Lessons from a Successful Medicare Program Integrity Demonstration Soon to Expand Nationwide

[PREVIEW CLIP FROM AMY CINQUEGRANI]

When we're talking about Medicare, it's always important to think about the sustainability of the Medicare Trust Fund and these are all of our taxpayer dollars at work. And we want Medicare to be around for when we need it. And so thinking and implementing ways to strengthen program integrity, so that Medicare is around while making sure that the patients can receive the care that they need is really crucial for what we do.

[J.B. WOGAN]

I’m J.B. Wogan from Mathematica and welcome back to On the Evidence, a show that examines what we know about today’s most urgent challenges and how we can make progress in addressing them. On this episode, we’re going to talk about a surprisingly successful government initiative that sought to curb wasteful spending in the Medicare program without harming access to care or the quality of care that patients receive.

Almost seven years ago, the Centers for Medicare & Medicaid Services, or CMS, launched an initiative aimed at preventing the improper use of ambulance rides, which had long been cited as a major source of waste and abuse in the Medicare program. Specifically, the initiative focused on something called repetitive, scheduled, non-emergent ambulance transport, often referred to by its acronym “resnat.” When patients use RSNAT services, they’re typically going for things like dialysis treatment, chemotherapy, and treatment of non-healing wounds. Medicare covers RSNAT services so long as the patient meets certain criteria, such as being confined to a bed. Under the initiative, companies that provided RNSAT services needed to seek prior authorization from CMS before providing the service and submitting a claim for payment.

Eight states and the District of Columbia participated in the initiative, and my colleagues at Mathematica evaluated the impacts. Because non-emergency ambulance service use is rare in the overall Medicare beneficiary population, the study focused on the Medicare beneficiaries who are most likely to use RSNAT services—those with end-stage renal disease and/or severe pressure ulcers. Those two groups account for about 85 percent of Medicare RSNAT claims, so any change in use and spending would be most detectable among those patients.

The most recent Mathematica evaluation found that, as a result of prior authorization, RSNAT costs declined by over 70 percent since the initiative was implemented. Importantly, spending less on RSNAT services resulted in a decrease in overall Medicare fee-for-service spending of more than 2 percent, equal to about $1 billion since the model was implemented. All of this occurred without any evidence of a decline in access to care or the quality of care beneficiaries received.

Late last year CMS announced that it would expand the initiative nationwide.

My guests for this episode are Amy Cinquegrani from CMS and Andrew Asher from Mathematica. Amy directs the Division of Payment Methods and Strategies at the Center for Program Integrity within CMS. Andrew is a senior fellow at Mathematica who led the evaluation of prior authorization for RSNAT services. During the interview, Amy and Andrew discuss lessons from the evaluation that could inform the next phase in the national rollout. We also touch on novel approaches using data analytics to further reduce improper health care use and payments in the future.

I hope you enjoy the conversation.

[J.B. WOGAN]

Much of what we'll be discussing today has its roots in the evaluation Mathematica conducted of the Medicare Prior Authorization Model for Repetitive, Scheduled Non-Emergent Ambulance Transport, which is a mouthful. Folks may not be aware what some of those terms mean, so I was hoping we could start by defining some of them. In laymen's terms, what do we mean when we say "Repetitive, Scheduled Non-Emergent Ambulance Transport," or, as I think sometimes we'll refer to in this conversation as -- is it "R-SNAT" -- R-S-N-A-T?

[AMY CINQUEGRANI]

So, we usually say "RSNAT," as...

[J.B. WOGAN]

RSNAT. Alright or...

[AMY CINQUEGRANI]

Yeah, we don't make it easy.

[J.B. WOGAN]

Okay. And what's it -- Yeah, if you could -- Also, what's a common example of this kind of non-emergent ambulance transport?

[AMY CINQUEGRANI]

Sure. So, these are the non-emergency ambulance rides. So, nobody is calling 911 in these circumstances that are scheduled in advance, and they occur either three or more times during a 10-day period, or at least once a week or three weeks. And the most common example of these types of trips are patients that receive dialysis treatments, but they can also be for something like wound care or cancer treatment or a number of other, you know, services that are needed.

