[rules-users] Best (better) practice question

I have an application that classifies glucose levels in a patient, looks at the insulin dose, and provides advice. Works reasonably well (using 4.07 and local drl file). However, my "decision object" has a text field in it where the advice and explanation is put, by the relevant rules. This means that the RHS always modifies the decision object, so no-loop does not prevent looping. For this reason, I have created a "state object" that has a bunch of Booleans in it, and the RHS sets a flag as well as modifying the decision object. This works. My question is whether feeding a state object with a bunch of Booleans is really the right thing to do, or should I be asserting new facts (one fact object for each current Boolean). I could have a "helper object" with inner classes that represent these facts. It feels like I ought be be inserting and retracting facts rather than the approach I am using, but not sure it matters. Thank you. - Mike