This article is sponsored by NextGen. This article is based on a discussion with Brian Wheelan, CEO of Transformations Care Network, and Heather Martinez, Director of Specialty Solutions at NextGen Healthcare. This discussion took place on October 8th, 2024 at the BHB INVEST conference.

Heather Martinez: I’m the Director of Specialty Solutions with a focus in behavioral health and integrated care at NextGen.

Brian Wheelan: I’m the CEO of Transformations Care Network for outpatient mental health. That includes both psychiatry and psychotherapy, we have almost 1200 providers, 60 locations at the end of this year, and about 100 and 50 million revenue, probably 1.1 million visits this year in terms of claims.

Martinez: The focus of our session today is around our revenue cycle and leveraging technologies that help the revenue cycle. I see a lot of clients on a daily basis. There is a big resistance to adopting technology, especially with larger organizations. I’ve been in the revenue cycle space for over 25 years, a little bit longer. A lot of times our staff are saying, ‘only I understand this, there’s no way to automate this, right?’ And I think we need to challenge that narrative and we’ve made some progress on the clinical side with providers leveraging technology, but we’re a little bit behind on adoption for solutions that can really impact your revenue cycles.

Brian, you guys have seen some great success in adopting innovative technologies. Could you share some specific examples of how AI and automation have been incorporated into your operations to tackle growing complexities of health care management?

Wheelan: So the easy answer is in adopting some kind of industrial strength RCF, right? With NextGen as a small provider. That means going to the big leagues, which was, spending money. It’s time, it’s talent, it’s painful, right? At the end of the day, RCM is that important? But we did. So we did you guys have an internal partner in RCX Rules, not internal, but a designated partner just sent you a Rules engine. Everyone kind of thinks generative AI and the more tangible accessible part of AI right now is just robotic process automation. Is anything that you do a lot of frequently with a human that’s taken on a screen and moving maybe multiple screens means in and out of things, logs in, logs out, there’s extraordinary amount of labor content there, there’s a lot of uh mistakes there. So RCX Rules is basically trying to take all of those repetitive processes that essentially manifest in touching a claim. You’ve seen a patient, hopefully you verified that they’re eligible.

Now, all you’re trying to do is get the same 35 data points into the various permutations of a box to get it paid, which sounds easy. But you know, any one Blue Cross entity probably has 50 to 60 different ways just to get the same claim paid depending on what group or funder it is. Whether it’s a blue card or all those things you add that you start taking Medicaid, Medicare all the different payers crossing state lines, your permutations explode RCX Rules Basic. He tries, I think does a very effective job of getting around that. We went from touching 86% of claims to now touching mid teens. As we went from four or 500,000 claims to over a million claims a year. So that scaling was, was massive right now. Truth is that RCX rules will not fix bad processes. RCM is not after the point of service, it starts at intake. It is throughout the entire process. If you are not clean, end to end RCX rules can help you cover some sins, they can help you catch some common ones, but it is not a panacea and you’ll wind up trying to rule your way out and use AI to get out of a bunch of things, the candidly or self inflicted problems that started at the front desk or intake or in documentation or whatever you have to do both.

Martinez: You point out right, that I don’t think there’s a tool out there that fixes bad processes. You’re, you’re spot on there. So, operational efficiency, right? That it’s specifically in behavior, health, low reimbursement rates. Can you talk about how leveraging this sort of technology has improved the efficiency of the team and how it’s impacted your overall financial operations?

Wheelan: Everybody kind of focuses on how many claims are you touching? Even if you do or don’t touch it first pass, you get it paid. Is it getting paid accurately? Is it getting paid quickly? Right. And so on all of those dimensions, I can honestly say we are touching way fewer claims which has allowed me to either add fewer people or actually cut people. I’ll come back to that because it’s hard because the implementation requires deep subject matter knowledge from people you’re going to eliminate, which is a really interesting organizational issue.

Our denial rate actually spiked super high initially only in the positive because we were able to resubmit a bunch of claims to actually get reconsidered and ultimately get paid. So it’s initially, you look at you like why all of a sudden I got 20% denial. It’s because you resubmit a whole bunch of stuff. They either got to deny the original or the res submittal. Your denial is going to go sky high over time. Our denials have gone from kind of 11 12% down to kind of four or five, not zero.

