I'm targeting specific codes based on comparing utilization patterns against Medicare's national averages for my physicians' specialties. For example, if they're billing a lot more 99214's than what's typical for family practice, those are the only ones I pull. My focus is on teaching them to document better; versus grading their current documentation skills.
When I find things that are wrong or could be improved, I provide detailed rationale to support my comments (beyond just re-stating the rule). I try to provide insight on the purpose behind each rule, and give examples of how they could have re-worded the phrase (or what they needed to add/leave out) to make it better; I also identify every phrase that was eligible for credit, and what credit would be applicable, even if the phrases didn't impact the code selection. By doing so, I give them examples of things that meet guideline-criteria, from their own words, for their future reference.
I have my audit tool set up with all of the HPI, ROS, and PFSH elements, etc. listed, and I copy and paste the statements that get credit, next to the item credited. If there are 10 potential HPI elements, then I list them all where they could fit. For the exam, since we only use 1997 guidelines, I've got all of the multi-system bullets listed in light gray, and in plain font; for every bullet that is earned, I change the font-color to black, and made them bold. It's fast, and it shows how I got my answer, as well as what bullets they missed.
Then I have a comments section, where I detail places that fell short of credit, or any other issues that I noticed. My goal isn't to be overly-picky; I'm just going off of how I learn things - when I do something wrong, I have to understand why it was wrong, what would have made it right, and why it matters, in order to commit it to memory and prevent repeating the mistake. Hope that helps!
