Subscribe
Share
Share
Embed
340B Insight provides members and supporters of 340B Health with timely updates and discussions about the 340B drug pricing program. The podcast helps listeners stay current with and learn more about 340B to help them serve their patients and communities and remain compliant. We publish new episodes twice a month, with news reports and in-depth interviews with leading health care practitioners, policy and legal experts, public policymakers, and our expert staff.
Narration [00:00:04]:
Welcome to 340B Insight from 340B Health.
David Glendinning [00:00:12]:
Hello from Washington, D.C. and welcome back to 340B Insight, the premier podcast about the 340B drug pricing program. I'm your host David Glendinning with 340B Health. Our guest for this episode is Kristen Chupka, the System Director 340B program for Dartmouth Health, based in New Hampshire. Kristin was one of the experts kicking things off for the Pre Conference workshops at the most recent 340B Coalition winter conference. In her panel, she spoke about the concept of applying automation and artificial intelligence to 340B. So we had her come on the program to discuss that exciting topic. But before we get to that interview, let's do a quick recap of some of the latest news about 340B.
David Glendinning [00:01:03]:
340B Health and two of our member hospitals have asked a federal court in Washington, D.C. to throw out lawsuits from four drug companies that are seeking to impose 340B rebate mandates on covered entities. Our association filed that motion along with Genesis healthcare system and UMass Memorial Medical Center. Regular listeners of the show will know that the judge overseeing the cases granted permission for the three of us to intervene in those cases as defendants, which allows us to participate more fully as a party in the case. You can find more information in the show notes about this litigation, including the first motion the government filed in these cases since President Trump took office. That motion indicates the new administration believes HRSA made the correct decision by declining to allow drug makers to impose unilateral rebate schemes. And it notes that the agency has long envisioned upfront discounts as the preferred price reduction mechanism for 340B covered entities, the drug companies in the cases, and the government. We'll have a chance to respond to our motion by April 1st.
David Glendinning [00:02:26]:
And now for our feature interview with Kristin Chupka with Dartmouth Health. Stories about harnessing the power of automation and AI are all over these days, and that includes in the world of 340B. Attendees of one of the pre conference workshops at the Winter Conference participated in a lively discussion led by Kristen about the potential benefits and pitfalls of this technology, and we met with her at the conference as it was wrapping up to continue that discussion. Here's that conversation.
David Glendinning [00:02:58]:
I am here with Kristen Chupka, who IS system director, 340B program for Dartmouth Health. Kristen, welcome to the last day of the 340B Coalition Winter Conference and welcome to 340B Insight.
Kristen Chupka [00:03:11]:
Thanks for having me.
David Glendinning [00:03:12]:
So I had the pleasure of attending your Pre conference workshop in which you spoke about automation and AI in 340B. But before we get into that, tell us a little bit, if you could, about Dartmouth Health and the patients you serve.
Kristen Chupka [00:03:27]:
So Dartmouth Health is a small rural health system. We are expanding and have brought more hospitals in this last year, which is really exciting. We have an academic medical center in Lebanon, New Hampshire, which is a disproportionate share hospital. Also under that dish is a hemophilia treatment grant Ryan White, Part C, part D. So that makes things really fun for us to have some different aspects to that large academic medical center. We have two rural referral centers. One of them is in Bennington, Vermont. The other is in Keene, New Hampshire.
Kristen Chupka [00:04:02]:
The rest of our system, hospitals or critical access, spread out around Vermont and New Hampshire. We really pride ourselves on our rural environment and our healthcare services that we can provide. We've got lots of specialties in the south of New Hampshire, but also really doing some expansion in our academic medical center with a new patient tower to expand mental health services in the state. So we really value the opportunity in that environment to care for our patients.
David Glendinning [00:04:34]:
Thank you for running through all that. Now, when it comes to the subject AI and automation, you had said during the workshop that you are not an expert, but an enthusiast, which I liked. You did say you were a Harry Potter expert.
Kristen Chupka [00:04:48]:
I am, which is great. I am an expert there. Thank you.
