Harnessing AI to Enhance Oncology Care Efficiency

Dr. John Sweetenham, Dr. Larry Shulman, and Dr. Rebecca Maniago discuss the integration of clinical pathways and decision support tools into the cancer center workflow, challenges to implementation at the point of care, and the promise of AI to further unlock these tools for clinicians.

TRANSCRIPT

Dr. John Sweetenham: Hello, I'm Dr. John Sweetenham, the host of the ASCO Daily News Podcast. Over the last decade or so, there has been a great deal of work and a lot of discussion about the implementation of oncology clinical care pathways at the point of care, which are designed to reduce variability in care, reduce costs, and improve the quality of care and outcomes. Although clinical pathways aim to guide treatment decisions, current data suggests that the utilization of these pathways at the point of care is very low. There are many reasons for this, which we will get into on the episode today.

My guests today are Dr. Larry Shulman and Rebecca Maniago. Dr. Shulman is a professor of medicine at the University of Pennsylvania Abramson Cancer Center. He's also the immediate past chair of the Commission on Cancer and serves on the National Cancer Policy Forum of the National Academies of Sciences, Engineering and Medicine. Rebecca Maniago is the director of clinical oncology at Flatiron Health, a technology platform that collects and analyzes real-world clinical data from electronic health records to facilitate decision making and research.

Our full disclosures are available in the transcript of this episode. Larry and Rebecca, welcome to the ASCO Daily News Podcast and many thanks for being here.

Dr. Larry Shulman: Thank you, John.

Rebecca Maniago: Thank you for having me.

Dr. John Sweetenham: Larry, I'm going to start out, if I may, with a question for you. You and I, in a previous podcast, have discussed some of these issues regarding pathway implementation before. But to start out with, it's certainly, I think, helpful for the listeners to remind us all of what are the benefits of oncology clinical pathways and why are we still talking about this 10 years or more on.

Dr. Larry Shulman: Yeah, and that's a great question, John. I think the good news is, and all of us who live in the oncology sphere know this, that there's been tremendous progress in cancer therapies over the last decade. But what that has entailed is the introduction of many new therapies. Their complexity is becoming really very tough for people to manage.

And so what we have are oncologists who are really trying to do their best to deliver care to patients that will give them the best chance for survival and quality of life. But it's really, really hard to keep up with everything that's happening in oncology in the context of what we all know is a very busy clinic schedule. Lots of patients coming through and decisions need to be made quickly. Pathways really could help us to guide us into recommending and delivering the best therapies for our patients for a particular disease. You know, cancer is complicated. There are many different types and there are many different therapies. It's just a lot to deal with without some assistance from pathways or pathway tools.

Dr. John Sweetenham: Thanks, Larry. So, knowing that's the case and knowing that these tools reduce variability, improve costs, improve quality of care as well. Starting with you again, Larry, if I may, why do you think it's been so difficult for so many oncologists to use these pathways effectively at the point of care?

Dr. Larry Shulman: So, I just wanted to step back a little bit. There are very extensive guidelines that tell us what the best therapies are for really all of the cancers. These guidelines come from the National Comprehensive Cancer Network or NCCN and the American Society of Clinical Oncology or ASCO and other professional organizations. And they're there. They're there, in free information off their websites.

But the problem is how to translate those pretty dense documents into something that will work in the clinic for a patient, for the physician who's working in the electronic health record. And the tools that are available, and there are a number of tools that can integrate with electronic health records, are expensive. You need to purchase them from the vendor and there are yearly fees.

And they're also difficult to implement. You need to work with the vendor to integrate them into your own rendition of your electronic health record. And there's a lot of customization that needs to be done. So, it's a financial challenge and it's also a time challenge for people to integrate these tools into their workflow, into their electronic health records.

Dr. John Sweetenham: Thanks, Larry. So speaking from my own past experience of pathway implementation, it certainly has been a major challenge for the reasons that you mentioned and also because of the, I think resistance may or may not be too strong a word, of many of the clinicians to use these for a number of reasons, part of which are the time it takes, part of which many of them feel that the pathways aren't really changing decisions that they might make anyway. So, you know, the uptake of pathway utilization, even in those centers which have been successful in getting something installed and plugged into their EHR, on the whole, hasn't been as good as it could have been. So maybe I'll turn to you, Rebecca, because I know that this is something that you've worked on a lot.

