Latest / Off the Chart: A Business of Medicine Podcast / S1 Ep76: What AI can (and can't) do for prior authorizations, with Brad Boyd of BDO USA
Transcript
- Speaker 1 00:00:00 There's a lot of bets that are made on AI to alleviate some of these burdens. revenue, the whole of revenue cycle areas is is ripe, for opportunity. Speaker 2 00:00:21 Welcome to Off the Chart, a business and medicine podcast featuring lively and informative conversations with health care experts, opinion leaders and practicing physicians about the challenges facing doctors and medical practices. My name is Austin Littrell. I'm the assistant editor over at Medical Economics, and I'd like to thank you for joining us today. This week, we're looking at one of the biggest headaches for physicians: prior authorizations. Designed as a cost control tool, the process has grown into a major source of red tape, eating up hours of staff time, delaying care, and frustrating both doctors and patients. Medical Economics managing editor Todd Shryock sat down with Brad Boyd, principal of management consulting at BDO USA, to talk about whether AI and other technologies can finally make prior auth less painful, but it explains how new tools are already pulling data from EHRs, flagging potential denials and helping practices skip unnecessary steps altogether. Speaker 2 00:01:09 Thank you again to Brad for joining us. And now let's get right into the episode. Speaker 3 00:01:22 Joining us today is Brad Boyd, a principal of consulting firm BDO, to discuss prior authorizations and how I might be able to help streamline the process. Brad, thanks for joining me. Speaker 1 00:01:34 My pleasure. Thanks for having me. Speaker 3 00:01:36 So, Brad, currently, from your perspective, what are the biggest pain points in the current prior authorization process for physicians and their staff? Speaker 1 00:01:45 Okay, so coming out of Covid, there's been several just headwinds that have that have challenged, financial, you know, provider organizations and, labor shortages has been one very, impactful. And, unfortunately, it's an issue that's here to stay. So when you look at things like prior authorizations and just most revenue cycle functions, overall the really administrative functions, they're necessary because providers need to get paid for all the clinical work that they do. It's not adding value. It's not why they went to medical school, but it's an unnecessary part of the health care system. Speaker 1 00:02:25 And making sure providers are getting are getting paid. There's also been several kind of over the last year, year and a half. There's just been some some trends. I mentioned staffing costs, mentioned some just challenges with administrative work, but there's also been a significant increase in denials from payers across the board. one of those the sources of those denials relate to prior authorizations. So the volume of denials has gone on. There's also been a shift as more Medicare lives have shifted to Medicare Advantage plans. Those Medicare Advantage plans have much more, have greater requirements regarding prior authorizations. And so the the administrative burden, as the number of covered lives under Medicare Advantage plans has increased, the administrative work of all the prior authorization requirements over, you know, larger population and then denials that result after whether it's all the prior auth work has been completed or not. The increase in denials has just created an extra pain points for provider organizations. So labor costs have gone up. Now there's additional administrative costs, whether it's prior authorizations or denials. Speaker 1 00:03:53 Sales and at the same time reimbursements being cut. So we have higher operating costs, more administrative work regarding claims to get paid, and higher, you know, denial volumes. It's just led to pressures on provider organizations that, these, you know, provider organizations have always operated on razor thin margins when they're now they have higher operating costs, more administrative work. Regarding prior authorizations and appealing of denials, it's really left, the industry in a, in a uncomfortable position where it really is getting to the point where, razor thin margins to begin with are, are are not really sustainable. And unfortunately, the administrative work here is, it's really not adding value. It's all it's all really paperwork, unfortunately. Speaker 3 00:04:47 So where do you see I having the most immediate and practical impact on this prior authorization process. Speaker 1 00:04:54 So I think the overall revenue cycle area, prior authorizations, denial management are ripe for automation, including artificial intelligence, because we want to take our most valuable resources, our our clinical clinical staff. And a lot of prior authorizations start with with clinical staff, and nursing and care managers are very involved in that. Speaker 1 00:05:16 So the extent that automation we that we can shift some of these administrative tasks to, to an automated solution, it frees up those resources to do more patient facing work, whether it's patient care, more coordination of care. So it'll it'll help, you know, moving that to a digital or automated experience will really help, reduce the overall either reduce labor costs or the volume of resources that are