Aver Insights recently spoke with Scott Orme, Aver’s expert on all things analytics, about the power of custom episode design for payers. The following transcript has been lightly edited for length and clarity.
Aver Insights (AI): Before we get into custom episode design, can you tell us a bit about Aver’s off-the-shelf episode solutions?
Scott Orme (SO): Sure. Aver has gathered open source and licensed episode definitions and meticulously developed logic to capture the business rules so that we can take any claims data, apply that logic, and build out all of the defined episodes. So, for example, if a client is interested in episodes for a working-age commercial population, we can apply the PROMETHEUS episode suites against all that data and provide them all the relevant information needed to set up an episode payment program. We also have some definitions that are used less frequently, such as by some state Medicaid programs, but the dominant two with the largest market share are PROMETHEUS and Bundled Payments for Care Improvement (BPCI), and that’s where we put the majority of our efforts.
AI: With these standard solutions available, why would a payer consider customizing?
SO: Custom solutions are superior to off-the-shelf models in several situations. First, a payer might have, and wish to maintain, a very unique relationship with a provider, such as a unique pricing structure or agreements about the types of services included in their standard working agreements. This could require specific modifications to things that would normally be considered part of an episode for which the provider would be at risk.
Second, a payer might have a very specific program in which they assign members to physicians and want to use those attribution rules for episode assignment, whereas the stock model has algorithms for attribution. Aver can easily apply the payer’s attribution rules, customizing episode assignment.
Finally, a payer and provider may wish to enter into an arrangement with an episode that doesn’t really exist in the literature, perhaps it’s a rare, esoteric episode, or maybe a hospital is specializing in advanced cardiac procedures and they have a specific valve bundle. Aver can help these potential partners create an episode that doesn’t otherwise exist.
AI: Can you offer an example of a modification Aver has made to an existing episode definition?
SO: For example, a payer might work with a hospital that doesn’t directly offer physical therapy (PT) services and must contract with an external source. That hospital may want to negotiate PT out of the bundle because they have less control over it. The payer may disagree, but the parties can at least have the discussion because our tool can easily make that modification. In another example, a payer client started with a stock model episode and, in viewing the data we provided, realized a hospital surgeon was using a procedure not included in the business rules but that was clearly an acceptable practice. We created a modification to add specific codes that better reflected the care pathway for that episode.
AI: What about bundles Aver has built that didn’t previously exist?
SO: Just a couple examples are spinal fusion bundles and some specific heart procedure bundles that are proprietary to different clients. We’re working with some payers to create new episodes on everything from carpal tunnel to hammer toe. In this case, we are doing what we would call “discovery bundles.” These are broadly defined bundles meant to capture everything that happens in an episode. Then, we statistically evaluate the prevalence of certain claims and practice patterns. From there, we define an operational bundle, which is the bundle definition these providers are evaluated against for reconciliation.
AI: For those newly created bundles, are you looking at historical data or real-time data?
SO: For the most part, we use 2-3 years of historical data to get a sense for how things operate today. Moving forward, we are developing some operational episodes with clients that are more real time, in which we’re gathering and evaluating claims as they are submitted.
AI: What types of payers should think about customizing?
SO: A large payer with a broad geographic footprint is more likely to need the ability to modify definitions. Depending on where their network providers are located, they may experience different negotiating leverage in parts of their footprint with more competing providers than in areas with fewer providers. One of the issues in customizing episode definitions is that both sides must agree on how these things are defined.
AI: What are some of the challenges a payer might face in implementing that customized solution?
SO: First, it is much easier to defend a national known model against challenges from providers. With a customized episode definition, trust issues come into play. It is imperative that providers use a defensible rationale. And yet, we’ve found that a lot of the modifications payers make are usually conciliatory towards their provider partners, often in response to providers not wanting to work in the stock model, and they negotiate to something with which both parties are comfortable.
Second, if you customize dramatically, there’s a point where you get back to fee-for-service because everyone has their own episode flavor. At some point, you have to set some limitations. Maybe you have three classes of episodes types for all the providers you work with. But if you have 50 providers and 50 episodes, you’ve probably not solved your problem.
AI: Are payers trying to bring providers in the door in the hopes of increasing risk over time?
SO: I think those that have thought about it do have a strategy for moving toward shared risk and shared reward. Phase 1 might be simply learning, seeing what the claims history illustrates. Phase 2 might be shadow pricing to show providers how they would have done in an episode payment model. Phase 3 could move to shared risk and a reconciliation model involving a target price and quality metrics.
AI: In what ways can Aver customize an episode?
SO: The great thing about our tool – and I say this as someone who comes from a place where we used other tools – is the ease, flexibility, and nimbleness we have around multiple dimensions. Time is very easy for us to modify, but that can be very difficult with other software products. Our modifications can be as detailed as specific codes being in or out of the bundle, or as broad as changing time windows or changing the logic for attributing an episode to a provider. We can create and build a threshold for excluding or capping episodes, and even modify thresholds or rules for kicking an episode out for manual review. For example, the way the PROMETHEUS stock model works, what used to be a readmission may no longer appear as a readmission because it is rolled up into a different episode. We can create an all cause readmissions flag on every episode and quickly add that into the rules. If there’s a way to think of a rule that uses claims, we can model it very easily. It’s hard to think of a scenario where we couldn’t model it.
AI: How does Aver work with payers as they work through the process?
SO: Our process is responsive to client needs. A few clients are users of the software and we help them with their methodology. One client is building a unique cancer bundle with some fairly complicated steps, and we are consulting with them on how to use our software to best match the clinical rules they want to impose. Conversely, we have another client where we are their backshop, so to speak. We take their questions, run the data, provide a report or visualization in response, meet with the client and review to determine if we’ve met the need and answer follow-up questions. There is quite a bit of freedom for as much as clients want to invest into doing the work versus consuming the work.
AI: Is there anything else for payers to know about custom episode design?
SO: I think if payers really want to get the most value out of the experience, they should learn about how our tool works and about the episode models they are considering. Once they dig into the data, we assist them in creating levers and toggles that they can manipulate and look at the “what if” scenarios. If I’m a payer getting ready to implement bundles with a provider group, I would want the ability to run four different versions of an episode to do some “what if” comparisons – consider several different modifications and the tradeoffs involved. Our software can run those things simultaneously and provide the results; they don’t have to submit a request to an IT shop and run one change at a time. If I were a payer, I would start thinking about my “what if” scenarios and find a partner like Aver that can help me work through them.
AI: Thanks so much for your time, Scott.
The next and final blog in this episode design series will focus on accounting for risk in bundle design. Sign up here to have it and all future posts delivered to your inbox.
Want fresh insights on Value-Based Care delivered right to your Inbox? Become a member of the Aver Insights Community.