Regulatory Open Forum

 View Only

  • 1.  Sampling method in finish product inspection

    Posted 14-Apr-2023 16:40
    Hi, 
    I have a question regarding the sampling method in the final inspection step of class II MDs (product realization) in manufacturing premises. Is it acceptable? For example, AQL=0.25

    Kr,
    Shahram




  • 2.  RE: Sampling method in finish product inspection

    Posted 14-Apr-2023 18:57

    Using a sampling plan at final inspection is acceptable with a caveat. Device class is not a factor.

    If you don't fully verify the process output, such as by using a sampling plan instead of 100% inspection, then you must validate the process.

    There are two views of 100% inspection. In one view, the predominant view, 100% inspection means checking every product for the critical to quality, CQT, characteristics. I've had people say that 100% inspection requires checking every product for every characteristic.

    You didn't say anything about the sampling plan, so I'll infer it is a lot acceptance plan. In the US, the common plans are Z1.4 and c=0. The c=0 plans have a strange OC curve such that the probability of acceptance drops off quickly. I've seen cases where poor quality requires rejection of almost every lot. In this case you are back to 100% inspection to screen the lot.

    Both Z1.4 and c=0 have provisions for switching rules. I recommend you implement them since they can dynamically adjust the sampling plan. A good quality history can reduce the sample size and still provide the desired level of protection.



    ------------------------------
    Dan O'Leary CQA, CQE
    Swanzey NH
    United States
    ------------------------------

    Original Message


  • 3.  RE: Sampling method in finish product inspection

    Posted 18-Apr-2023 11:07

    Shahram,

    There are many factors which go into inspection and testing of in-process product and finished product when using a sampling approach.  There are sampling methods and plans which can be applied such as ISO 2859 or the newer ISO 2859x series or the older military standard which is still used today.  However, sampling of finished goods does depend on many factors such as if the test and/or inspection is visual, qualitative, quantitative, using automated system, manual, etc.  When determining if sampling is needed or can be done this also should rely on the verification and validation testing which has been performed of the device to determine if there are things like CTQ - Critical to Quality attributes.  You might want to seek some expert advice or research into this as there are many articles and books on sampling methods, but it would be specific to the device.



    ------------------------------
    Richard Vincins ASQ-CQA, MTOPRA, RAC
    Vice President Global Regulatory Affairs
    ------------------------------

    Original Message


  • 4.  RE: Sampling method in finish product inspection

    Posted 19-Apr-2023 07:47

    A little over 18 years ago, I developed a general guide for setting sampling plans and acceptable quality level. I used FDA criteria for much of it. Also, CSP means Continuous Sampling Plan and differs from the concept of lot/batch acceptance.

    Suggested quality assurance levels and various means for achieving the levels for the various numerically designated "Specification Levels".

    For a level 1 Spec. Level

    1. FDA considers 4.0% as an acceptable quality level for patient examination gloves in terms of leakage.  Single sampling is done on lots less than 1200 gloves.  At 1200, a sample of 80 is taken, and passes if 7 or fewer defects are encountered. (< 8.75%). Variables plans for AQL may be used to reduce sample size.
    2. The average outgoing quality limit should be 4% or less.
    3. The minimum value of process Cpk is (2.06/3.0) =0.69
    4. A CSP-1, CSP-2, or CSP-3 plan with an AOQL of 5 to 7 could be used.

    For a level 2 Spec. Level,

    1. FDA considers 2.5% as an acceptable quality level for surgeon's gloves in terms of the attribute "leakage". Single sampling is done on lots less than 1200 gloves.  At 1200, a sample of 80 is taken, and passes if 5 or fewer defects are encountered. (< 6.25%) Variables plans (Z1.9) for AQL may be used to reduce sample size.
    2. The average outgoing quality limit should be 2.5% or less.
    3. The minimum value of process Cpk is (2.24/3.0)= 0.75
    4. A CSP-1, CSP-2, or CSP-3 plan with an AOQL of 2.9 to 5 could be used.

    For a level 3 Spec. Level

    1. FDA guidance for condoms is 0.4% as a maximum acceptable quality level or AQL based on the attribute leakage.  Typical single sampling (Z1.4) with 0 acceptance number is 32.  If we want to specify a 90% chance of acceptance with 0.4% defectives, the actual plan should use a sample size of 26 and acceptance number of zero.  Such a plan will reject 90% of the lots that are 8.48% defective. Variables plans (Z1.9) for AQL may be used to reduce sample size.
    2. The average outgoing quality level should be 0.4%
    3. The minimum value of process Cpk is (2.98/3.0) = 0.993
    4. A CSP-1, CSP-2, or CSP-3 plan with an AOQL of 0.3 to 0.8 could be used.

    For a level 4 Spec. Level

    1. Assurance by 100% automated inspection techniques, and/or validated processes with appropriate Cpk (> 1.0, 1.33 preferred) is required. 
      Cpk and percentage units outside specifications (normalcy assumed) are:

    1.0             0.13%

    1.33         0.003%

    1.67         0.0001% (1ppm)



    ------------------------------
    David Manalan
    Principal, INQC Consulting
    Acton MA
    United States
    ------------------------------

    Original Message


Related Content