David Glendinning [00:04:50]:
Well, for all the Hufflepuffs out there, what is the difference between automation and artificial intelligence?
Kristen Chupka [00:04:59]:
So this is what I love about it, because technology is advancing, right? And I kind of had bucketed the two together, thinking, right, well, you know, they're going to work together and they will. But when you think about AI, it's a computer that you've told what to do and it's doing all the work for you and analyzing and giving you predictions and kind of running itself. When you have automation, it's similar, but it's not making decisions for you. It's going through the algorithm and what you've taught it to do. So, yes, you can have them together. But it's really important to understand automation is not AI. And AI has some challenges in terms of that emerging technology and what we want it to teach us, tell us, predict for us. And that could get you into some maybe ethical or legal things that you might need to consider.
Kristen Chupka [00:05:51]:
But I do like this technology and I think it's been fantastic for us. So I'm happy to talk a little bit more about, you know, how we're using it or different things related to AI and pharmacy even, or healthcare.
David Glendinning [00:06:04]:
Yeah, well, thank you for running through that difference. It's an important one and it did come out, come up through the workshop. So let's talk about some of those applications. How might you use automation or AI in a pharmacy setting?
Kristen Chupka [00:06:19]:
One of the big ones right now, which I really think is a game changer, is in the specialty authorization, right? With your specialty pharmacies, AI can go out on the Internet, pull the correct PA form, integrate with your emr, fill that form out for you and submit it. Right. You just have to teach it where in the EMR to go, how your EMR is configured and what payer plans or whatnot that you need to make sure you're pulling the right form for. So if you can automate that, now you have all your pharmacist or technician resources that aren't filling that out manually and digging through the EMR and you can actually spend more time with your patients. So that's one big one that I think is emerging with the AI in healthcare. Also, you know, thinking about repetitive tasks in a pharmacy and how you can do other things with documents and kind of free up time. So the repetitive task portion of utilizing AI is really amazing right now in healthcare.
David Glendinning [00:07:22]:
And the podcast is called 340B Insight. So we are mostly interested in that subset of pharmacy. So when it comes to 340B, specifically, how might entities bring in automation or AI?
Kristen Chupka [00:07:36]:
So I get really conservative, right? I don't want to teach AI to audit. For me, I think maybe in a few years I might get more comfortable with it. But I get nervous about an emerging technology that might tell you that you have no compliance issues because it's learned something incorrectly or you've taught it something incorrectly. And then what? What are you going to do? Right? So I really am endorsing the automation side of being able to teach a logarithm, if you will, to go through and say, I want to check my provider, I want to check my location, I need to run an orphan check. And that's very black and white, right? And it can shoot you out a bunch of examples that are fine, right? There's no issues, everything checked. We like to do ours at no variation. So if every single metric matches great, we may find that on the other side of the coin, when you start talking about automation, it might drop something and say, well, this is not the same provider, or this is not the same location. And we go and audit that and it's still an okay claim.
Kristen Chupka [00:08:50]:
But I don't want to teach an AI to override that. I think the challenging part about putting AI into the auditing space is the integration with the EMR and some of the considerations there for patient information. And our organization does really value that. We want to keep our patients safe. And so, you know, as this technology emerges, right. And more of the linkage with AI and this technology comes forward, I think there'll be more opportunity to think through, through that and how we can have it be compliant in a way that makes us feel comfortable.
David Glendinning [00:09:28]:
Yeah, I'm hearing opportunities there potentially for better compliance, for more efficient 340B operations. That all sounds great, but you've alluded to some downsides. So what are the downsides? Why not just turn on automation or AI throughout the 340B universe?
Kristen Chupka [00:09:45]:
I think the big one really is ethical, right. It's an emerging technology. I, I mentioned in our, our session, right, about Star Trek and the Borg and like the computers taking over the world. And that's not really happening here, but people are nervous about it and it, it makes some hesitancy in adopting a new technology when you think about that. And so the big one, you know, that I think about a lot is informed consent and the patient knowing, right. That you are using a tool that is not your brain and do we need their permission? At what point do they need to be involved in that conversation? I also think about, right, if I have a provider and they're using AI to make a decision about my healthcare, should I be aware of that? Do I think less of them for using something that isn't what they've been been taught? Right. And when you think about AI, is it going to Dr. Google and pulling the incorrect information and then leading my provider down a path that's not best for me? So those are some of, I think, the ethical considerations related to AI and healthcare.
