It seems to me that a decision to backport any -current patch to -stable is fairly easy:
Does this fix something in -release? Or does this fix something in -current or add new functionality there?
The decision of whether or not to add a -stable fix to the errata page is probably much more of a judgment call. As I'm not a developer, I don't have insight into how that decision is made, but I am guessing each proposed errata is discussed by those responsible for the area concerned, and a consensus is formed.
I see -stable patches as they appear in the source-changes@ mailing list, but I have never counted them and compared the number to errata patches.
----
@bsdfan:
Some folks who run -current never build a kernel or userland. They just upgrade from snapshot to snapshot. The mergemaster package makes upgrades manageable -- and now with sysmerge(8), mergemaster may be dropped. Also, the availability of "snapshot packages" may mean that unless there's a synchronization issue, these lazy people don't even have to build a port.