normal → Reduced: after 10 consecutive lots accepted (with steady production and responsible-authority approval)
• Reduced → Normal: when a lot is not accepted, production becomes irregular, or other conditions warrant
• Discontinue inspection: after 10 consecutive lots remain on tightened inspection

• Scenario: Incoming inspection of tablet packaging. Lot size N = 2,500 units. Agreed AQL = 1.0% for critical defects (wrong label, damage). Use Normal Inspection Level II. Determine sampling plan and decision criteria.
• Step 1: Find Sample Size Code from Table I.
  ANSI/ASQ Z1.4 Table I Lookup

  Lot Size: 2,500 units
  Inspection Level: II (Normal)
  Code Letter: K
• Step 2: Find Sampling Plan from Table II-A (Normal Inspection).
  Table II-A Lookup: Code Letter K, AQL 1.0%

  Sample Size (n): 125 units
  Acceptance Number (Ac): 3 defects
  Rejection Number (Re): 4 defects
• Step 3: State the Sampling Plan and Decision Rules.
  Normal Inspection Plan:
  • Inspect 125 units randomly from 2,500-unit lot
  • Count defects (critical issues: wrong label, physical damage, etc.)
  • If defects ≤ 3 → ACCEPT LOT ✓
  • If defects ≥ 4 → REJECT LOT ✗
• Step 4: Example Inspection Outcomes.
  Scenario A: Inspect 125 units, find 2 defects → 2 ≤ 3 → ACCEPT
  
  Scenario B: Inspect 125 units, find 4 defects → 4 ≥ 4 → REJECT
  
  On Rejection: Supplier is notified. If 2 of any 5 or fewer consecutive lots are not accepted, switch to tightened inspection for subsequent lots (same or larger sample size with a stricter acceptance number — never a smaller sample). May require 100% inspection or supplier audit.
• Step 5: Inspection Level Switching Strategy.
  Ongoing Scenario (Multiple Lots):
  Lot 1: Normal Inspection → 2 defects → ACCEPT
  Lot 2: Normal Inspection → 1 defect → ACCEPT
  Lot 3: Normal Inspection → 0 defects → ACCEPT
  Lot 4: Normal Inspection → 0 defects → ACCEPT
  Lot 5: Normal Inspection → 0 defects → ACCEPT
  
  Five consecutive acceptances is the tightened → normal criterion. Switching to REDUCED inspection (smaller sample, cost savings) requires 10 consecutive lots accepted, steady production, and responsible-authority approval.
  
  If a lot is not accepted on reduced inspection → revert to Normal and investigate the supplier.
• Interpretation and Action:
  AQL Sampling Plan: n=125, Ac=3, Re=4
  
  This plan provides statistical confidence that the lot meets AQL 1.0%. Its operating characteristics (binomial, n=125, Ac=3): a lot at the 1.0% AQL is accepted with ≈96.2% probability (producer\u2019s risk ≈3.7%); the 10% consumer-risk point (LTPD) is ≈5.3% defective, and a lot at 6.1% defective is accepted only ≈5% of the time. Continue sampling, with severity switching as supplier performance warrants.