Ob Hospitalist Group CMO Mark Simon, M.D., on the Value of Autonomous Coding

Ob Hospitalist Group (OBHG) is the nation's largest provider of OB/GYNs with nearly 2,000 physicians. Mark Simon, M.D., OBHG’s chief medical offer, recently spoke with Healthcare Innovation about the potential for artificial intelligence to aid clinicians with administrative tasks such as coding. 

Greenville, S.C.-based OBHG conducted a recent pilot with the Commure Autonomous Coding solution. Within three months, OBHG clinicians reduced time spent entering charges by 83%, and AI now codes over 85% of all charges. The next step is rolling the solution out enterprise-wide. Joining the conversation was Commure Chief Operating Officer Deepika Bodapati.

Healthcare Innovation: Dr. Simon, could you describe how your organization partners with hospitals around the EHR and data capture and sharing, and then maybe how you've traditionally handled coding, billing and quality reporting, so that we can contrast how the new mode is impacting that?

Simon: We contract with hospitals to provide OB hospitalist coverage. And the vast majority of our work is done in the hospital facility itself. And, of course, from a patient safety and patient care perspective, our goal is to document the work that we do within the hospital’s EHR because it's best for the patient, the information is available for everyone to see, et cetera. And then we do our own professional fee billing for the services that we provide. For a number of years, we’ve used PatientKeeper as a charge entry system. [Commure purchased PatientKeeper from HCA Healthcare in 2021.] Our physicians and midwives would see the patient, document the note in the hospital's EHR, and then they would go into PatientKeeper and enter the charge for each billable visit.

HCI: And were there some common coding workflow frustrations expressed by the clinicians about the manual data entry?

Simon: Absolutely. Clinicians, physicians in particular, don't like coding. It has its own set of rules that need to be followed, and they may not be as intuitive to a physician as one might hope they would be. It's an extra piece of work that they have to do. They have to go into a different system and then think about the visit in a different way and enter those charges. Coding for physicians can be a very burdensome activity.

HCI: How did your group come to do a pilot of the autonomous coding solution from Commure? 

Simon: Commure came to us, and explained that they had this offering where they could start to potentially autonomously code our visits. That is something that is very appealing to us, in terms of ease of physician workflow and midwife workflow. If we could do something as a company to make the physicians’ lives a little bit easier, we wanted to investigate that and look into how this could work for us.

Bodapati: At Commure, we see ourselves as an AI partner to OBHG, asking how can we relieve pressure and create more efficiencies in existing work streams? This was such a clear place for optimization and uplift when it comes to provider time being spent, when it comes to revenue leakage, and just a clear place where automation can be introduced to relieve providers’ pressure. It made a lot of sense to deploy that technology here. 

We have a massive physician shortage, so this is very top of mind for everyone, and likely will be for decades to come. We are going to want to try to help providers find the purpose back in their work, and that is not through doing documentation and coding and going back and forth with billers on why some charge got accepted and some charges didn't get accepted. That is not why they went into the profession, but the fact that these people, who are really hard-working people are burnt out, means that there's something very wrong. They just don't want to work on this administrative nuisance. So we see our solution as a way to augment providers in that capacity, to help them code, to help them document and create a very seamless workflow to optimize revenue at the end of the day.

HCI: In pilots, did you identify that it captures more codes than the clinicians might themselves, and then they can just approve the codes that were captured?

Bodapati: We found that the amount of time that clinicians spent coding reduced by 83% and we're doing about 85% of all the coding there.

Simon: From our perspective, we absolutely hear from clinicians that it makes their lives a lot easier, and that there are certain encounters where they could have easily forgotten the code. Clinicians want to do what's right, but they're very busy, and they can get easily distracted by the clinical work that they're providing. Sometimes they'll write the note or document the visit, thinking that they want to go back and put that code in, but they forget about it. But with the AI tool, it makes it so much easier. Now the note goes through, it gets coded, and they have an opportunity to look at it and make sure it fits with their understanding of what it should be. And then it goes off to the revenue cycle team.

HCI: Does having this new AI partner require any training for the clinicians, because it is changing their workflow, even though they are doing less administrative work now?

Simon: What they need to be refreshed on is that documentation is key — how they document the note about what they did and why they did it, and for whom, and all that thought process that goes into it, which is something they should be doing anyway, right? So this isn't like it should be new, but it's just a reminder to make sure that their documentation is going to make the AI tool work the best for them and where it causes them the least headaches. So, for example, making sure that their note type is correct. If they're doing a delivery, that it's really a delivery type. That helps the AI tool accurately code this, which means they have less work on the back end. At the same time, making sure that their documentation covers all the appropriate points of what they did during that visit, so that the code can be as accurate as possible, especially when you start talking about diagnosis, making sure that their assessment includes the appropriate diagnoses that are relevant for that patient. Therefore, it will be picked up by the AI. So it's a refresh flash training on documentation. Part of my role is risk mitigation and malpractice and all that, too. So I'm a big fan of better documentation regardless.

Bodapati: This is really meant to empower providers. There are new codes. Codes change relatively often. It's not something that's static, right? So then everyone needs to get updated on these changes. And what we're finding is the case is that often doctors actually drop secondary diagnoses, and actually don't accurately portray the complexity of the procedures they're doing. If you don't say that your patient has COPD as a secondary diagnosis and hypertension as a secondary diagnosis, and there's a slight complication, that's actually a higher acuity procedure that you're not getting paid for, but you are doing all of the work of  higher acuity procedure. Physicians should be compensated in a way that's absolutely commensurate with the amount of work and the complexity of the work that they're doing, and that is not happening today.

Simon: I think one of the other reasons why AI as a tool for coding in particular is beneficial is that not only do the codes change, but sometimes the rules around how you pick the appropriate code on the CPT side, those get changed and updated, and it's not top of mind of a physician to stay abreast of exactly all the nuances of specific codes rules. I think it's a perfect use case for AI in healthcare, because the AI can keep up with that, and the physician doesn't really need to do that. It's not a good use of their time.

HCI: So Dr. Simon, are you now scaling this up across all your footprint?

Simon: Yes, we're looking to implement this solution at as many hospital partners as we can, as quickly as we can. When our IT leader announced at our national clinical leadership meeting in the fall that it is OBHG’s intent to implement autonomously aided coding at as many sites as possible, as quickly as possible, he was the hero of the entire conference, pretty much, because they were so excited that this is something that's going to make their team's lives easier.