[J.B. WOGAN]

So, in the context of healthcare, what do we mean when we say "prior authorization"?

[AMY CINQUEGRANI]

Prior authorization -- it sometimes can be called "precertification" or "prior approval," depending on the particular health plan, but generally, it's a process where a provider needs to obtain advanced approval from a patient's health plan before a specific service can be rendered to the patient in order to qualify them for, you know, payment coverage of that services. So, it's a way to make sure, basically, that a service is actually covered by the plan before that service is performed.

[J.B. WOGAN]

Okay. And I understand that one of the things that this is trying to do is curb the improper use of non-emergent ambulance transport or repetitive non-emergent ambulance transport. So, give me an example of that.

[AMY CINQUEGRANI]

Sure. So, I'm just going to back up just a tad, just for some context, because it's really important for this benefit to understand for non-emergent ambulance coverage and Medicare, it's essentially a benefit of last resort. So, if a patient can safely travel to their destination by any other means, that's what they need to do. Patients have to meet Medicare's medical necessity requirements in order to qualify for these transports, and so what that means is either that the patient is bed-confined and using another type of transportation could endanger their health or that, regardless of bed confinement, they have a medical condition that requires services during the transport -- requires the services of an EMT.

So they need something like oxygen support or tracheostomy support or there's an injury that needs to be supported. Also, the reason for the transport has to be to get to a Medicare-covered service or return from a Medicare-covered service. And Medicare only covered ambulance transport to certain destinations, such as the dialysis facility, like I mentioned before, or a hospital or a skilled nursing home. A doctor's office, for an example, for a regular doctor's appointment -- that's not a covered destination. So, all of those things need to be met in order for Medicare to pay for the service, and so the most common types of things that we see is when the patient doesn't meet those medical-necessity requirements and they could have safely used some other, you know, method of transport to get to their appointment or service.

[J.B. WOGAN]

Okay, great. And I want to understand better what -- how prior authorization works and how it would save money. What's the concept or rationale behind using prior authorization?

[AMY CINQUEGRANI]

Sure. So, all those things I mentioned before -- those various, you know, medical-necessity sort of criteria and transport destinations, et cetera, being appropriate - prior authorization helps make sure that all of those things are met before the service is rendered, before a provider, you know, submits a claim to Medicare or any other health plan, and before that claim is paid. So, we're making sure, you know, in this instance, that Medicare's only paying for medically necessary services to eligible beneficiaries who meet the coverage and payment requirements of that particular service.

[J.B. WOGAN]

Okay. And what's the backstory with using a prior authorization in this way, or this -- I guess, in this specific model? What did CMS hope would be accomplished? How did it come about? And we've already talked a little bit about how it works, but if you want to get into any more of the nitty-gritty details, that would be great.

[AMY CINQUEGRANI]

Sure. So, kind of the overall goal of the model and what we want to accomplish is to test if prior authorization could reduce expenditures for these services, these repetitive, scheduled non-emergent ambulance transports, by reducing improper payments, which we'll talk -- I can talk a little bit more about that -- what we mean by an improper payment -- while maintaining quality of care. And so, non-emergent ambulance services have sort of long been associated in Medicare and, you know, sort of beyond in the healthcare space with overutilization and improper payments. And we have a program here at CMS called the CERT program -- means the Comprehensive Error Rate Testing program, which reviews a statistically valid sample of Medicare-paid claims each year in order to estimate the amount of improper payments that Medicare pays.

So, like I mentioned before, if Medicare pays for a service that it shouldn't have -- one of those, you know, examples of an ambulance transport where the patient could have safely gone by some sort of other method -- you know, it wasn't actually medically necessary or the medical documentation didn't support how it was medically necessary. We call those improper payments. And so, that's what our CERT program does -- it reviews these claims to determine the amount of improper payments. And these findings are -- they're publicly reported each year. It's actually part of a law that we do this. But we use those findings to determine where we at CMS need to focus our program integrity efforts. And so, just to give you even more history, in the year leading up to the model, Medicare paid over $1.2 billion for non-emergent ambulance transports, with CERT estimating an overpayment rate of over 20%. So, that was a projected about $256 million in overpayments in just one year for these non-emergent ambulance transport services.