We’ve also been able through the rules and other things to get a better sense about payment accuracy. Both understanding whether or not that claim goes out, when it comes back, whether it’s actually paying the contract and automating that and then also using kind of the tools in the, I’ll call it both NextGen and was our kind of tunnel to make sure I know where the claim always is. I guarantee you if you go into your RC, your RCM engine and really ask them, I’m creating the claims going to the clearing house. It’s getting kind of scrubbed, it’s going out the door, it’s come back. If I posted, there is leakage in every single one of those steps and a half a point here or, you know, a basis point there doesn’t sound like a lot but you wake up and that’s 5% of, of revenue right there and, and literally, that’s when things are going well, right? And so the visibility, we’ve been able to pick up on not losing stuff in the process because the minute you automate you actually increase complexity, even though you’re improving accuracy, I have more touches. Now, no human touches up between the clearing house and the scrubber and RCX rules and A I and everything. I have more places to possibly lose a claim, right? And so that surveillance is critically important coming out the back end with better metrics. But again there’s nothing easy about it. It’s not cheap, it’s not easy to implement, it works. And we are way better on the other side of it, both kind of on a net cost perspective and especially on a net revenue basis. But don’t let anybody tell you, it’s an easy implementation, it’s cheap to do or it doesn’t require a lot of expertise.

Martinez: Can you give a little bit of details there? It’s a complex equation, right? You need to understand your business. You need to understand where the breakdowns are happening in claims or you, you know, unique scenarios which your cycle, your team, they, they understand that they know because they’re in there and they’re, they’re digging through it. So basically, that’s what you’re creating is these equations of how to find things. If it’s this scenario, do this with it, hold it here. That means somebody has to, it needs to go back to uh point A and get addressed and then, oh no, this is fine. Everything’s passing, you’re having this check. It’s actually replacing a person, either running a report and downloading into a spreadsheet and sorting through it, putting a pivot table on there and finding how many errors they have and then going back and trying to fix them.

Wheelan: This is a universal truth. If you go spend real time with your RCM team one, the Byzantine knowledge that exists across the individuals. Like, oh she’s my united person. She knows what to do when you’ve got the Medicare Advantage plan for this kind of crossover payment, etc. You also learn, they’re making triage decisions every day about what they’re not going to do. They’re like, oh shit, that one’s 100 and 20 days. I can wait. This one’s about to go. I got denied if I don’t do it. And if they’re being honest with you, they will tell you what they are not getting to all the time and they look and you may not spot it because your CFO is booking a percent of charges and not really to contract. You got a whole bunch of ways in which people are hiding, which are not really getting paid. But when you go in there, they are absolutely leaving money on the table and, and knowingly doing so in order to chase the bigger deal or the faster burn or whatever it is, and this just opens up your eyes to what you’re not getting or the complexity of the processes that they are solving for in innovative ways. Just trying to make that as, as automated and repeatable as you possibly can get to work.

Martinez: So given that there’s ongoing staffing shortages, how has this impacted the team?

Wheelan: If I would say this is going to be a gross rule of thumb. But if you look at a team, you’ve got a third of people who you’d like to promote and do something else because they’re really capable. You got a third who does what they do, they do that well, and you’ve got a third, you probably should be moving on anyways, right? And you know, we’ve probably, that’s kind of what happened, right? We took a third like, hey, I’m a little behind on credentialing. I want to build a payment accuracy function. I need to do more on eligibility and because their job ceased to exist, we move them to other stuff. There is still, you know what I would say is even if your automation engine is perfect, you’re still the payers change things, right? You add a new payer, something occurs. If there is a ‘bot manager’ function that they need to exist and that’s your middle third and then the other third, you probably would have moved on if you knew you could hire somebody else anyways or if you weren’t kind of trying to hang on and you move them out.