needed, or take the resources that are already employed and shifting them to to more higher value added administrative work. So there's a lot of hope that the industry is placing on automation and AI. And I think prior authorizations is is ripe. and this was actually in, you know, automation in this areas was was being focused on well before the pandemic started. But now that I think as artificial intelligence has evolved, there's a greater, there's greater solutions out there to have a meaningful and sustainable impact, on reducing the, the cost of the administrative burden, with prior authorizations. Speaker 3 00:06:26 Would the AI be more of a kind of simple rules based, system, or do you see the AI getting, you know, a place at the table where it's maybe having a little more leeway and how it's, operating. Speaker 1 00:06:42 So a great question. And I use the, you know, I reference automation and I, and I separately because they are very they're related but they are different. So. So rules based logic. RPA now there's a genetic kind of RPA. That's really where your rules based kind of approach goes. And that's been there's been a lot. RPA has done has made a real impact in revenue cycle management. I think the holy grail really comes around a genetic AI and looking at behaviors, looking at patterns of proactively looking at patterns of where claims have been denied for, say, different areas of prior authorization, and knowing in advance what a claim might, you know that that the this claim will most likely be denied based on past performance? And what are the steps necessary to appeal that? And doing some of this work kind of behind the scenes in an automated fashion, in some cases, AI from a, you know, the true agent, a genetic AI, can look at claims as physicians or nurses are doing documentation and doing the documentation and coding of the claim and behind the scenes, they can really say, okay, based on our recent patterns. Speaker 1 00:07:58 These are areas where that claim can be denied, and that could either provide alerts to the clinician or better yet, it can start to organize some of the documentation. So in that, you know, when payers allow documentation to be submitted with initial claim, it can say, we're going to submit this information with the claim, because we know based on history that this is a high likelihood that it's going to be denied for these reasons. So we're going to attach those parts of the clinical note with that claim up front, where it's where it's possible just to just to to mitigate the the delay in that cash and all the, the work that's involved, you know, involved in in in working that denial. So I think genetic AI is very different from, from rules based. It really is more of the when we when we talk about AI today, I think a lot of people might not know the term genetic, but that's really where where we're looking for where, where the computer is working at a higher level of intelligence. Speaker 1 00:08:58 based on that past history, much more advanced than than traditional rule based logic. Speaker 4 00:09:07 Say, Keith, this is all well and good, but what if someone is looking for more clinical information? Well. Speaker 5 00:09:13 Then they want to check out our sister site, Patient Care Online. The leading clinical resource for primary care physicians. Again, that's patient care online. Com. Speaker 3 00:09:26 Do we have the data inputs, that exist today in in with payers and practices to make all this work. Is it going to have to be additional data input into the system, or do we already have what we need to work with? Speaker 1 00:09:42 Another good question. I think we have the the data exists from practice management systems or revenue cycle systems, electronic health records. However, based on our work, what we see with organizations is there's not a, a a sound understanding of the prerequisites for an AI solution to work. The data might be there, but it might not be stored in a warehouse or a a organized fashion with sound data governance that then again, I can work off of, in a in a more automated and streamlined fashion. Speaker 1 00:10:18 So the data might exist, but it might not be stored in the right process. With standard data sets. There requires some like data management and governance around that data. That's a big step that is often not understood for for a true AI solution to work. So the data is in your systems, but it typically needs to be stored or staged in, say, a warehouse or repository with sound control and governance around it, so that then the AI solutions can really can really, work in a, in a C, in a, in a streamlined fashion. Speaker 3 00:10:54 Well, these systems be on the provider side. Or could it possibly be on the payer side where you're putting in data and it's already flagging it for you? I guess you know, which side of the fence will these systems be on? Speaker 1 00:11:08 Well, most of our work is on the provider side, so I think there's a little bit of a battle going on in, in AI, between the providers and payers. The data exists. And there's also there's also a lot of tools and automation regarding prior authorizations that technically exist, but unfortunately the incentives aren't really aligned between providers and payers. Speaker 1 00:11:30 So there are there are solutions out there that can