There's also a couple of sort of government oversight bodies. We have the Office of Inspector General, the Government Accountability Office, and they also publish a number of reports warning that our non-emergent ambulance benefit is highly vulnerable to abuse. So, you know, that's sort of where the idea for, you know, doing something -- we needed a little bit more focus on these ambulance rides. And so, historically, Medicare -- specifically, traditional Medicare fee-for-service Parts A and B has always used a retrospective approach to identify these improper payments, meaning that the services have already been rendered to the patients, claims have already been paid, and then Medicare needs to sort of go back in and claw back those payments. We call that the pay-and-chase method. It's useful in certain circumstances, but it really shouldn't be the only tool, you know, in our toolbox. And prior authorization, as a concept, has been used in private insurance pretty frequently. Most of us have probably had to go through prior authorization for, you know, some service that we've received, you know, in our lives, but it's really new to Medicare.

And, you know, years ago, we implemented, for the first time, a prior authorization process for certain complex power wheelchairs in Medicare, and it was tremendously successful. There was a lot of fraud, again, in that benefit, and spending really, you know, dropped significantly -- improper spending dropped significantly when the model was implemented. So we were looking for other services, and specifically, sort of non-emergent services that might lend themselves well to this concept of prior authorization. So, between the CERT findings and the OIG and the GAO reports, you know, repetitive scheduled, non-emergent transports seemed like a really good candidate. And so, again, it's just kind of another tool in our program integrity toolbox to help protect the Medicare Trust Fund and, you know, just sort of testing it out in another venue, you know, another service type -- that's how the model came together.

[J.B. WOGAN]

Okay, that's a perfect background. And, Amy, we're going to give you a break for a second and turn to Andrew. Andrew, I know you have extensive experience with program integrity. When we first talked about doing this podcast, you described this prior authorization, or preauthorization model, to me as ambitious. For listeners who aren't as familiar with this line of work, why did you characterize it as ambitious? What is unusual about this model? And give me some sense of whether it is, say, more ambitious than other efforts to reduce waste, fraud, and abuse.

[ANDREW ASHER]

Sure. Well, to be approved for national expansion, CMS requires that the model demonstrate that it reduces expenditures and does so without adversely impacting beneficiaries. And that's really unique for most program integrity initiatives. Most don't face that strict criteria. And while the goals can vary, such initiatives usually are considered successful if they reduce unnecessary use and cost and fraud in the areas that they're targeting. But for the model, so, CPI needed to demonstrate that it reduced the total cost of care, not just the cost for RSNAT services, and it also needed to establish that there is no appreciable impact on quality of care for beneficiaries or their access to care. The model of it was also ambitious because it requires CMS and their Medicare administrative contractors to reestablish and enforce its medical-necessity criteria for acceptable RSNAT ambulance transports, and that causes some providers concern. While some providers were intentionally abusing the program, many were not even aware of the criteria. And one risk that the model faced is that it would cause ambulance providers to leave the Medicare program to such an extent that it could impact the overall ambulance capacity. We found, though, that the model caused primarily RSNAT-dependent suppliers to leave the program and appeared to leave sufficient capacity to meet the needs of beneficiaries.

[J.B. WOGAN]

So, I imagine that some listeners will find that the topic of repetitive, non-emergent ambulance transports to be a pretty narrow topic, and I wanted to try to get at how this connects to larger themes within the world of waste, fraud, and abuse and trying to make sure that our healthcare system is, you know, as efficient and effective as possible, while still providing good quality care. So, are the -- I guess I'll just put it simply -- are the integrity challenges with repetitive ambulance services emblematic of a larger problem that CMS and other healthcare payers are trying to address?

