That works a little bit when we're talking about one satellite poofing in a year based on a collision with another satellite, and not at all when we're talking about thousands of events a year, many of which are satellite-debris collisions too small to track (you only get one orbital vector), or between pieces of debris.

Every collision generates hundreds, maybe thousands of pieces of debris, only the largest of which are trackable.