For us, it is from a labor content perspective. It’s probably about a third of what it was getting to more. Obviously, the automation that goes with NextGen other things, it is not free. So don’t in your head score 60% savings. That’s not true, right? But when scale, right, I am very confident as I go to 2 million claims or you know, 3 million claims that we will actually add way less staff or be able to put focus on other issues, right? So what I would say is the automation AI experience with NextGen did create a can do attitude for us. So we did use a different vendor but it was the same idea. We said RCX Rules was good, it proved out it worked. We went with thoughtful AI to do two other kinds of AI exercises based on the strength of the NextGen experience. You think about it, you’ve got somebody in intake and probably get in the front office who every day goes into a Blue Cross website or a Medicaid website or whatever. And especially if you’re on the public side, it changes all the time. So they may do it five or six times during an appointment cycle. That is totally 100% able to be automated. And then the other thing and this is, NextGen whoever you use, you always have a problem. You are often collecting a copay before you create a claim, right? Because I can’t create a claim until the appointment has been completed. So I collect copay because the person showed up or they’re registering for telehealth or whatever the case is, I’ve got their 20 bucks or whatever it’s going to be. I can’t actually tie that back to the adjudicated claim until like a week later. And so this posting challenge to be able to evaluate whether or not you were paid accurately or whether or not you collected, the patient liability has to be tied at the end. I guarantee you if you go inside your companies, you have people doing that manually all day and it’s never fast enough and they can’t tie it out. And as a result statements are going out wrong or you can’t tell whether you’ve collected money makes patients mad. And so what we’ve been able to do is take on the other end to tie together a full RCM picture of patient liability and the payer side. Super eye opening. But previously massively labor intensive.

Martinez: I mean, just so innovative, very impressive. How have solutions like the RCX engine contributed to reducing errors specifically? Are you catching and speeding up the process? Are you able to address your front end processes that were causing errors? Because you don’t have somebody just quietly in the corner fixing this while this is continuing to grow.

Wheelan: You can’t run in the red zone all day, and so what I would say is because more claims are going out with less touch and we built some surveillance engines. I don’t know if it’s automation or just the NextGen environment. We do just seem to catch things when they break faster, right? The reality is we seem to know sooner because one of the most exasperating things is if somebody called up like, ‘hey, I just realized the Highmark claims haven’t been going out for 30 days, how does anyone know the claims haven’t gone out for 30 days, right?’ But it happens all the time, that’s the reality.

We seem to just catch them faster because of bot maintenance. Which is you will have to go back and reprogram your AI bot to kind of do it a little bit differently. That notion of more quickly finding when the payers make changes or something and, and often it’s, it’s not your fault. Blue Cross is a great example. Blue Cross is an employer friendly plan. They love to specialize around self-employed groups or this union or this and that right. They’ll create some special benefit and they never think at all about what it means to get a claim paid uh against that particular thing. So for that new funding group, the NP goes in a different place or the diagnostic order is different or something, maybe it builds to your parent clinic and everything else is built on that. There’s something stupid and you always have to back up how it happens. We’re able to spot that faster, get it reprogrammed to get the claims in and manage heterogeneity in a way that previously would have been either you’re just counting on somebody who makes 18 bucks an hour to be industrious, or you’re counting on having more people who make $18 an hour. Or often you just swallow the denial, right. That’s the reality. And you’re kind of happy with the net revenue. That’s 95% of the contract, right? And then most people declare that a victory. We went from 94 to 90 you know, 95 and we’re, I think we’ll probably budget this coming year to 97.5. And everybody thinks that’s ok. But, you know, that’s actually pretty good by almost any benchmark.

Martinez: Do you know how many rules that you guys have? If you had to guess.

Wheelan: That’s a great question. I don’t know exactly. I bet it’s at least 1000 now, and it grows rapidly. There is now a Scrum weekly about where to create rules and go through. There’s been kind of a management cadence created. What are we finding? Is this our error? Is it a pay or change? Is it a rule or do we have to go further upstream? I will tell you our CFO is a hawk about not overly relying on rules to fix fundamental problems. So she’s kind of gone back and, and she’s like, so help me if you refer to RCM as a freestanding unit again, I’m going to strangle somebody. It’s like this is, this is a team sport. It is end to end and it’s quite, you know, is operations everywhere and there’s a tendency to let it go to the back of the house to get picked and, and kind of cleaned later as opposed to dealing with it up front and training against it. You have labor issues in both places and you’re going through multiple systems to get there.

