Aver Insights recently spoke with Steve Kohlmann, Aver’s R&D expert, about the use and importance of risk adjustment in a bundled or episodic payment program. The following transcript has been lightly edited for length and clarity.
Aver Insights (AI): Starting at a very basic level, what is “risk” in the context of an episode or bundled payment model?
Steve Kohlmann (SK): At its core, an episode or bundled payment model is asking providers to accept a set amount for taking care of a patient. This payment model requires a mechanism to recognize that not all patients are exactly the same. Risk adjustment in a bundled payment model seeks to identify historical risk factors that may cause the cost of a person’s care to exceed the standard bundled payment amount. Because the bundle is disease or condition specific, the risk factors are tailored to the specific condition.
These risk adjustment factors can be thought of as additive scaling factors that adjust the budget. If a provider treats a riskier patient, they have more money in the budget to care for that more complex patient.
AI: Why should a payer make adjustments for risk in their bundled or episode payment program?
SK: Risk adjustment normalizes the playing field among providers. Payers need to be responsive to the reality that providers face when an actual patient walks through their door. Some providers have higher costs because they treat a more complex patient population. If those higher costs are justifiable based on the risk of the patients they are treating, it makes sense to offer a higher budget to that provider.
AI: In a previous post we touched on risk adjustment when using BPCI and PROMETHEUS models. What are some general approaches payers might take to measure risk across different types of episodes?
SK: All risk adjustment models identify risk factors, assign those factors a weight based on how they impact the cost of care, then crunch the numbers to identify a payment amount. Risk factors can be identified with a patient questionnaire or health survey and also through claims history. Risk factors are generally additive – more risk factors usually increase the bundled payment amount.
The critical question is whether risk factors identified in a sample are applied to a specific population. To that end, it is important to identify risk factors using the same, or a very similar, population that will be involved in the bundled payment program. For example, if a payer is contracting with a specific hospital, Aver would want to run that hospital’s own data to show their own history of the impact of risk factors on the cost of care provided by their facility.
AI: How can payers and their network providers be assured that any risk adjustment methodology is “sound”?
SK: This is a real challenge, because risk adjustment is not standardized across the industry. Based on what we’re seeing in the industry now, risk is primarily driven by historical claims diagnosis data available from payers. This means we don’t have access to many risk factors that impact cost of care, such as clinical factors available in medical records. In the future, to improve risk adjustment accuracy, we will need to marry claims data with clinical factors. For example, it is incredibly rare to have a diagnosis of smoking or obesity appear on a claim because it is not what is being treated clinically, so it is not documented.
In the future, I think we’ll have some sort of a national certifying body that endorses risk adjustment methods, similar to the way the National Quality Forum endorses quality measures. As we see continued growth of value-based care and episodic payments, there will have to be some sort of agreement on a common language and understanding of how we apply these concepts.
AI: So, should payers stay away from risk adjustment for now?
SK: No, I think the industry is onboard with risk adjustment and it is a core offering of Aver’s solutions. Payers see it as a great tool in driving conversations with providers about taking on risk. Sometimes, when payers open this discussion, providers’ first reaction is to say, “but our patients are riskier.” With risk adjustment, payers can acknowledge that some patients are riskier, but continue moving to value-based care. It may not be that every contract is risk-adjusted, but it is absolutely vital as part of the conversation.
AI: How does Aver help payers operate risk adjustment?
SK: Aver supports our payer clients’ use of risk adjustment in a number of different ways. Some clients use an internal risk score, usually not specific to bundled payments, and will send that to Aver as part of their data feed. We can make that available as part of their analysis. For clients using the PROMETHEUS payment model, we run that risk adjustment package, incorporating any elements of the episode that the client has customized.
AI: What are the data requirements to get meaningful risk adjustment results?
SK: We typically want to have two to three years of historical claims data from a payer. It’s important to have longitudinal data, but we also need a significant volume of patients and episodes. To get a good fit, there is a base of 25-35 episodes that is the absolute minimum to get a reasonable level of statistical significance. Even then, if there is a known risk factor that doesn’t seem to fit statistically as predicted, most risk adjustment models will eliminate it.
AI: What are some other ways Aver is working with risk adjustment?
SK: We’ve been talking about an infinite spectrum of risk. At Aver, we think that on a practical or operational basis, this infinite risk spectrum can be challenging to manage, for payers as well as providers. Providers want and need to know the budget to which they are tracking each patient. We’re seeing increasing interest in putting some parameters around this infinite risk by grouping patients into different pools or tranches based on their risk scores. For example, we might identify four different risk levels and pool each patient into one of those risk levels for provider budgetary purposes. This helps providers better predict their payment rates and establish care processes based on those risk levels.
AI: Thanks so much, Steve, for sharing your insights on risk adjustment with us.
We hope you’ve enjoyed this series on episode design, and encourage you to sign up to have all our posts delivered to your inbox. Next week, we’re launching a series of posts on pricing episode payments - stay tuned!
Want fresh insights on Value-Based Care delivered right to your Inbox? Become a member of the Aver Insights Community.