Continuous Flow Mfg Skip-Lot Sampling

1
Continuous Flow MfgSkip-Lot Sampling

• Domenick Amato
• Frank Gomez
• Lynn Torbeck

2
21 CFR 210.3

• 210.3(20) Acceptance Criteria means the product
  specifications and acceptance/rejection criteria,
  such as acceptable quality level and unacceptable
  quality level with an associated sampling plan
• 210.3(21) Representative samples

3
21 CFR 211.84

• Requirements
• Representative samples
• Number of containers to sample
• Amount of material to be taken
• Variability Process average
• Confidence levels
• Past quality history

4
Continuous Flow Mfg

• How to define a continuous process? Types of
  processes?
• How to define a lot or batch in a continuous
  flow process?
• Assumption of homogeneity of product?
• Assumptions about defects?

5
Types of Sampling Plans

• Variables Plans
• ANSI/ASQ Z1.9
• Attribute Plans
• ANSI/ASQ Z1.4
• ISO 2859

6
Sampling Plans Defined

• Population of units, lots or batches
• AQL Acceptable Quality Limit
• LQ Unacceptable Quality Limit
• Sample size
• Accept number
• Reject number

7
Risk Management

• Producers Risk The probability that a good lot
  is rejected by the customer. (Type I error or
  alpha)
• Consumers Risk The probability that a bad lot
  is accepted by the customer. (Type II error or
  beta)
• Levels not fixed but common values are 5 and
  10.

8
Assumptions for Sampling

• Representative samples
• Beginning, middle and end
• Top, middle and bottom
• Every 30 minutes or 100 units
• Random Sampling (difficult)
• Defects must be randomly distributed !
• Sample inspection is 100

9
Reduced Testing

• Smaller sample size. Using General Inspection
  Level I in place of Level II or III in Z1.4.
• Not inspecting all characteristics on the CoA or
  listed in the USP.
• Can use reduced testing in Z1.4.
• Use the switching rules in Z1.4.

10
Skip-Lot Sampling

• Not all lots are tested.
• Only a fraction of lots are inspected.
• Goal is to save money or time and still achieve
  quality products.

11
Types of Skip-Lot Sampling

• Test every nth lot, not random.
• Test i lots, if all pass go to a fraction, f,
  of future lots. Select i and f in advance. Lots
  selected at random.
• ASQ S1, i and f are determined from the data
  collected. Lots are selected at random using
  dice.

12
Assumptions for Skip-Lot

• Supplier and customer agree on a plan. Good
  communications.
• Faith and trust in an honest supplier. Good
  relations.
• Proven history of quality with a stable and
  continuous process.
• Clear criteria for action plan.

13
Discussion Topics

• Compendial requirements
• Regulatory expectations
• Current state, pro/con/concerns
• Desired state for excipients
• Fraud, counterfeiting, adulteration
• Processes Out of control.
• Changes without notification

14
References

• NIST
• http//www.itl.nist.gov/div898/handbook/pmc/sectio
  n2/pmc22.htm
•  
• http//www.itl.nist.gov/div898/handbook/pmc/sectio
  n2/pmc27.htm
•  
• http//www.itl.nist.gov/div898/handbook/pmc/sectio
  n2/pmc23.htm
•  
• ASQ (1996). ANSI/ASQ S1, Attribute Skip-Lot
  Sampling Program.
•  
• Schilling, E. (1982). Acceptance Sampling in
  Quality Control, New York, NY Marcel Dekker, pp
  443-451
•  
• Juran, J. and Godfrey, A. (1999). Jurans Quality
  Handbook, Fifth Edition, pp 46-31, 46-32, New
  York McGraw Hill.