In usability testing (at least summative testing) we do not design the study to force use errors - we focus on the risk based tasks.
Yes, I agree that in summative testing you do not "force"use error, but you may be missing my point -you are still looking for them. Only if your HFE/usability process was done correctly, you won't find them. It is still the same problem as I mentioned, before the summative testing, you do not have a "guarantee" that your device is (mostly) error-free.
An analogy that I use might help. If you are not sure of error and put people in a plane to travel, the outputs might be: the plane crases and everyone dies, then you can confirm that the errors exist, or the plane might land safely, then you can confirm that it could have taken off.
Not a great way to confirm something. It's more or less the same thing as using a clinical evaluation to confirm usability (as I mentioned, it obviously can be used, as well as post-market data fro real use, as a way to confirm).
I do not see a reason to exclude usability data.
I did not say that you should not use usability data either from the clinical evaluation or any other source. My problem is to link this directly as being part of the clinical evaluation. It would be better. in my point of view, to require that this data is fed back into the usability engineering process. The way the
MEDDEV says it, it seems that the usability engineering process is part of the clinical evaluation. Worse, the MEDDEV clearly says that the clinical evaluation includes "confirmation of usability, that the design adequately reduces the risk of use error as far as possible, and that the design is adequate for the intended users". That's what I do not agree with, that the clinical evaluation confirma usability. This makes no sense.