[ANDREW ASHER]

Yes. Given limited resources, to a large degree, payers have to rely on providers to deliver high-quality services and do so in a manner that's compliant with their coverage benefits and their billing guidelines. And while most providers are honest and comply with these guidelines, not everybody does. The challenge payers face is how to support care delivery while also doing their best to make sure that the millions of services they pay for each year are properly paid. Prior authorization is one strategy to do this, but there are others that do not examine quite as many services and do so in different ways. Payers can use prepayment edits to outright deny concerning services, they could use prepayment analytics to identify high-risk services for further review, and they can also use analytics to identify providers for audits and investigation. They can also use analytics to identify specific services for closer review and follow up with self-audits. One approach that's particularly appealing is what I'll called "guided self-audits," where payers engage providers in their own compliance to enable them to learn more about the issues that they face and how, together, the payers and providers can work to achieve more compliance.

[J.B. WOGAN]

Okay. We've talked about the design and intention of the model. Let's talk about the impacts of this work and what Mathematica found. Andrew, from the evaluation, did you find that the model realized savings?

[ANDREW ASHER]

Yes, the model realized substantial savings. We estimated that the total cost of care for the study population declined by 2.4% and that the savings for RSNAT services, specifically, declined by 76%. To put these findings in context, let me share two things -- first, of all the models CMS tested during the past 10 years, only one other model realized greater savings than this one. Second, RSNAT costs make up only a very small share of the total costs of care for beneficiaries, about 2%. The model's impact on RSNAT use and cost was dramatic enough to impact overall costs substantially.

[J.B. WOGAN]

Okay. And beyond the savings, Andrew -- I mean, it's good that it saved money, but what did it mean for people in the healthcare system? For example, how were beneficiaries affected, in terms of the quality of care they received or even just their access to care?

[ANDREW ASHER]

Well, J.B., this is, perhaps, the most important part -- most utilization control strategies, including prior authorization, raise the possibility that beneficiaries will experience delay in receiving care or denied care that they need. In the model, quality of care and access to care were largely unaffected. Given the dramatic impact the model had on RSNAT use and cost and the importance of the destination services, such as dialysis, to beneficiaries' health, adverse impacts might have been expected. The model did not appear to cause an impact on emergency department use, hospitalizations, or death. Among beneficiaries with ESRD, we did find very small impacts on dialysis service use, including scheduled and emergency dialysis, but no impacts on hospitalizations or complications of untreated ESRD. Our focus groups in surveys did identify examples where beneficiaries were impacted and suggested that providers were concerned, but our data analysis did not find that these were widespread problems. Beneficiaries who needed scheduled transports but no longer received RSNAT services through the model may have been able to find another way to get to these appointments -- one of the other modes of transport that Amy had identified.

[J.B. WOGAN]

Amy, can I just ask -- how important was it from CMS's standpoint that not only did this save money, but that it didn't affect beneficiaries -- or didn't have a negative impact on quality of care or access to care?

[AMY CINQUEGRANI]

Yeah, that was definitely, you know, something that we're concerned about and that we were anxious to see in the results of the evaluation. But we knew that there were issues, you know, within this benefit -- you know, that's why we implemented the model. So, you know, we expected that there were, you know, a number of people using the services that didn't necessarily, you know, meet the requirements of the service. So we were, you know, certainly pleased to learn that, you know, that there weren't significant impacts in that area.

[ANDREW ASHER]

And, JB, one important consideration when gauging the impact on quality and access is the measures that you choose and whether you have a narrow range of measures or a broad range of measures, which we did, because quality and access can show up in a number of different ways. So, in our study, we included several standard and well-established outcome measures that looked at a broad range of services, but we also designed some novel measures that were trying to get at treatment delays, which are not typically measured as frequently. For example, we looked at the average number of days between scheduled dialysis visits.

[J.B. WOGAN]

Mm. Okay. Right. So that was one of the ways that you were trying to, again, go beyond just finding out whether this reduced spending, saved money, but, also, was having at least a "do no harm" kind of impact on the beneficiaries.

[ANDREW ASHER]

That's right.

[J.B. WOGAN]

So, when I posed the earlier question to Andrew about people being affected, we started by talking about beneficiaries, but another group of people in this ecosystem would be providers. And, Amy, I was wondering if you could explain the concerns or the considerations about provider burden and, you know, what steps did CMS take to try to avoid providers leaving the Medicare program.

