qtree growth rate abnormal vs. qtree almost full

We have script which generates incident ticket for "qtree almost full" and "qtree full" events.  We have seen where a process that might be runaway might be better served by using the "qtree growth rate abnormal" warning event.

I am interested if anyone else uses qtree growth rate abnormal in a similar way and how much lead time in terms of detection it gives over the qtree almost full, and is it reliable.