And it's a kind of double-barreled question. I think the first part of it is, you know, what do you think are the major roadblocks to high physician uptake in the use of these pathways platforms? And maybe you could talk a little bit about what the various software platforms do to make them more physician-friendly and to enhance utilization right on the front line.

Dr. Rebecca Maniago: Yeah, that's a great question. And so, you know, I've worked with a number of customers and physicians over the past five and a half years on implementing these pathways. And the number one pushback is really about the time it takes in the workflow. So, if I had a dollar for every time I heard "every click counts," I'd be a rich person and it does come down to clicks. And so, you know, as a software vendor, we really have to focus on how do we reduce that friction?

How do we make sure that the clicks we are asking for are the ones that actually matter? And how do we continue to streamline that process? And so, you know, while there is a fine balance, because as part of a Pathways platform, at the end of the day, we do need to understand some data about that patient. You need to understand the clinical scenario so you can surface the right treatment recommendation, which means there is some amount of data capture that has to happen. In some circumstances, you know, we can pull some of that data in from the EHR.

But unfortunately, the reality is that a lot of that data is messy and it's sort of stuck in documents and unstructured places. And so it doesn't easily flow in, which means we rely on the provider to give us that information. And oftentimes they've already entered it other places. So what's more frustrating than entering data twice? But, you know, I do see a great opportunity here. And this is certainly where software companies are focused is with AI.

So, know, for, especially for this data aggregation, a lot of these AI tools can actually scan through the chart instead of relying on the physician to sort of manually skim through and aggregate and find all that pertinent information. That's what AI is really good at. And almost instantaneously, it can find the messy data that lives in those unstructured documents. And wouldn't it be nice if that was automatically populated within these applications so that really all we're asking of the clinician is to validate that that information is accurate. And then choose the treatment that cuts down on the number of clicks, it cuts down on frustration. You know, again, the physician will be the one that needs to make that decision. AI is not there to replace that, but it certainly has a great opportunity to reduce some of this manual documentation and the things that physicians find the most frustrating, especially as it relates to using these pathways tools.

Dr. John Sweetenham: One of the pretty common pushbacks that I heard during my time in a couple of institutions was, "Well, you know, I'm sitting here at the point of care with my patients and I already know what I want to do and how I'm going to treat that patient if it's not in the context of a clinical trial. So I don't need to go through, you know, X number of clicks to get me to where I know I'm going to be anyway."

Does either of you have any thoughts about that? I think you've sort of partially answered it, but what do you think, Rebecca? Do you think that this is something that is more easily overcome-able, if that's even a word, than it was a few years back?

Rebecca Maniago: Yeah, I do. And I think this is where the customization comes into play. So while they may know what an appropriate treatment for their patient is, there are more options now than ever, which means at a local level, there may be multiple options that are clinically equivalent. And so when you think about things like payer pathways or drug margins as an organization, they have to drive some of that from within. But having the capability to do so can then start to sort of sell the value to the provider that, yes, you may know what you want to order for your patient, but would you consider something else if it was clinically equivalent, but it had other benefits to either the patient or the organization?

Dr. Larry Shulman: The other thing I would add to that, John, if I can jump in here is that the data is the data and the data shows us that guideline concordant care is not always prescribed to the US. And in fact, in some circumstances, the gaps between what should be prescribed and what is being prescribed are quite wide. So, you know, people feel like they're always doing the best job and making the best recommendations. And I think, you know, I think I am. But, you know, like many of my colleagues at academic cancer centers, I'm highly specialized. I only see patients with breast cancer. But many oncologists throughout the country are more generalists. They're seeing patients with multiple diseases. And it's harder for them to be completely on top of what the current recommendations are in any particular circumstance. Our diseases are complicated. They're getting more complicated all the time with molecular and genomic testing and subcategorizations of different cancers. So, I don't think that we can be too cocky about it, quite frankly. I think we ought to use technology that Rebecca describes for the tools and for AI to really help us. I think if we turn our backs on that, I think we're making a big mistake. You just got to look at the data. The data is pretty convincing.

