Regulatory Open Forum

 View Only

  • 1.  Sampling

    Posted 21-Oct-2016 15:04

    What will be the number of failures, a percentage of failure or acceptable quality level necessary to fail a lot of material?

    ------------------------------
    Brenda Miller
    Combat Medical Systems
    Harrisburg NC
    United States
    ------------------------------


  • 2.  RE: Sampling

    Posted 21-Oct-2016 15:36

    For the U.S. market, the FDA requires that this decision be made using a statistically-valid rationale.  To help you with this, I recommend the standard called ANSI/ASQC Z1.4 - Sampling Procedures and Tables for Inspection by Attributes (i.e., ISO 2859-1).

    ------------------------------
    Kevin Randall, ASQ CQA, RAC (U.S., Canada, Europe)
    Principal Consultant
    ComplianceAcuity, Inc.
    Golden CO
    United States
    www.complianceacuity.com

    Original Message


  • 3.  RE: Sampling

    Posted 21-Oct-2016 16:19

    You are asking about acceptance sampling, which is a rich topic. (I teach a two-day workshop on the subject.)

    In the most common model, determine the characteristics for the product, often called Critical to Quality, CTQ, and determine if you would use attributes (conforming/nonconforming) or variables (actual measurements).

    For attributes, the most common published plans are ANSI/ASQ Z1.4 or the Zero Acceptance Number Plans (c=0) by Nick Squeglia.

    For variables, the common published plan is ANSI/ASQ Z1.9.

    FDA-CDRH recognizes consensus standards including Z1.4 and Z1.9. If you use either one, then you must implement the switching rules, or you are not following the standard.

    If you use sampling plans for in-process of final acceptance, then you will also need to implement process validation following 820.75.

    ------------------------------
    Dan O'Leary
    Swanzey NH
    United States

    Original Message


  • 4.  RE: Sampling

    Posted 24-Oct-2016 13:04

    This is a super complex topic, with no "single" answer. The answer will vary based on the size of the lot, the risk of the component in question, the known variability of the component and its related manufacturing process etc. Once that is known and assessed as part of your broader risk management system, you can select an appropriate statistical sampling methodology. That methodology will give you the "answer" in your specific case.

    g-

    ------------------------------
    Ginger Glaser RAC
    Vice-President, Quality and Regulatory Affairs
    Maplewood MN
    United States

    Original Message