Kristen Chupka [00:10:55]:
And I think you can extrapolate those into the 340B considerations as well. Legally. Right. How many of the state board of pharmacy have adopted regulations about utilization and pharmacy? What about all of the security functionality about putting that data out there? And then if we go back to the patient and provide a relationship like how does that relate to liability? Right. If, if I make a mistake as a provider and use AI and, and really harm someone, what happens if I come out and say, well, oh, the computer told me to go ahead and do this, right? That's not going to bode well.
David Glendinning [00:11:31]:
Okay, so you've spoken about some of the legal and ethical considerations that need to be brought to mind when thinking about AI. Where are the areas in which Dartmouth Health might be considering turning on this technology?
Kristen Chupka [00:11:45]:
So since we're feeding it all of our claim data, right, to be able to identify is this a compliant claim in our auditing processes? One of the, I guess interesting ways that it has evolved is why can't we do this to put our financial packages together, build KPIs about trends and where our patients are filling and how many claims we're capturing and what you know, pharmacies are being utilized across the country. And I think there is an opportunity there for AI to do some modeling for budgeting, take into consideration the rebate models and give us a couple of different pictures of what our next year year might be. And it could even take those into the KPI space. Like do we think that we're going to have an increased capture this year based on modeling? So once we get that project through and really kind of see those financial dashboards, that's the next step. And I think that will get us really comfortable with thinking about other way to utilize AI in the auditing space. With mixed use. We haven't really done much auditing with the automation there. And so I think about the volume of claims and how much there is to look at.
Kristen Chupka [00:12:54]:
But is there an opportunity for AI to evaluate the different databases on the NDC files and package sizes and quantity multipliers that people struggle with. And that might be a really good way to check some of those things in a very easy way if the AI is searching all of those, those databases and providing feedback. So I think there's a lot of opportunity there to kind of get started slowly.
David Glendinning [00:13:19]:
And what about what Dartmouth Health has already done in the area? It sounds like you're kind of maybe starting to dip your toes into, into thinking about AI, but automation, you alluded to having used a bit of that already. So you know, how has Dartmouth Health used automation in its 340B operations?
Kristen Chupka [00:13:36]:
So when we started, I think about two years ago in trying to build this resolution with our partner, we were only auditing about 10% of our claims. We have two entity owned pharmacies that are under our disproportionate share hospital. And just for that entity alone, it's about 16,000 claims a month. So 10% really isn't much. And you know, we really get nervous about, you know, what we're missing, you know, and we're doing bulk checks on those claims. But you know, when you really start diving into the emr, something that Might look good, isn't necessarily compliant. So now we're auditing at 100% across all of our nine covered entities, we were averaging about 30,000 claims a month, right? So thinking about the system, that's a lot. And what we've been able to do is really get that count down.
Kristen Chupka [00:14:31]:
So for our dish covered entities, we're averaging about 2,000 claims a month that actually drop what we call flags for any of those items that don't match. And then that's what my team audits. What we have found is kind of interesting on the back end, right? Once everything has gone through, and that is One of our TPAs, we have three actually has a sustained flag rate over our 10% auditing where we were in the past. And that's kind of fascinating, right, that you have a TPA that is also, right, supposed to be taking in all of these elements and providing you a compliant output. And we're finding that we have all these unmatched things that need to be audited. Now, that doesn't mean that they're all not compliant. And I know I mentioned that before, but I want to stress that, right, we could probably think of different ways to tie that back, but we like knowing we didn't have the matches that we were expecting. The last thing you know, there I want to mention is we didn't want to set it and forget it.
