Arcadia CMO on Data Challenges Facing Health Plan Leaders

Healthcare data platform company Arcadia just released a survey report highlighting a widening gap between the data capabilities that payers need and the tools, infrastructure, and analytics strategies they have. The company’s chief medical officer, Luke Hansen, M.D., M.H.S., recently spoke with Healthcare Innovation about some of the data challenges payers are dealing with.

Earlier in his career, Hansen worked at United Health Group as the Illinois Market Medicare Advantage chief medical officer and a senior medical director for population health at Optum. He also held health system roles as a population health CMO, including medical directorship over capitated risk-based contracts at an MSSP ACO.

Healthcare Innovation: Dr. Hansen, it occurred to me that perhaps you have personal experience with some of the challenges or frustrations around data access or data integration from your times working for payer organizations.

Hansen: It's funny, because I had worked in provider organizations my whole career, until I joined United. The main reason I made that step was because I felt like as a provider interested in value-based care and improvement that the health plans held all the levers, certainly the payment levers and and so I thought if I move over to that part of our industry, maybe I can be closer to opportunities to engineer better levers. And then you get there and you realize there's a whole other set of challenges over on this side. I think this report speaks to that — it's not simple anywhere in our business. So I I'm not particularly surprised at what the respondents shared.

HCI: Let’s walk through some of the key survey findings and delve into some of the implications. One is that data integration lags behind data sharing. The report noted that 66% rated their experience in sharing data as good or excellent, but only 57% said the same about integrating data from external sources, and 85% said they have yet to integrate all available data into a centralized platform. Why is it they'd be better at data sharing than integrating external data? Does it involve technical or resource issues?

Hansen: I would say that the the challenge — and this is true on the provider side, too — is that there are different facets of our industry. There are business owners and there are clinical owners. The thing about the value-based care paradigm is that it integrates those two in a much tighter way than has historically been the case. It was not that long ago that the medical staff did what it's going to do and the business leaders would make sure that the lights stayed on. That's not at all the case anymore. To be a success in VBC, which is really what health plans are endeavoring to do, it requires that tight integration in a way that hasn't been expected before.

HCI: The report notes that AI and advanced analytics are relatively untapped. Less than 20% of plans use AI to develop care plans, summarize records, or drive member engagement. Is it still too early to expect to see those numbers higher? Are there vendors out there focused on the payer space offering those type of tools, or are some large payer organizations developing their own AI tools internally? 

Hansen: I think it's important to make a distinction between the AI tools and the underlying data that they rely on. And the coverage often focuses on the tools, which are novel and great, but they're only going to be as good as the underlying data asset. When you have fragmented data, the AI can't really leverage the complete story. And without that, it's really going to be limited in its utility. It is also a question of where to invest. Which tools that leverage the technology will be most helpful to plans? I think some of that is about document processing. Probably that's the easiest one, but I think there's real opportunity around prior auth advising and automation and that will come sooner rather than later, given there's a lot of policy movement around improving the prior auth experience for everybody involved, but that is all dependent on a good data asset underneath. If you don't have good data informing those models, then it's not going to be a helpful addition.

HCI: Is it difficult to make generalizations about health plan data sophistication? Are larger organizations investing more in this space? 

Hansen: At their heart, health plans are insurance companies, so their tooling is strongest around pricing risk and member acquisition. They have the most robust analytics there. It is not their primary function as an insurance company to manage clinical care in the system as it has evolved. That is something that they end up doing, de facto, for members that they have risk for, and those members may not be engaged in care otherwise. 

That's a great example of where there's no one else doing the work, because the provider side of our industry is still relatively reactive. And unless you as a patient are going to reach out for help, they're not going to come get you. So health plans, I think, lag in that set of competencies. I think that there are plenty of small, local regional health plans that are doing a great job. I think that the benefit they have is they understand their communities in sort of a fundamental way, and they have longstanding ties and interdependencies with local providers. Now, on the other hand, the national plans obviously have scale and resources available to them that a small plan doesn't. So there's a little bit of opportunity on both ends of that spectrum.

HCI: One of the things the report talks about is member 360-degree profiles. Can you talk about why that's so important and why incomplete ones might lead to missed opportunities for health plans?

Hansen: In the report, we were very intentional not to talk about the EHR, but an EHR is a 360-degree longitudinal clinical record for a person. We would see a health plan using a 360-degree profile with some similar components to it. Because it's longitudinal you have more history coming into it, and it's much easier to predict the future. The business of pricing risk and predicting risk and acting proactively — you’re not going to be very good at that without it. The comprehensive nature of it is equally important. I alluded to this before about risk. As you know, there's been this balance in the MA space of revenue drivers and cost drivers. And for a long time, frankly, the revenue drivers were sufficient to be successful as a health plan, which is to say, you grow your membership, you support appropriate documentation of risk so that you're getting revenue coming in through the risk channel, and then you manage your stars, you manage quality, and you get the quality payments and that was enough.

