I recommend creating two smart rules for each database containing indexed items: one with “Kind is Any Document” and “Item is Indexed”, and a second one for “Kind is Any Document” and “Item is not Indexed”.
This way, you automatically know if you have non-indexed items because some operations in indexed items could create non-indexed items, like summary creation, extraction of annotations if you do them out of an indexed folder containing indexed folders. It is not the first time I’ve lost documents (*) because of this reason when deleting or messing with databases.
(*) Well, not really lost because I have copy of all, but risk to lose them.