Regulatory Open Forum

 View Only
  • 1.  Complaints /Adverse events trend analysis

    Posted 07-Nov-2017 19:08

    Hi ,


    Does anyone have experience with performing trend analysis for complaints and /adverse events. Could you share the best practices / guidance for the same.


    Thanks & Regards,


    Rashmi Pillay
    Regulatory Affairs Associate


    3-4 Second Avenue

    Mawson Lakes SA, 5095


    T + 61 8 7074 8105


    One Powerful Vision.

    Confidentiality: This e-mail is from Ellex Medical Pty Ltd, ABN 35 008 276 060. The contents are confidential and intended only for the named recipient of this e-mail. If the reader of this e-mail is not the intended recipient you are hereby notified that any use, reproduction, disclosure or distribution of the information contained in the e-mail is prohibited. Viruses: Any loss/damage incurred by using this material is not the sender's responsibility. No warranty is made that this material is free from computer virus or other defect. Ellex Medical Pty Ltd entire liability will be limited to resupplying the material. If you have received this e-mail in error, please reply to us immediately and delete the document. 



  • 2.  RE: Complaints /Adverse events trend analysis

    Posted 07-Nov-2017 20:27
    Hi Rashmi - I have a lot of experience on this topic for Medical Devices. Analysis of complaints data, including adverse events, is an important part of the Quality system. There are many analytical methods (trending, time series analysis, disproportionately analysis, test analytics etc.) that can be used to analyze complaints data. Here is a FDA white paper that provides some guidance:

    Data Mining at FDA -- White Paper

    However, it is more important to first clearly define the objectives of this analysis. Are you doing this as part of a post-market surveillance process? How do you plan to identify potential quality/safety signals based on complaints data? What criteria would you use to take action (or not take any action) based on data? 

    I will be happy to discuss more if you like. Please send me an email. 

    Best regards

    Naveen Agarwal

    Naveen Agarwal
    Creative Analytics Solutions, LLC
    Jacksonville FL
    United States

  • 3.  RE: Complaints /Adverse events trend analysis

    Posted 08-Nov-2017 01:55
    Hello Rashmi,

    Might be something in there that you could use as a benchmark.  After you have completed the trend, use the MHRA form found here for reporting trends (Trend Report).

    Best of luck, and do let me know how your report turns out.


    John Beasley, RAC (US)
    Founding Member and Senior Consultant
    MedTech Review, LLC

    257 Garnet Garden Street
    Henderson, NV 89015

    USA:  +1 612-889-5168
    SKYPE:   medtechreview

    All views given by MedTech Review consultants on the interpretation of medical device regulations represent our best judgment at the time. Such views are not meant to be a definitive statement of law and we would advise you to seek the views of your own professional advisors. 

    This electronic transmission is strictly confidential between the sender and the intended addressee.  It may contain information which is covered by legal, professional or other privilege. If you are not the intended addressee, or someone authorized by the intended addressee to receive transmissions on their behalf, you must not retain, disclose in any form, copy or take any action in reliance on this transmission. If you have received this transmission in error, please notify MedTech Review, LLC as soon as possible and destroy this message.

  • 4.  RE: Complaints /Adverse events trend analysis

    Posted 08-Nov-2017 07:11

    A good place to start is the EU MedDev on Vigilance Reporting, MedDev 2.12-1 Rev. 8. It has a section on Trend Reporting and says, "Appendix C of GHTF SG2 document N54 'Global Guidance for Adverse Event Reporting for Medical Devices' provides useful guidance."


    You should expect that every adverse event and nearly all complaints are associated with at least one row in your ISO 14971:2007 Hazard Analysis (assuming the typical Excel workbook approach). The row will have steps from Hazard, Sequence of Events, Hazardous Situation, to Harm. For each harm you would have estimated the severity and its frequency of occurrence.


    For the analysis, I recommend classifying each complaint or adverse event by the unique identification number assigned in the Hazard Analysis. Part of the implementation of ISO 14971:2007 Clause 9 determines if the frequency of occurrence (this is a rate not the number of events), is greater than your initial estimate.


    An SPC like approach is a good technique to start. Using the GHTF guidance, set the baseline at the estimated rate from the Hazard Analysis. Set the threshold high enough to detect signals, while reducing false alarms. Start with the threshold at 1.5 times the baseline rate. Also, consider using a moving average (three months) to smooth out spikes in the data. For example, some reporters may send reports in a batch instead of when they occur. This would cause a false alarm from the spike. Plot the monthly (smoothed data) as a time series along with the baseline and threshold. Apply "Shewhart rules" to detect a signal. (While not technically correct, the method is pretty robust for signal detection.)

    Dan O'Leary
    Swanzey NH
    United States