Medical Device Notified Body Feedback – EU MDR: A Case Study

Getting notified feedback on your medical device clinical evaluation can be overwhelming.

Even if the changes required are minimal, the feedback sheet typically doesn’t look minimal.

They do like their text and technical expressions, don’t they?

Knowing how to handle notified body feedback requires understanding what they want.

And often, what they want is relatively easy to fix (depending on the quality of the original clinical evaluation, of course).

In this article, we will review real notified body feedback sent to us by a client looking to fix their clinical evaluation report.

Obviously, all names, identifiers, and otherwise compromising details have been removed, and text may be reworded to protect our client, but the feedback and solutions are real.

This particular clinical evaluation was relatively easy to fix, and we did not have to go through any significant reworkings.

It is not always this easy, although we often see some of these errors.

And remember – clinical evaluations are hard.

They’re complex.

They require specialized skills and experience to get right, which is why many companies outsource them if they can.

There are no “dumb” errors, and we all make mistakes.

Notified body feedback is a learning experience, and while some might be farther ahead in their learning journey, there is nothing wrong with being at the start of the process.

So, let’s get to it!

Non-conformity 1: The choice of search terms is not clear.

Description: the client had included four similar devices in their clinical evaluation plan and report but had only searched for three during the literature search.

The notified body is looking for an explanation for why the fourth search term has been excluded from the literature search.

There is no explanation as it happens – the fourth device was simply overlooked during the search.

Solution

Performing an extra search and including the fourth similar device as a search term.

 Non-conformity 2: Methods used to avoid bias were not described.

Description: the clinical evaluation did include a line about bias, but it needed to be adequately clear how the manufacturer planned to avoid bias during the clinical evaluation.

Solution

Include more details about avoiding bias during searches and writing the clinical evaluation report, especially when determining the benefit-risk profile and its acceptability.

In this case, avoiding bias was done by the clinical evaluation report being reviewed by a person from a completely different department than regulatory from the manufacturer’s side, as well as hiring an external company to perform the clinical evaluation.

From our side, a second and senior medical writer continuously reviews all of our clinical evaluations to ensure we avoid bias and that all information in the evaluation is correct and appropriate.

 Non-conformity 3: Parameters of similarity were not established.

Description: the manufacturer had included four similar devices in their clinical evaluation, listed in bullet points. There were no justifications or rationale for those similar devices’ inclusion.

Solution

Adding a paragraph about how similar devices are identified and the parameters of similarity, as well as a simple table showing the similar devices, their intended use, and the technical characteristics they share with the device being evaluated.

Non-conformity 4: The methodology of evaluation and appraisal of relevancy to the subject device is not presented.

Description: There were no established inclusion/exclusion criteria for adverse events picked up during the adverse event search.

Solution

Adding a list of inclusion and exclusion criteria for adverse events, such as:

Inclusion criteria:

• Adverse events related to the target device and/or similar devices.
• Adverse events related to similar technologies or functional characteristics.

Exclusion criteria:

• Adverse events unrelated to the target device, similar devices, technologies, or functional characteristics.
• Adverse events with insufficient data to clearly identify the device in question.
• Adverse events with insufficient data to clearly identify the impact of the adverse event on the patient.
• Adverse events outside the established period.
• Adverse events that appear to report on the same event as another report (i.e., duplicates)

 Non-conformity 5: Not all retrieved studies were appraised for data contribution.

Description: The manufacturer had only extracted data from articles on their own device, not on articles referencing similar/competitor devices.

Here, the notified body is looking for either a justification for why not all articles were appraised or data appraisal for all the relevant articles.

Solution

Extracting data for all the relevant articles identified for the target device and similar and competitor devices.

It has previously been enough to only extract data on the target device for some notified bodies, but after the introduction of the MDR, that ship has sailed.

All articles related to the target and/or similar devices should undergo data extraction and appraisal.

Non-conformity 6: Deviations from the protocol, if any, are not identified in the literature search report.

Description: there were no deviations from the protocol, which was not stated in the literature search or clinical evaluation report. Even if there have been no changes, specifically saying so is always worth it.

Solution

Added a section in the literature search and clinical evaluation reports stating no protocol deviations.

 Non-conformity 7: Discrepancies in the number of included articles.

Description: the number of included articles is included in various places in the literature search report and clinical evaluation report; in this case, the numbers differed between reports and in the same report.

While this is a simple mistake, it is a non-conformity we often see.

The number of included articles can change until very late in the clinical evaluation process, and there are usually one or two places in the report that people need to remember to change.

The last thing we do at Citemed before sending any document is check all the numbers everywhere in our reports because, somehow, it is so easy to get it wrong.

Solution

Review all numbers everywhere and make appropriate updates.

Conclusion

It can be so daunting when you get that email from your notified body, and you go to open the Excel sheet, and it’s over 30 rows long.

But once you go through it, what typically happens is that you’ll notice that you’re missing a clarification here and there, a justification or rationale in a few places, and a few re-counts.

Likewise, many non-conformities can be fixed with the same corrections.

Take a few breaths, start from one end, and sooner than you realize, you’ll be ready to send it back!