Dr. John Sweetenham: You know ever since we started looking seriously at decision support through pathways a number of years ago, the word has always been around the payers role in this and the day will come where we are going to get reimbursed based on pathway and concordance and I'm not sure that that day has arrived. So I have a question for both of you in this regard actually. And the first of those is maybe I'll start with you for this part of it, Larry. Where do you think we are in that regard? And are you hearing more and more of payers starting to look at pathway compliance? And then on the other end of that, and maybe I'll ask Rebecca about this, is one of the other pushback issues that I used to experience from physicians I worked with was they may go through the pathways platform and come up with a treatment recommendation. The best example of this I can think might be that the recommendation might be a biosimilar. Let's just use that as an example. But the next stage in the process would be to find out whether the patient's insurance would actually cover that particular biosimilar, which opened up a whole new can of worms. So there are two kinds of payer aspects of that. Maybe Larry, I'll ask you to start off by talking about that kind of coverage issue. And then I'll ask Rebecca, if you have any thoughts about the flow the other way in terms of getting drugs approved and what we can do to help from an insurance perspective.

Dr. Larry Shulman: Sure, that's really an important point, John. Our current state of affairs with the payers and their attempt to be sure that we're providing responsible, guideline concordant care is the use of prior authorization processes, which are incredibly costly, both for the oncology practices and for the payers.

They have an army of nurses sitting at the phone talking to us in the oncology practices to decide whether they're going to pay for something. And frankly, generally, the payers will pay for things that are part of either the NCCN or ASCO or other professional organizations' guidelines. But you need to prove to them over the phone that in fact the patient qualifies for that.

We have actually had some experiments with some of the payers to prove that to them in different ways by auto transmission of data. And this would be a big savings for them and for us, it would take away some of the delays in therapy while we're waiting for prior authorizations to go through. And we shouldn't have to do this by phone.

The EHR and the pathway tools should aggregate the data, aggregate the potential treatment and be able to transmit those data to the payer. And if in fact it meets the appropriate criteria for guideline concordant care would be approved. Right now, it's a terrible, costly, timely manual process that they should be able to fix.

Dr. John Sweetenham: Thanks, Larry. And have you, you know, from a broader perspective, so not thinking necessarily about individual patients and specific issues around prior authorization, have you seen any movement among the payers to kind of get more aggressive about this and say, okay, you know, we are going to want to see your numbers, we want to know how many of your physicians are now using their pathways platform and so on. Are you seeing any word that that might be happening? Because certainly a few years back, that was the word on the street, as it were, that this day was coming.

Dr. Lawrence Shulman: And that's the proposal that we've made to several of our payers. Let us give you the aggregate data. If our guideline concordance is above a certain level, give us a gold card, give us a pass, and we won't need to do pre-authorizations. We've actually done that at my institution in radiology. Aggregate data gives individual physicians that pass if their guideline concordance was appropriate. I got to pass. So I don't need to go through those radiology pre-authorizations for my patients. And I think we can do the same thing with therapeutics. It's been a little bit more cumbersome to do it, and there's some detailed reasons why that is. But that's really what they want to know. And the payers want to know that patients are getting guideline concordant care, but they also realize it's not going be 100%. There are always a few outlier patients who require some variation from the guidelines. But if we get above 80% guideline concordant care, I think many of the payers would be happy to accept that as long as we continue to feed them the data. And that's the case in our radiology process with one of the payers is, you know, I get a gold card, but they continue to look at my data. And if I don't continue to perform well, they'll take that away.

Dr. John Sweetenham: Thanks, Larry. And Rebecca, just returning to you, this issue of prior authorization and facilitating life for the physician at the point of care in terms of knowing, you know, which specific treatment might be covered for a patient. Do you have any thoughts or maybe you could give us some insights on what software vendors are doing to facilitate that part of the process, the communication back to the payers to take some of that burden off the physician and the physician staff?

Rebecca Maniago: Yeah, absolutely. And this is a problem we've been trying to tackle for years. And it's not easy. We've tackled it in a couple ways. So first, we try to sort of link up to the payer portal where the information that was being attested to within the application could then be automatically sent. Because at the end of the day, the data points that are being collected to surface treatment recommendations ultimately are the same data points that the payer wants.