automate many of the the prior authorization tasks, requesting a procedure or an imaging test. All the informations in the chart based on the payer rules, that that prior authorization request could go to the payer that payer could see. Must could get access to the information from electronic health record. If both parties had the necessary, security set up and access to set up, so that could automate that, that that prior authorization request from days to seconds. However, the reality is the providers don't necessarily trust the payers to look at that chart in a independent fashion. What information might they look at that might lead them to deny that claim? payers aren't really incentivized or payers aren't really incentivized to to automate that prior authorization, because the longer it takes for that authorization to be approved, the cash is in their bank earning interests. So it really is a lack of those lack of alignments in incentives. so payers will use AI tools to scrutinize claims before they're paid or to scrutinize, prior authorization requests. Speaker 1 00:12:51 and providers are just really trying to figure out what's the most efficient way to convert their clinical activity to cash. It's a real unfortunate lack of a lack of incentives, because there are tools that exist today that if both parties wanted to, to work very collaboratively. But unfortunately, it's the incentive piece there. what what what works best for the providers means payers are going to pay claims faster. And that's where that that breakdown of incentives lies. Speaker 3 00:13:23 It makes me wonder, is there any danger of a world where AI is is sending in the the claim the payer AI is rejecting it? The provider AI is, appealing it. You know, where it's like two AI systems arguing with one another. Speaker 1 00:13:41 Unfortunately, I think that's already happening, whether it's, you know, I or some type of rules based logic. there's been a significant increase in, in, in denials over the last year, year and a half. Some industry, you know, reports about 20% increase in denials. That's not because providers are are not doing their work, or that they're trying to, you know, provide services that might not be medically necessary. Speaker 1 00:14:08 There's always going to be, you know, there needs to be checks and governance around compliance. But to I don't think it's realistic that there's been a significant increase in poor behavior on the provider side where they're providing services that are not medically necessary. so I already think they are battling and it's a lot of gamesmanship. And, and for these businesses that are still struggling financially coming out of Covid. Higher operating costs, lower reimbursement. Now you add this administrative burden on prior authorizations. there it's really something that is it's it's really impactful on provider organizations. And I think, the when you're, when you're operating on such thin margins to begin with, when you delay cash, or, or the cost of collecting that, that claim increases dramatically. It it's a, it's an environment that's not sustainable. And so hopefully I and the provider side can help address some of it. and ultimately though there I think there just need just to be some better alignment in the incentives between purveyor and providers because, we're all consumers of that medical services, even though we're professional, you know, administrators on that, that work with providers and that work with payers. Speaker 1 00:15:36 At the end of the day, we want, we want our, our providers to, to spend their time taking care of us and our family members. And, it's it's something that's, there's there's a lack of balance. that's, that's occurred over the last couple of, you know, a year, year and a half. And I know from the provider side, I is, for many areas in the whole continuum of patient care, I is there's a lot of, bets that are made on AI to alleviate some of these burdens. revenue, the whole revenue cycle area is, is is ripe, for opportunity. Speaker 3 00:16:14 Well, these AI tools for prior auth be integrated into the existing EHR workflow. Or is it going to have to be some additional step for a physician to avoid a prior auth? Speaker 1 00:16:28 I think it absolutely needs to be embedded into the clinical workflow. The administrative workflow, the patient access side. And when scheduling is, you know, when schedules are, are booked and the an extension of the eligibility verification process to understand prior auth, it really does need to be embedded. Speaker 1 00:16:48 we, we, we work with a lot with organizations that are taking different approaches to figuring out what are the right tools. whether it's going to be your EHR or your revenue cycle system vendors, they're working on some functionality that that includes a genetic I there's a lot of third party vendors that have, you know, might address a couple of different use cases. and they have AI solutions that has to be integrated into the, into the overall kind of end to end, clinical workflow. as well as, you know, organizations are looking at maybe using some third party tools immediately to to do a lot of learning regarding AI and get governance programs and and prove the concept to get some short term wins and demonstrate that value. But longer term, I think they're looking for