Sorry for the delay in replying, been backlogged.
I'm not sure I understand the question. A hazard analysis in some form is expected irrespective of the level of concern. Of course, with increasing level of concern, there's the implied increased level of detail.
Software controls quite often are used as mitigations. And your testing needs to show that the mitigation (software control) operates as designed and is effective in mitigating the risk. Putting the device in a "safe state" is one method but is not necessarily a cure-all. For example, with an infusion pump (major level of concern, of course), stopping the pump ("safe state") would not be acceptable for all error conditions. This would potentially cause a delay in treatment.
So you always need to re-assess the risk AFTER controls to determine if there are any risks added as a result of the controls.
Hopefully this "shotgun" answer gave you some useful information. If not, maybe give some more specifics about the device and the controls and I'll gladly try again.