All of those things are in a period of uncertainty. Whether it's Version 28 for risk adjustment and more regulatory scrutiny, in the last year there has been a lot of retrenchment both on the part of of health plans, but also providers are not as sure that they want to be robust participants. So all that said, it puts pressure on the cost containment side. And in that world, risk has a different definition. It's not just about HCC [Hierarchical Condition Category]. It's about that social risk and all the things that would determine how a person is going to use care. If you can't profile a member and understand where they're going to go, it's very hard to intervene and get them tracked to the most appropriate, effective site of care.

HCI: The report mentions the value of SDOH data, episode-of-care data, and pharmacy data sources. Are those all relatively new sources of data for the payers?

Hansen: They’re each novel in their own way. So for health-related social needs, the plans have been looking at that information for a long time, but it's not systematically captured. So patient-level specificity on that remains elusive. 

The episode-of-care stuff is super interesting. At Arcadia, we think of episodes in a certain way, which is to say, a certain condition and an associated arc of care. Although the respondents in this survey said that they use episode-of-care data, I'd be interested if they're able to really leverage that in terms of an observed to expected performance? I think they have observed to expected performance at an aggregate level for a member for PM/PM, or per member per year. But particularly in the specialty sector I think there's opportunity to look at how effective is a cardiologist at managing an episode of heart failure, which could be 365 days? Are they managing it above expected or below expected? The 360-degree profile gives you that longitudinal view that would help inform a longitudinal episode. 

On pharmacy data, it is surprising to me that a relatively small number of respondents said that they were using pharmacy data because several of the big plans have pharmacy benefit managers, so they're in the mix. But I think there is a huge opportunity in terms of management, and the plans are going to be key to this. A physician can prescribe the right medicine in the confines of the exam room. But then, so often, members or patients show up at the retail pharmacy counter to fill that and they're told that there’s this out-of-pocket obligation and they walk away. That moment is such a critical place where alerts need to be generated. When I think of how pharmacy data could be better used, that's the moment where the plan will know that the script wasn't filled, and the provider won't necessarily know that. There's so much emphasis on ADT alerts around discharges. If we had a similar sort of ADT alert around pharmacy no-fills, I think it could be so powerful. 

HCI: In the press release about the report, you're quoted as saying that when payers and providers work from a common data foundation, it becomes possible to manage risk, optimize utilization, and enhance physician performance more efficiently. So are payers opening up their data and analytics to share with provider organizations around managing patient risk and preventive services?

Hansen: To render it very basically, the plans hold the claims and the clinicians hold the clinical data, and it's only through a recipe that includes both that you can establish the most appropriate care plan and the most realistic, actionable care. It’s one thing when a doctor in the confines of the office says we should do A, B and C. But if the plan doesn't include insight around the benefit structure and the availability of services, they are going to have problems in seeing that plan realized. 

So certainly, communication is important. And I think it should be said that those two sectors sit across the contracting table from from each other. So there is an inherent skepticism and even suspicion, which I think is part of the barrier that needs to be overcome. I think both sides would say that, but at the same time they are business partners, they also are negotiating against each other much of the time, so it's a complicated relationship, and I think that's held back data sharing.

HCI: Was there anything else that surprised you about the findings or that you would want to highlight?

Hansen: What I would say is that going forward, there is going to be a differentiation in performance for plans, depending on how well they can integrate these different data streams and and to some extent, that integration is dependent on provider collaboration.

So the ability to collaborate well with a provider partner will in many ways bear fruit for them, including the fact that they'll have better access to clinical data in real time that can allow them to support members better. They will then be able to feed that forward through advanced analytics back to providers in real time, so that at the point of care appropriate decisions can be made to achieve everybody's shared interest.

There will be plans that continue to operate in this siloed way, and I'm not sure that that is sustainable. For instance, if you are going to evolve down a road toward delegated risk and eventually capitation, I don't know how you do that well when you're delegating that risk to a provider without that kind of tight data integration. 

HCI: Who is it in a health plan leading this decision making around data strategies? Is it usually the CIO or is there somebody in a different role or a team?

Hansen:  I would say that it depends on the size of the plan. What I would like to see happen is that the segments of plans that are most directly tied to clinical performance would be able to make the case that awareness of front-line clinical activity is critical to their work, so it's critical to the plan success. When the fuel for it is coming from a clinical place, I think that can be quite powerful. There is this common clinical interest between the health plans and the providers, and if clinical excellence is what we're setting our compass to, then we won't go wrong.