developing more proprietary AI solutions. We. We conduct an annual survey of of healthcare CFOs. And this year, about over 90% of the CFOs reported that their longer term strategy would be some type of proprietary AI, platforms that they would develop based on their own data sets and their own, you know, electronic health records and revenue cycle systems. Speaker 1 00:18:06 they view that will be, you know, they know their business, their patients, they, their payers, their providers, the best. but they, they view, a lot of the pre-work that's necessary to do that. They don't want the, the pursuit of perfection to get in the way of meaningful progress. So they're starting to invest with a balance of some third party solutions or using some of the tools that their, their, their EHR or revenue cycle vendors are providing while they are learning about this and that learning is going to go towards to, to, you know, to development of some internal proprietary programs. I think the the type of organization you are really depends on your approach regarding, you know, buy, build versus buy. If you're a big academic medical center, you have all this research, you have so much data you're and a little bit more of resources. You're probably able to, to develop and maintain, you know, a proprietary AI platform a little bit better than, say, a, a smaller practice group practice that doesn't have all those resources. Speaker 1 00:19:07 You're more dependent on your some of your existing vendors or a third party, solution. So those are, you know, very common conversations that we get engaged in, daily across the country as organizations. Are we we they're, they're planning, around AI and how and what are the the best use cases where it can start to have an impact on revenue cycles. Logical. But they want to do so very strategically, to make sure that they're going to learn and demonstrate value and build and go and grow is like is how we call it to to learn and then apply those learnings to it to expand your your AI program. Speaker 3 00:19:46 What about the small independent practice or the rural practice? Are they going to get left behind or will they be able to take advantage of these tools? Speaker 1 00:19:55 Well, they certainly they certainly don't have the the resources from an investment that say a larger health system or academic medical center would have. I think there'll be a little bit more, dependent on their, their existing EHR rev cycle vendors to develop and, and extend those tools to them. Speaker 1 00:20:15 so I think that's number one is, is, just the reality is that they're going to they'll be a little bit more dependent on their existing vendors as opposed to, say, homegrown development. we also see a lot of, clinical alignment work happening where, say, a rural health centers or community practices, have becoming, not necessarily employed or owned by larger systems, but there's been a lot of, clinical alignment activity and leveraging, say, the stark laws where those bigger systems or medical centers can extend their EHR systems to those community providers. And if they're involved in those types of relationships, they those practices or, say, a rural health facility, then they could, you know, if they're going to be going on their clinical partners electronic health record system, they often benefit by using the ecosystem of IT tools, which could include AI. But but if you really are independent on on your own, you're, you're somewhat, limited because you don't have the, the the the the capital to to make, you know, investments like a world class academic medical center. Speaker 1 00:21:29 it could be something that, it would be unfortunate if, if they're left out because they, they, they still had the same issues of getting a bill clean bills out the door and getting paid timely. And that could really grow some of the the gap in financial performance for some of those, rural or independent community providers out there. Speaker 5 00:21:56 Hey there folks, my name is Keith Reynolds. I'm the editor of Physicians Practice. And for this P two Management minute, let's wrap about three red flag behaviors that mean it's time to let an employee go. Number one repeated compliance violations. Snooping in EHR sharing passwords. Any privacy breach invite six figure HIPAA fines. Lost hospital contracts. If someone ignores written policies after counseling, your duty to protect patient data outweighs the second chance. Number two toxic disruptive behavior, gossip, chronic negativity, outright defiance. Drain productivity and chase good staff away. Patients feel the tension to document every incident and coaching attempt if the attitude persists. Swift termination protects your culture. Number three dishonesty or suspected theft. Speaker 5 00:22:40 Missing deposits or altered ledgers. Signal embezzlement practices have lost hundreds of thousands to trusted insiders. Theft warrants immediate dismissal and usually a call to your attorney, insurer, and possibly law enforcement. Protect patients, team morale and your bottom line by acting on these warning signs. For more bite sized practice tips, make sure you visit Physicians Practice. Thanks for watching, and I'll see you tomorrow on the Peta Management Minute. Speaker 3 00:23:09 Do you feel that we're at a tipping point with with prior us in