Kristen Chupka [00:15:40]:
If you set it and forget it, you don't know that the automation isn't working properly. And I could even extrapolate that into any sort of AI, right? So there has to be a process to check. So we call it audit the auditor. So for anything that doesn't drop that compliant zero flag, right, we check about 30 claims a month per our entity, each one to make sure that the logic's working and that when it says it's fine, it's fine. And that gives us a real good deal of comfortability with the process and what we've implemented. So the other thing to remember, right, you're teaching an automation to do a service for you. And the other part that we expected, but also maybe didn't anticipate the checks and balances of keeping that in line. So monthly updates and having a process to make sure that your new providers get added, your Medicaid plans get updated, and that those tables in the background are going to continue to provide you a solid output.
Kristen Chupka [00:16:43]:
So that's definitely a consideration. In terms of workforce, I would not say that implementing this technology decreased my need for FTEs, because we're still auditing a good amount of claims every month, but what we're able to do is now audit all of them and still hit our mark for compliance.
David Glendinning [00:17:04]:
Yeah, it's interesting to hear you speak about the FTE issue because that certainly did come up during the workshop in terms of the concerns over AI, for lack of a better term, taking people's jobs. And so it's interesting to hear that you're able to turn this on and get those results without really having a sizable impact on your staffing considerations.
Kristen Chupka [00:17:28]:
I think that's also because we are doing 100%. If you're going to use this kind of technology and then still only audit a small percent, you know, you're missing the opportunity and having a fully compliant program. And, you know, we knew with the size that we had, there was no way we would need a team of probably 50 or 60 analysts to complete all those claims every month. So this was a good solution for us, and our team really enjoys working with the system and. And the level of transparency it gives us.
David Glendinning [00:17:59]:
Great. Let's speak a little bit more about the workshop, because some of the attendees there had their own experiences, had their own views on. On automation, on AI. So how did you feel about that conversation? How do you think the other workshop attendees viewed the opportunities and risks for AI, specifically in 340B?
Kristen Chupka [00:18:20]:
So I really wanted to poll the group. So it was an interesting, interesting session. I was able to ask that of the audience. And. And one person said they disagreed with me in trying to implement AI in the auditing space, but another person said, it absolutely scares me. I don't want to think about it, which, you know, valid. Right. It's a new technology and it's out of our comfort zone, but you do need to get comfortable with the uncomfortable.
Kristen Chupka [00:18:47]:
So I really hope that, you know, with that talk and people listening to this podcast will hopefully start thinking about ways to not be scared about it and ways that you can get comfortable with a solution that will help your covered entity maintain compliance. You know, I think about opportunity, and I think with this technology in this space, next year we could probably have the same talk. And maybe I thought about some different things, so I would love feedback from the community. What else could we do with AI and automation in this space so we can work harder, easier, and get everything done that we want?
David Glendinning [00:19:28]:
Well, Kristin, this is a very interesting topic and potentially, as you say, you know, great opportunity here for covered entities if they proceed thoughtfully and with the appropriate amount of cost. So really look forward to seeing where Dartmouth Health can go with this. So we'll we'll have you back on and talk more maybe in a year or more and see where you're at at that point. But in the meantime, thank you for giving us a view into this world.
Kristen Chupka [00:19:54]:
Well, thank you for having me. I again, I I hope that people will warm up to the idea after listening to this and maybe we can advance this space together.
David Glendinning [00:20:06]:
Our thanks again to Kristin Chupka for being with us for this look into a fascinating technology subject. We expect that many share some of the views expressed about automation and AI in that pre conference workshop, ranging from cautious to excited about the future. We look forward to having her in the booth again to check back on where Dartmouth Health might be in that journey. How does your health system use automation or AI in the 340B setting? We would love to hear from you. You can email us at podcastree40bhealth.org we will be back in a few weeks with our next episode. As always, thanks for listening and be well.
Narration [00:20:47]:
Thanks for listening to 340B Insight. Subscribe and rate us on Apple Podcasts, Google Play, Spotify or wherever you listen to podcasts. For more information, visit our website at 340bpodcast.org. You can also follow us on Twitter @340BHealth and submit a question or idea to the show by emailing us at podcast@340bhealth.org.