How to address unresolved Medical Device Software Anomalies?

[Roland chung]

Hi all,

We got feedback from FDA that the submission should include a list of all unresolved software anomalies for our devices (Moderate Level of Concern).

But we believe that all the bugs or defects have been solved before the product release.

Can you please share your experience on handling this problem?

Many thanks,
Roland

 

[yodon]

I typically create a Version Description Document for each software release that describes the complete configuration, how it was built, etc. I have a section for "Known Anomalies / Open Issues" where I list anything related to the software from the issue tracking system. Along with the known issues, I describe why (risk based) it's ok for the software to be released with the anomalies.

If there are none, I state "There are no known issues with the software at the time of the release."

One fine point to consider: if your software contains SOUP / COTS, it would be good to also check with the supplier to see if there are any known issues. You would list those and also do a risk analysis to justify using the SOUP / COTS with the anomalies.

So far, that's been acceptable.

I can just envision a dilbert-esque auditor asking for a list of unknown anomalies.

 

[MedTechSoftware]

Hi Roland,

It is generally accepted that, even with only moderate levels of complexity in software systems, it is not possible to eliminate all bugs and issues. There are always residual issues, even if they have not been discovered.

If your software verification activities have been related to your software requirements specification, it may be that your requirements are a bit light and therefore test cases could be too superficial leading to a conclusion that you have eliminated all bugs.

I suspect that the FDA inspectors may be alluding to similar thoughts.

Hope this helps to point you in the right direction.