general? I mean, there's been there was a big announcement this week about for the second time that they're going to finally address prior auth. It's getting a lot more attention. you know, from regulators, there's been a lot of public uproar about it. And do you feel like we're at a tipping point and technology might help finally solve this problem that no one seems to to like? Speaker 1 00:23:36 I hope so. I hope so, I think. I think the, the financial stress of the health care provider world, we feel it's a little bit more sensitive now than it was coming out of Covid when there was some government funding that was coming out. Speaker 1 00:23:52 or in other periods of time, 2008, 2009, with interest rates going, high, I think the higher operating costs of these organizations at this point, you add on the complexity of, of or the the additional costs of getting paid. It's not sustainable. We do. Now, there are technology tools out there. there's some of these tools around. Prior authorization. Some of them existed before. Before the pandemic that hadn't been adopted. And some of the reasons for that, a lack of of broader adoption deals with either issues of trust between providers and payers, which are reasonable or or lack of incentives. I do think we're at a tipping point where some programs, whether it's, you know, gold card programs or on prior authorizations for that, those types of programs, some of the regulations that have been announced or proposed regulations this week, around, you know, possibly, eliminating some prior authorization requirements. there has to be a better mousetrap. It's and and it's not a matter of putting a third, you know, a person on the moon. Speaker 1 00:25:05 It's getting. It's there. I unfortunately, I think regulators and Washington, D.C. can play a role in pushing this through. which is hard for me to say. because I don't think that people that don't understand health care should be making policy decisions about health care. But this is an area where, the volume of denials is out of control. And the impact that's having on provider organizations who treat all of us and our family members is is significant. Speaker 3 00:25:40 Are there ethics issues involved with getting AI into the prior auth process? I mean, we're talking about medical necessity, very personal decisions and, you know, life threatening decisions in many cases. What are your thoughts on the ethics of all this? Speaker 1 00:25:58 I think there are a lot of ethical issues and real concerns with, with, with. I think for one, within clinical care and clinical decision making, I'll push that aside for a moment. But the responsible use of AI, I think everyone One that we work with, I think recognize the importance of of, you know, using AI responsibly. Speaker 1 00:26:23 governance plays a significant role. And we do see organizations, that are aware of not just the governance requirements, but the ethical use of AI. providers are you go to medical school to take care of patients, so they want to do right. There are lots of risks when it does come to where data is stored and how that data can be protected, because AI tools typically, while they can pull information and interact with electronic health record, they're often pulling information, say, from a data warehouse or a third party tool. Cybersecurity is another major, major, challenge in healthcare that is widely understood as an issue. So how that data can be stored. How it can be accessed, read, write very in a manner which doesn't expose, protected health information to bad actors, to making sure that it's used appropriately. You're not sharing information with, say, a payer that might be used for other purposes. that all plays a very, very important role. I think it's recognized as an issue and it's more compared to, say, two years ago. Speaker 1 00:27:45 I think there's such an awareness around protection of information and also responsible use of AI, that it's not as much of a barrier to moving forward with these programs today as it could have been a year and a half ago, two years ago. Speaker 3 00:28:01 If a healthcare organization is looking to streamline their prior auth process. Where would you advise these start? Speaker 1 00:28:10 Number one is it starts with having a strategy for I. I think, you know, there's areas of revenue cycle that are already being approached, whether it's automation advanced like RPA or true AI. It starts with having a strategy. Organizations shouldn't look at a tool or a product just to address prior authorizations. What are your longer term objectives? Build a program and a strategy around a collection of use cases that you can understand. How do we prioritize where we want to make those investments? Start off. Demonstrate success. Demonstrate value. Learn. Have the proper governance in place. Monitor performance. Set our targets and how we're measuring our benchmarks, and then use that to work to expand the program to address additional use cases. Speaker 1 00:29:02 If some organization jumped to just address prior authorization, it's a logical starting point. There's a there's a lot of pain points there, but don't start just with with the end game of just addressing prior authorizations. Because if you're going to pick a vendor to to to be your partner, you don't just want to have one vendor