It’s faster and more dynamic. You’re relying on the programmers upfront to create a data governance structure. That’s always going to be the case. It works but it breaks faster. So the AI piece it, you know, basically the way I always think about it is if you had somebody who would log into a screen and make a decision, they’re like, oh, you know, the Blue Cross is that it’s got to go in this cell, not that cell, right? And those things do change. It’s silly between what you thought was the same funding group or whatever. Basically the bot will go and spot that and it’ll make that change actively based on what it’s seeing on the screen. It is able to log in as if it were a person to a, you know, a, a blue cross website, a united website, net, new website, state Medicaid site, whatever.

I think the thing to think about and this is sort of different, we’re transitioning with technology. I am a person that has to explain something to a developer, meaning I need to take my knowledge to another human and I need to transfer that and I have to do it continually. When you think about it, you’re sort of replacing that need. I can still configure, I can communicate, I can set up something, I don’t need that developer, right? I can in my own language say how I want something to be handled. If it comes in like this, I want it to reject or this is OK or this is a path. So I think the reality of these technologies, they get better and better and better. We’re sort of replacing that, right? We’re replacing that need to explain it to somebody more technical who could then write code to find it to do this.

So 85 we went from touching, you know, kind of mid eighties claims to now touching mid teens of claims that does mean I still have exceptions. But the person who is actually responding to the exception now makes 18 bucks an hour, not, you know, whatever, 2022 whatever, but it’s not a programmer. It’s not a scrum. It’s not a group of folks kind of up the food chain. It’s somebody who’s still a claims specialist. They understand RCM. An exception came out. It’s probably something fairly stupid. There’s a lot of different things they’re capable of fixing just the way they used to, they just have less to fix, right? And then you’ve given them a tool if that is something more routine change. Now they update the rule. The bot is now a little smarter and you’re going to have to not touch it the next time. And so it’s just, it’s, it’s almost like working at the top of a license, right? Or whatever the case is.

Martinez: Looking forward, in what areas can we improve this process even more?

Wheelan: Part of the reason we’re talking about this automation is because most of the pairs we work with have 15 different systems from the seventies with real limitations around how they’re able to do what they say they’re going to do. I would not be crazy optimistic here, what I am realistic about is mental health. My margins are tight and my payroll is big, so if you think about it, we’re about a $100 and $50 million company. It takes me almost 1200 clinicians to do that. If I were an orthodontist or an orthopedist, that’d be like 30 people, right? And they would make $3 million a year and work three days a week. I’ve got a whole bunch of people working really hard and I’ve got a ton of SGN to cover that? To get the eligibility to get the claim right, to actually make sure that it reconciles the right way in some ways, this is what allows us to stay in business, right? Without this ability to get those costs out and be a little bit more efficient and have operating leverage, you might ask the question whether or not a scaled outpatient platform can make it right because there is this issue like if you look at our space, life stance is getting better, but they don’t really make a lot of money. I got some locals and all these other guys kind of doing it right. So there’s an open question about whether folks who are becoming better operators are sustainable. So when I look at this, I do feel more optimistic that there are ways to take the cost out and actually ensure that the money is going to my clinicians and the support I need in the clinical workforce is something that you can do in the long run.

Martinez: I would say for this technology in the future, in order to remain financially viable, you have to adopt it, you have to find a way to make a plan of how you can leverage it. I think getting ahead and understanding how you can leverage different automation bots, whatever is out there on the market, just specifically in your revenue cycle is so important. Your clinicians are going to demand it pretty soon. No one will want to type out a note anymore. They’re going to want to record their conversation and do all of that.