Unfortunately, there are a lot of data interoperability challenges within that space. So that was not something that was going to be sustainable. However, in current state, because as I mentioned, the customization is key for these products, focusing more on how can we allow practices to embed payer pathways within the application. So again, you kind of start with the backbone of your standard guidelines but then having the capability of adding in a payer pathway that will only show up as that preferred option for a patient who has that insurance, at least at the point of care, the provider sees what the insurer would then approve. So while it's not automatically assuring authorization, we are at least steering the decision in a direction where we think most likely this is going to be approved based upon the pathway that they have access to. So that sort of current state, I agree. We've been talking about this idea of gold carding for years.

Presumably the data is there today, right? Like we are able to capture structured data with every order placed to recognize concordance to Larry's point. All those reports are available to provide to payers. I just haven't seen a lot of practices have a lot of success when they tackle it on their own from that direction.

Dr. John Sweetenham: Right, thanks. Larry, you and I were at the NCCN annual meeting recently and I know that you've been quite heavily involved in the policy program and in the policy forums and so on at NCCN. Are you able to share anything from this year's meeting in terms of care pathways implementation and what you think might happen next in that regard?

Dr. Larry Shulman: NCCN, in my own opinion, has really led the way in defining what guideline concordant care is through their guidelines, which are very extensive, covering basically every cancer and every situation with every cancer. And it's really an astounding amount of amazing work that all of us use and the payers largely use as well. But they've increasingly understood that there's a gap between their guidelines and the implementation of their guidelines. And they are working on some things. They are working on the digitalization of their guidelines to make them more accessible, but also thinking about ways that they may, in fact, fit into the work processes that all of us have when we go to clinic.

They're acutely aware that the country is not where it needs to be in regard to a translation, if you will, of their guidelines in the practice. And I think we're all thinking really hard about whether there are things that we can team up to do, if you will, to try to close those gaps.

Dr. John Sweetenham: Great, thank you. Just switching gears a little bit back to you, if I can, Rebecca. I think you've said a little bit about this already. What do you think are the next steps that we need to take to more effectively implement these tools in the clinic? I think we've discussed a little bit some of the roadblocks to that. But where do you think we need to go next in terms of getting better use of these pathways?

Rebecca Maniago: Yeah, I will say one thing that we haven't really touched on is the pharmacy team. So the biggest blocker that I see is actually the pre-implementation. So there's a lot of focus on how do we get physicians to use this? How do we increase adoption? But often the first barrier is the regimen library. So no matter what the pathways platform is, the backbone of it will be those regimens. And so, really helping organizations and we partner with pharmacies, they're doing all the backend configuration. And so how can we make that piece of the technology easier for them to implement because that's really the lead up and there's a ton of cleanup and maintenance. You know, as a pharmacist, I empathize, but really that's where it all begins. And so I think, you know, continuing to focus on not only the front end user and the physician, but everybody that's going to be involved in order to make a pathway program successful needs to be, you know, at the table in the beginning, helping set up those processes and, and buying into the why this is important.

Dr. John Sweetenham: That's a great point.

Dr. Larry Shulman: So could I just jump in one quickly here, John? So pathways, as we've discussed, the tools are expensive. There is a person cost, as Rebecca is just describing, about customization and implementation. But there are very good data in the literature to show that when you follow pathways, care is less costly. Survival is better, which is obviously our primary goal, but also cost is less. And the payers can benefit from that. And the question is, can they figure out ways to use that to help to fund the purchase and maintenance of pathway products that will give their patients better care, but also less costly care? And so I think that is a potential solution. I've had that conversation with some payers as well. And it would be great to see that happen. I think that would be a huge step.

Rebecca Maniago: Yeah, we have some, if they're able to set it up in the right way and really optimize, you know, from the pharmacy perspective, we have practices who the application is more than, you know, paying for itself just by way of using it to the fullest potential that it has.

Dr. John Sweetenham: Yeah, that's a really great point. A couple of other more general questions. I'm going to start with you, Rebecca, and Larry ask you to respond as well. Are you hearing anything from patients around this issue? Are they aware or becoming more aware that pathways are being used in the clinic when they're seen by their physicians? And do they have a say, are there patient advocates involved in this part of the process? Rebecca, maybe you could start.

