Software change management

[Galac]

Hi,
In this transition period between MDD and MDR, our notified body requires us to submit all our changes and our analysis of these changes (significant/non-significant) before any release of the software product.
This situation leads us to incredible situations where the NB ends up blocking everything under the pretext that it impacts usability. Moreover, they themselves admit that they do not understand everything about our technologies. This leads to numerous exchanges by phone or email that take weeks and weeks. It's impossible to get feedback in less than 2-3 weeks and to have direct contact with the person who reviews our files. It's endless.
On our side, we want to do things right 1/ for our users 2/ for regulatory compliance.
Would you have any advice to unblock the situation please?

 

[indubioush]

Maybe consider switching notified bodies if you are unhappy.

 

[Tidge]

Hi,
In this transition period between MDD and MDR, our notified body requires us to submit all our changes and our analysis of these changes (significant/non-significant) before any release of the software product.
This situation leads us to incredible situations where the NB ends up blocking everything under the pretext that it impacts usability.

Even if you aren't formally required to have prepared/submitted a software architecture because of safety classification (per 62304) or level-of-concern (per FDA), having one allows you to establish traceability to the tests necessary for regression testing... including any usability testing.

I can't make a personal assessment of what you are changing (in specifications or implementations) and have no idea of the level of summative testing that was originally done, so I don't have a "hot take" on whether or not the NB is over-stepping. When it comes to NBs and NRTLs there is a difference between "hand-waving" and "evidence-waving". My experience with NRTLs has been incredibly salty with respect to consensus standards that are not of the 60601-1 family (i.e. collaterals and particulars). I agree that it is quite frustrating to have an outside person at an authority (such as a test lead at a NRTL) ask a specific question in an area that involves a consensus standard like 14971, 62304, 62366-n, and when given the answer they don't know how to process it... because of ignorance and unfamiliarity with the consensus standard in question.

My response/approach has evolved to go something like this: "The thing you are asking about is something we put a lot of critical thought into and have lots of evidence to support our actions. The way you phrased the question inspires us to want to dump a LOT of records on you but we feel like you might need as much time to make your assessment as we took in ours along the path, and that is almost certainly not a good use of time. We aren't intended to fool you, but we realize there is likely a mere handful of documents that would satisfy your request. Do you want to consult with the home office, or would you like our process expert to walk you through the general process at a high level to help you decide which records would be most valuable for you?"

In the specific case of software and user interface questions, I'd have the architecture and project planing documents ready, along with the final reports... and assuming that my team didn't repeat summative testing I would have the past summative testing available, along with any formative testing done specific to the recent change.