that does one use case and another vendor that does another use case, because the end of the day, you'd probably have a dozen different vendors that all have to interface with your your data warehouses, your electronic health records. That becomes very expensive. And there's a lot of risk around data breach. So have the longer term strategy prioritize your use cases, then pick your vendors. Are you going to be using your EHR or your revenue cycle system vendors? One of them might be the appropriate third party vendor that can support some of your your your your prioritized use cases. Then broader speaking, what might you want to focus on a proprietary solutions if this is done regarding, you know, laid out for a longer term strategy. Speaker 1 00:30:10 That is where organizations are are seeing more value and more return from these investments than just cherry picking one use case or two use cases. Funding becomes a big piece of that because you don't want to just fund one product or one solution. You do want to have executive level commitment to these programs so that they can be, you know, expand not just not just be sustainable, but expand them over time. And that requires just more thought. And, there have been some vendors in the AI space that have focused on revenue cycle areas that don't exist today. a lot of money was raised. A lot of money was spent by providers. I'll call them early adopters, but they weren't really solutions. It was it was real custom development. And I think there was so much hope that these tools could, could, could help, you know, fix real revenue cycle problems that providers took some early bets. they didn't do so with a longer term strategy in mind. And that's where I think the organizations and there was an industry publication that said, organizations that approached, AI strategy, a genetic AI, programs that had a real strategy involved, strategy laid out, had a four and a half times return on investment from those programs. Speaker 1 00:31:29 Then the organizations that were just picking a product to solve a particular use case. And so I think that's where organizations certainly have opportunities to address prior authorizations and some of the and denial management immediately. But lay out a strategy and then figure out what are the right use cases to prioritize based on the pain point, the the investment required, the ease of implementation, because you want your first investments in AI to to have an outcome and to produce value, because that's how you're going to receive more funding to expand and expand these programs. And then when you get into the clinical care you've learned, you've got governance. There's more safety around it. And that's where I think the hope is for all of us, where you can then take some of the work effort, more work effort off of our providers to ease some of their, their burnout challenges. because there's a lot of hope that I can help can help there as well. But if you don't, if you do so in a very thoughtful process, hopefully there's this going to be more longer term success. Speaker 3 00:32:38 Is there anything else that you would like to mention that we haven't talked about? Speaker 1 00:32:41 No. I think that the strategy something was was something that I really wanted to emphasize because, the laying it out in a very thoughtful process as opposed to just focusing on, on one use case denial management, prior authorizations. the the defined strategy helps you really understand what are the real requisites around data, around governance? what do we need to do regarding our core systems? Take full advantage of our the technology of electronic health records and billing systems we have in place. I want we're as consumers of healthcare services, not just professionals who who help organizations in the space. We're all counting on on AI to to to improve, our overall health care delivery system. And we want these, early initiatives to succeed. And to really be successful, I think requires the strategy, a longer term strategy. There's there's many use cases that are ripe for solutions. Prior authorizations are being won, but do it in a thoughtful manner. so that these programs can not just be a, you know, a year long project. Speaker 1 00:33:56 They want to be sustainable over the long term. Speaker 3 00:33:59 Very good. Brad, thanks for joining me. Speaker 1 00:34:01 Great. Thanks for having me. Speaker 2 00:34:09 Once again, you just heard a conversation between Medical Economics managing editor Todd Shryock and Brad Boyd, a principal of management consulting with BDO USA. My name is Austin Littrell, and on behalf of the whole medical economics and physicians practice teams, I'd like to thank you for listening to the show and ask that you please subscribe to Off the Chart wherever you get your podcasts, so you don't miss the next episode. Also, if you like the best stories that Medical Economics and Physician Practice publish delivered straight to your emails six days of the week, subscribe to our newsletters at Medical Economics and Physicians Practice. Com. Off the chart, A business and medicine podcast is executive produced by Chris Mazzolini and Keith Reynolds and produced by Austin Littrell. Medical economics, Physicians Practice and Patient Care Online are all members of the MJH Life Sciences family. Thank you.