Nice, if you have SAM at your disposal 
We ended up using a combination of warning alerts and the hrsystemuptime OID for monitoring when a server actually reboots. Warning alerts give us a pre-emptive view of nodes that might be transitioning from up to down and the server hrsystemuptime OID lets us capture the actual server restart time, not just the SNMP service restart time.
Our biggest challenge now is to merge the warning alert and last reboot alert into a single business rule. I definitely don't want an incident for warning nodes and node down or last reboot events, especially if a node goes warning, then down, then triggers last reboot. Therein lies the art behind the science -- IMO! 