Rebecca Maniago: I haven't had as much exposure to that side of it. So, you know, I would love to hear what Larry thinks because most of my exposure is at the physician level, which of course they are the ones who are making the decision with the patient. So my assumption is that there is at least some level of understanding that there are options and that, you know, together let's decide on the best one for you. But again, I would love to hear what Larry has to say.

Dr. Larry Shulman: Yeah, so that's a really interesting question. I actually was discussing that at the cancer center last week, particularly around the utilization of AI in this process. And, you know, right now, as you know, if you submit a journal article or, you know, many other things, ask you whether you used AI to generate it. If in fact we use tools that include AI, we're not.

Are we obligated to tell the patient that you're making this recommendation together with computer assist, if you will, that helps you to make the recommendation you are making to them? Ultimately, I think it's the physician who's responsible for the choice, but should we disclose it? I have to tell you personally, I haven't thought about doing that. But I think it's a really, really good question is whether we should upfront tell the patients that we've had assistance in making the recommendations that we have.

Dr. John Sweetenham: Right, very interesting point. To close it out, one more question for both of you and again, it's the same one. Rebecca, to start with, we've all been, as I said right up front, talking and, you know, working on this issue for more than 10 years now. In 10 years from now, how would you like it to look and how do you think it might look?

Rebecca Maniago: Great question. I think we may get to where I would like to see it quicker than 10 years. I think AI provides a lot of opportunity and excitement. I'd love to turn a corner where physicians no longer see tools like this as a hindrance, rather they rely on them, they trust them, they help them get through their day. They continue to improve quality of care and reduce costs and patient burden. Obviously, that's the pipe dream, but I think we may get there before 10 years, given what I think AI is going to enable.

Dr. Larry Shulman: Yeah, I want to add to Rebecca's comments. One of the things that I worry about, and ASCO worries about a lot, is the oncology workforce, which is progressively strained in their attempts to care for all the cancer patients in the US. And for all of us who practice oncology, for many reasons, it's become more and more inefficient, whether it's use of the EHR, pre-authorization work, and so on.

And we really need to turn that around. We need to make practice not only better, which I think these tools can do, including AI, as Rebecca says, but make it much more efficient because that's going to allow us to both deliver more high-quality care to our patients, but also to care for more patients and have them benefit from our expertise and what we have to offer. So I think this is really an obligation on our part. I think it's an imperative that we move in this direction for both quality reasons and efficiency reasons.

Dr. John Sweetenham: Thanks, Larry. Well, I've really enjoyed the conversation today and I think, you know, it's been great to think about some of the challenges that we still have in this regard. But it's also great to hear what I'm sensing is quite a lot of optimism about how things may play out over the next few years. And it does sound as if there's a lot of hard work going on to bring us to a point where the clinical decision support tools are going to truly support what our oncologists are doing and no longer be seen as an obstruction. So, I want to thank you both for sharing your insights with us today on the ASCO Daily News Podcast.

Dr. Larry Shulman: Thank you so much, John.

Rebecca Maniago: Thank you so much.

Dr. John Sweetenham: And thank you to our listeners for your time today. If you value the insights that you hear on the ASCO Daily News Podcast, please take a moment to rate, review and subscribe wherever you get your podcasts.

Disclaimer:

The purpose of this podcast is to educate and to inform. This is not a substitute for professional medical care and is not intended for use in the diagnosis or treatment of individual conditions. Guests on this podcast express their own opinions, experience, and conclusions. Guest statements on the podcast do not express the opinions of ASCO. The mention of any product, service, organization, activity or therapy should not be construed as an ASCO endorsement.

Find out more about today's speakers:

Dr. John Sweetenham

Dr. Lawrence Shulman

Rebecca Maniago

Follow ASCO on social media:

@ASCO on Twitter

@ASCO on Bluesky

ASCO on Facebook

ASCO on LinkedIn

Disclosures:

Dr. John Sweetenham:

No relationships to disclose

Dr. Lawrence Shulman:

Consulting or Advisory Role: Genetech

Rebecca Maniago:

No relationships to disclose.