[AMY CINQUEGRANI]

Sure. Well, we are always concerned about provider burdenance -- definitely something that we have to balance, you know, when we're implementing new program integrity initiatives, but I wanted to point out and just make sure that everyone understands that the use of prior authorization didn't create new documentation requirements, so when -- the medical necessity criteria that I was talking about before -- that exists, regardless of whether we have a prior authorization model or not. Those have always been the rules in order for Medicare to pay for non-emergent ambulance service. So, ambulance suppliers, as well -- they've always been required to maintain documentation to support their transports and support payment of those transports. However, suppliers sometimes struggle in obtaining that documentation from the ordering physician. And so, we really tried to focus our efforts kind of around education for this -- you know, creating resources, both for physicians and ambulance suppliers, to remind them of the rules and remind them of their responsibility that, if you're going to request payment for these types of services, this is what you need to have. So just, you know, creating letters, having outreach, both, you know, from CMS and from our Medicare administrative contractors -- those are things that we really, you know, focused some efforts on.

We also educated the ambulance suppliers to alert us or alert their Medicare administrative contractor of any ordering physicians who sort of refused to provide documentation of transports, and we could sort of intervene separately from the prior authorization model. And so, again, just a lot of outreach, a lot of education, but, again, we implemented the model because of the improper payments, because of the concerns and the benefit, so we expected that, you know, while, again, the vast majority of providers are trying to do their best, there are providers -- you know, those bad apples -- that, you know, we expected would not be able to work within the model because they were providing services that weren't medically necessary or providing -- you know, had fraudulent documentation or some other things. And, you know, they wouldn't be able to continue within the model.

[J.B. WOGAN]

Okay. Andrew, I know -- you've already touched on the findings about suppliers leaving, but do you want to tease that out a little bit more of what exactly did Mathematica find, in terms of departure from the model because the burden was too great?

[ANDREW ASHER]

Sure. We found that close to half of the RSNAT suppliers left the Medicare program, but it's important to mention that our survey work suggested that providers also reported burden from the model, so both of those things were part of what we identified. But it's also important to mention that the profile of the typical supplier that left was one of -- was one where they had a substantial share of their revenue was for RSNAT services and they weren't providing other ambulance services to a large degree. And when it was all said and done, it appears that the overall capacity of ambulance suppliers that remained was adequate to meet the needs of the beneficiaries.

[J.B. WOGAN]

Andrew, as you reflect on the findings for the model states -- that is, the states that participated in the model or implemented the model -- what do you see are the prospects for nationwide implementation? Do you think a national version of this prior authorization program would achieve the same results that was found in the model?

[ANDREW ASHER]

Well, our findings suggested a national prior authorization program will likely be successful. The model realized substantial savings without appreciably impacting quality of care or access to care. The savings for both the first-year states and those that began a year later were substantial. That said, the savings were lower for the year-two states, and these states more closely resemble the rest of the nation than the year one states. So while I expect that the program will realize savings, the savings might be lower. That said, in our study, we found that the savings for the year-two states were still over 2% total cost of care.

[J.B. WOGAN]

Okay. And, then, looking again to the future and implications from this evaluation, Amy, how has the evaluation informed your plans or CMS's plans as you seek to implement the model nationwide?

[AMY CINQUEGRANI]

Well, it really showed us again that providers and some other stakeholders didn't understand the rules. And so, we -- it's really our goal is to make sure that all stakeholders -- the dialysis facilities, the ordering physicians, the suppliers themselves -- understand what's required, not only, you know, as part of the prior authorization process itself, but the existing Medicare coverage rules. And so, again, making sure that, when we're ready to move into other states, we can reach out through our Medicare administrative contractors to the different provider groups in that area, developing, you know, resources specifically for the ESRD population, since these are such high users of non-emergent transports. You know, sort of doing some better targeting of those education efforts, I think, would really help, too. And we also need some changes to the model based on sort of overall feedback. The model -- Because these patients are -- again, it's scheduled, it's repetitive -- they're going to these treatments or these services regularly, such as dialysis -- they don't have to request prior authorization for every single dialysis trip. That would be completely burdensome and unnecessary. So we have an approval period, basically, for prior authorization, where we say we'll approve a certain number of transports over a certain period of time. And they sort of have that bucket to work with. And then, once that's exhausted, they'll request prior authorization again. So we learned that there is, you know, a certain population where their condition's not likely going to change -- they're going to continue to need these services. And so, we extended that period -- you know, the number of transports that they can receive, and we extended the period of time for which we can -- they can receive those transports, just to, again, lessen the burden on those suppliers that are really trying to do the right thing for their patients.

