I have a question along similar lines, close enough that I think it can fit within this same thread.
What if you have a hazard and hazardous situation that has managed to cause a harm (cut) of a higher severity than you thought would be possible - from a Severity 1 within the assessment to a severity 3.
So now there is evidence of both severities occurring, with the severity 3 occurring at a less than 1 in a million uses, but still has occurred.
As there is now evidence of occurrence, every part of me is saying that this now needs to also be a part of the hazard analysis, so we have two cut harms listed with different severities. There are unlikely to be any changes in controls as a result of this, however it may now be necessary to disclose the residual risk to users in the user manual.
But receiving questions of "do we really need to do that, for one occurrence in over 20 years"
Do we?