Regulatory Open Forum

 View Only
  • 1.  European regulation 2017/745 and 2017/746 - Sampling of devices

    Posted 07-Jul-2022 09:15
    Dear all,

    I hope that all is fine for you.

    I have a question concerning the sampling of medical devices: do you have any methods or sources of information for establishing (and therefore justifying) the sampling chosen in particular in the context of design verification?

    Thank all for your help

    ------------------------------
    Pascale
    ------------------------------


  • 2.  RE: European regulation 2017/745 and 2017/746 - Sampling of devices

    Posted 08-Jul-2022 01:44
    Dear Pascale,

    This is only my suggestion :
    You can follow and arrive Sampling of Devices for Verification through Risk Based approach.
    You can create a small document for sampling plan for MD XXX and there you justify sampling chosen in particular in the context , you can also consider BOM /parts specially - Critical components  and arrive a strategy for Sampling plan for a Device. Later you can co relate Product classification also - as you know this also will be done based on the Risk ( Low /Moderate /High)

    Regards


    ------------------------------
    Jagadeesha Pampapathi
    Bangalore
    India
    ------------------------------



  • 3.  RE: European regulation 2017/745 and 2017/746 - Sampling of devices

    Posted 08-Jul-2022 08:26
    There are no specific requirements in the regulations. Fall back to EN ISO 13485:2016/A11:2021. It doesn't tell you how o answer the question, only provides a conditional requirement.

    Sampling is a big subject with a lot of tentacles. For example, incoming inspection, NB sampling for surveillance audits, design verification, etc.

    For design verification, you demonstrate that a design output matches the design input. The first problem is to ensure the design output has variability. For example, is the design input defines a power cord, the design output is a purchasing specification. You only need to read one copy of the purchasing specification to do design verification – no sampling. I've seen companies claim that they need to order power cords and check them at incoming inspection as part of design verification. This is not correct.

    Another common mistake confuses manufacturability with design verification. In most cases you don't not need to build the product for design verification.

    Use sampling as a last resort, since it typically requires making something – sometimes a lot of somethings. Use document review, drawing review, calculations, similar products, etc. whenever possible to demonstrate that design output matches design input.

    If you need to sample, there are two types: attribute and variable. In attribute sampling, set the acceptance number, number non-conforming, as 0, 1, 2, etc. Also set the confidence. Together, they determine the sample size. For attribute sampling, the sample sizes tend to be large.

    For variables sampling you need to use measurements and set the acceptance criteria based on the specification and the statistical distribution of the measurements. The sample sizes tend to be smaller, because you have information.

    A Google search will yield the techniques for both attribute and variable sampling for design verification. In particular, look for information from Steve Walfish; he does a lot of good work in this area.

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



  • 4.  RE: European regulation 2017/745 and 2017/746 - Sampling of devices

    Posted 09-Jul-2022 02:43
    Dear Jagadeesha and dear Dan,

    Thanks a lot for your valuable advice.

    Have a nice week-end.



    ------------------------------
    Pascale
    ------------------------------