Regulatory Open Forum

 View Only

  • 1.  Preventive Action Categories

    Posted 03-Sep-2020 12:31
    I'm seeking a little guidance on best practices here. We're (device manufacturer) putting some effort into developing our Preventive Action program and looking to create some categories to help us track and trend in a similar manner to our Corrective Actions. However, the categories for CA failure modes don't exactly translate to proactive measures taken in PAs. Some examples off the top of my head would be: d&d process improvement, manufacturing process improvement, QMS improvement, documents/records control, infrastructure, etc.

    Does it make sense to simply follow ISO 13485 QMS subsystems, or does anyone have recommendations for creating other categories? 

    Thanks!!

    ------------------------------
    Michael Thomas
    Regulatory Affairs Manager
    Salt Lake City UT
    United States
    ------------------------------


  • 2.  RE: Preventive Action Categories

    Posted 04-Sep-2020 04:53
    Hello Michael,

    Preventive action sources and activities can be a bit more elusive than the corrective action counter-parts.  From my own perspective you are heading in the right direction to identify preventive action categories originating from different "areas" instead of say like corrective action which would be related to a specific process or product.  You could theoretically have preventive action grouped by each product, though again the type of preventive action could vary so significantly.  We had used a general 'design improvement' preventive action category which was a suggestion by customer, raised during manufacturing, or even an engineer thinking about a new improvement in the design of the product.  What I think you will find is there is limited number of preventive action categories from a QMS trending perspective as compared to corrective action.  Though in same cases you could have a one-to-one relationship with preventive action and corrective action categories - just you might be looking at higher level categories resulting in maybe 6 - 8 categories instead of being able to drill down into having much more granular data like 15 - 20 categories.  Also keep in mind generally the number of preventive action instances is much, much less than corrective action instances, so only having 5 - 6 categories might already be enough.  Again you are on the right track where we had put in preventive action categories usually following areas of a QMS: development, manufacturing, suppliers/outsourcing, infrastructure, training, distribution, etc.  Then as you get more data, you could break those down further as needed.

    ------------------------------
    Richard Vincins RAC
    Vice President Global Regulatory Affairs
    ------------------------------

    Original Message


  • 3.  RE: Preventive Action Categories

    Posted 04-Sep-2020 07:37

    You are looking for metrics for the Preventive Action System. I recommend three.

    Preventive Actions are improvement activities in the QMS designed to keep nonconformities from happening. For categories consider the ISO 13485:2016 clauses are the X.X level. Classify each preventive action by the clause it affects and report the results with a Pareto Analysis. Note that the MDSAP grading system uses the same clause level, so this provides an opportunity to compare the Pareto charts.

    The second metric is Preventive Action aging. This helps identify any Preventive Actions that are open and not getting enough attention. If you only have a few (five or fewer) a line graph of the number of days open for the oldest one is a good approach. If you more than a few, consider side by side box and whisker plots.

    The third metric is time to closure. This implies you have a good definition of closure. In one case you could identify actions and closure means you have implemented them. Distinguish this from an effectiveness check which could have a lot of "dead time" while waiting for some event to happen. I recommend using the end of activity and not include the dead time. Because the distribution of activity time typically has a long tail to the right the metrics are the median and the ninetieth percentile. The mean is strong influenced by extreme values, so it could be distorted by one Preventive Action. For a graph you could use a histogram of the days to closure.



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

    Original Message


Related Content