We found we needed to turn off the heartbeats because APs kept getting failover requests from the controllers, and TAC is looking over our setup and the data we've gathered from it to see why; It doesn't look like, if there is acual packet loss, it should be enough to trigger the events. At the same time, replication of our setup at the TAC has not managed to recreate the behavior in the lab.
If you turn off the heartbeats, you still get HA failover with redundant pre-constructed tunnels from the APs, but only missing AP-to-controller heartbeats will trigger events, on a per-AP basis.
So basically it depends on whether you continue to experience these events and you can rule out actual packet loss triggering the events. If so, either turn off the inter-controller heartbeats for a stable workaround and report it to the TAC so you can be further diagnosed. If you can afford to play around, try adjusting the timer/threshold (that didn't help us, though.) If you are playing with the timers while experiencing events, you may want to turn off preemption so you don't get two events for every actual event, supposing you don't mind APs staying on the standby indefinitely.
If you are investigating packet loss issues, the heartbeats are sent outside IPSEC tunnels on UDP 8211 and are easy to sniff and a modern tshark/wireshark will be able to see their sequence numbers.