[J.B. WOGAN]

Okay. That's interesting -- the tweaks that you've made based on lessons from the evaluation ahead of national rollout. This question's really for both of you -- based on the rollout of the prior authorization model for, I guess, in this specific context, for non-emergent ambulance transport, what are some of the important takeaways for payers -- for CMS, but, maybe, also, for other entities, even in the commercial sector -- that they might want to think about as they're considering prior authorization in general and other strategies for reducing unnecessary use and cost?

[ANDREW ASHER]

So, prior authorization can be formulated and implemented to realize savings and reduce unnecessary use without negatively impacting beneficiaries for services with exceptionally high rates of improper payment. I think that's a key takeaway from the experience of the model. And CMS's approach is to target specific high-risk services -- individual services, such as RSNAT. And while that was highly effective, other payers may choose to target in different ways. They could target especially high-risk providers, providers that repeatedly seek payments for unnecessary services. There's other ways to target. The question is whether and how to do it, given the cost and expected improvement in savings. Really, with limited staff and budgets and provider concerns about burden, targeting is especially valuable, and there are a number of ways to do it, as we talked about. One that the AMA recommends is to "gold card" providers -- essentially, determine those that present a low risk of submitting a non-compliant claim.

[J.B. WOGAN]

And AMA would be -- is the American Medical Association?

[ANDREW ASHER]

Yes. Yes.

[J.B. WOGAN]

Okay.

[ANDREW ASHER]

And exempt them from prior authorization. Another strategy is to determine the risk level for individual providers and use that to guide the prior authorization strategy, to lower the rate at which low-risk providers are expected to obtain prior approval and increase the rate for high-risk providers. Using this approach, you have a strategy that could be considered fair and more efficient and can also be used to learn from the experience, to identify the overall rate of non-compliance and to look at the drivers of non-compliance. It also would enable payers to improve their targeting and to use analytics, machine-learning models, and other strategies to improve the targeting efforts and increase the return on investment for prior authorization.

[J.B. WOGAN]

Amy, what about you? What do you think ought to be the lessons, the major takeaways for reducing unnecessary use and cost or fraud at large?

[AMY CINQUEGRANI]

Yeah, I agree with much of what Andrew said, and, you know, those are things that we are -- you know, we try to figure out in the Center for Program Integrity, you know, every day -- you know, different ways to -- or ways to sort of best target providers, services, you know, what have you -- that are at risk for improper payments, you know, using the best use of our resources and lessening provider burden to the greatest extent possible. And the other thing I just -- it'll mention, and I'm going to sound like a broken record here, is just about the outreach and education and how, especially when you're beginning a program or when you're, you know, you're starting that to make sure that the folks that are going to be subject to that program really understand the rules, they understand what they have to do, they understand the benefit, they understand what pieces of paper or, you know, electronic documents need submitted, making sure we have really clear instructions where those things go. You know, at the beginning of programs, sort of despite our best efforts, we always get a lot of e-mail, a lot of phone calls with folks that are just trying to learn the rules, and so making sure that we at CMS partner with our Medicare administrative contractors the best we can, making sure that they're staffed and they're able to handle those -- you know, any sort of new initiative is really important for things to move smoothly.

[J.B. WOGAN]

Thanks to my guests, Amy Cinquegrani and Andrew Asher. In this episode’s show notes as well as the blog associated with this episode, I’ll link to a short fact sheet about Mathematica’s evaluation of the RSNAT prior authorization model. I also want to thank you for listening to another episode of On the Evidence, the Mathematica podcast. There are a few ways you can keep up with future episodes: Subscribe wherever you find podcasts or follow us on Twitter. I’m at JBWogan. Mathematica is at MathematicaNow.