Wheelan: So it’s always part of the challenge, because the EHR just essentially generates a recording. I don’t mean recording like a person but it’s like a recorded appointment. Somebody has now put their license on the line to say I saw this person and it happened. So the question then is how do I turn that into a claim that can actually go out and come back? It’s probably going to run through your clearing house, get submitted via NextGen and all the various pieces, all it does is take that recorded appointment that says 9083799214, whatever the case is with the right modifier and begin the process.

I would say make sure you invest in surveillance because that’s another place to lose a claim. Which is you should be asking whether all my recorded claims make it to a submitted claim and make it on the way back. Sounds easy. Do the math. I guarantee you’re losing claims. It’s shocking how much leakage there actually is. The simple answer is it doesn’t have to integrate as much as you just have to create uh kind of a tunnel.

Martinez: As long as you can, you can have access to your charge data. Obviously, NextGen has clinical solutions, rev cycle solutions, but you’re not limited to saying, oh, if you’re going to use rev cycle, you have to leverage the clinical as long as you have access to your charge data, whether it’s a HL seven, you know it can be brought in, then you can A B.

Wheelan: We’ve moved 99 different acquisitions onto Valant. That is its own skill set that is distinct from what you’re doing on the back end here. There’s a whole separate question around, ‘what do you do on the front end?’ Scheduling, collecting co-pays, all that type of stuff. That’s still an EHR conversation. I mentioned the challenge of posting on the back end, which is can I tie my copay back to a paid claim? Those are issues and that you do have to figure out some way to match those up. We actually do use AI for that as well because it’s laborious to do, but I don’t see anything here that would stop. In fact, I would say I would never acquire something if I couldn’t bring it onto the common EHR and RCM structure. It just is not viable in our world, at least from my opinion, the way we run things to not have one single platform. I’ve seen form can be multiple modular things, but you need to have one way of doing business.

Martinez: I’ve seen multiple times where the biggest impact is on your cycle, you need to be able to see that. So you’re looking for more of a holistic view and if you are doing acquisitions, you want to ensure you at least have multiple EHRs as you’re bringing them on. That way you’re only really vetting ‘can I get the charge data right, and get it out and, and be able to at least have an impact here and have one centralized rep cycle.’

Wheelan: The solution sounds simple and people tell you it is, but it’s not right because the data structures aren’t the same no matter what they say. So if you actually want to have a warehouse that’s accurate, you have to figure out how to have consistency.

So the decision to migrate to NextGen was made during the budget. During that year, I would say there was kind of a dark period in August of 2023 where you’re kind of like, all right, we’re really doing this right? And you have to make sure that you’re not losing claims. There was a whole bunch of resubmission activity. What I’d say is I can give you a lot of advice around and I would say we went back. We knew what we were going to do but you have to definitely give your payers a heads up about migrating. You have to ask for the one time dispensation, which is like, hey, I may miss timely filing with you right here. I’m starting now. Give me the grace until X and we have yet to have anybody not allow us to resubmit and actually get payment out.

So it’s October 2024 and I am still now cleaning up some of the collection disruptions that happened in the second half of 2023. But it’s coming in, right? It’s coming in, but you need to have the cash along the way and you have to be good about knowing what you’re missing. I keep coming back on the surveillance piece. So CFO for us only has RCM. So they do not have intake, they do not have operations over the clinic. However, I would characterize my CFO as an operating CFO. So like totally eyes wide open to, you know, and unafraid to kind of bring intake and front office and others to the table to be like, hey, we’re not fixing this on the back end like that is never going to happen. We’re going to leave too much money on the table. So we are a matrix. It is not this notion that if I were to magically put it all under the CFO magically, everybody’s going to operate with more attention to detail, better training. Let’s turn over all that. But you know our CFO controls the kind of posting RCM kind of true cash functions, including the kind of payroll and everything. But she is unbelievably good at engaging the front line staff and the intake team around the data that’s going to screw up your collections on the patient’s liability or otherwise if you don’t get it right. I mean, this is the whole thing. That’s appropriate, you can’t live without it.

For the seventh consecutive year, NextGen Healthcare is the highest-rated EHR vendor for physician groups with 6–99 providers and the top-ranking provider of PM solutions for physician groups with 1–50+ providers. To learn more, visit: https://www.nextgen.com/.