Should health plans expect more from their subrogation efforts? It’s worth a look!

Health plans have been relying on Subrogation for decades to recover healthcare claim payments that are a third party’s responsibility.  It is typically regarded as a highly manual and time-intensive process that relies heavily on member contact to verify accident and coverage details. It is not often thought of as a center of innovation.

But recent years have seen some health plans and subrogation vendors experiment with information technology and analytics to help improve the identification of claims with subrogation potential, reduce member abrasion across the subrogation process, and improve settlement rates. I recently presented a webinar in which I discussed ways that analytics and technology are heightening our expectations across three core steps of the post-payment healthcare subrogation process: case identification, investigation and resolution, and recovery. Should health plans expect more from their subrogation efforts? We think so and here’s why.

1. Subrogation case identification

Identifying cases with subrogation potential is a delicate balancing act. If too many cases are opened, the result will be excessive outreach to providers (e.g., for medical records) or members (e.g., for information about an accident that could be subrogatable).

The more a health plan or its representatives reach out to providers and members for cases that don’t ultimately generate value (often referred to as false positives), the more those communities get frustrated with the health plan. Additionally, this creates inefficiency, costing the health plan time and money and generating no value from it. On the flip side, if too few cases are opened, then recovery opportunities are lost.

Let’s face it—we’ll never have all the information we need to make a perfect decision about which cases to pursue for subrogation. But we do need to explore ways to gather as much information as we can to make better decisions without irritating our important constituencies or leaving money on the table.

How can we fill in the picture? By leveraging more of the data that is available today and using analytic models, we can rely less on member outreach and manual inspection, while automating and speeding up some of the decision processes. For example, what can social media tell you about your members?  What can you glean from external property and casualty databases? Can you build business rules based on past experiences and observations to generate analytics that more accurately identify cases? Can you improve these models over time as you feed back results from earlier efforts?

2. Subrogation investigation and resolution

Again, this is a typically manual process requiring outreach to providers and members for information about the case. It’s an area ripe for inefficiency and member and provider abrasion. Within this step, we have identified opportunities across four areas that could result in better results with less waste and abrasion.

  • Outreach modality: Modality refers to the optimal outreach methods for patients and providers. Would you expect that a retired, Medicare Advantage plan member in their 60s would be more likely to answer the phone in the middle of the day than a 25-year-old who is likely to be at work? Would you expect traditional outreach methods such as letters and phone calls to work as well for a younger generation fixated on texts and email?These are simple examples of how we can build models that identify the best way—and potentially the best time—to reach out to different types of members. We can incorporate these and other measures into models that help determine the best path to reach members and achieve a response.Modality in an important concept because the member response rate is key to being able to work these cases and bring them to resolution in a timely fashion.
  • Natural language processing for automated document review: Incredible advances have been made in text analytics and natural language processing (NLP), which allow us to read, comprehend, and analyze incoming correspondence (including incoming medical records) and limit the passing of that info onto your staff only when the analytics show there actually may be savings here.
  • Work prioritization: You have an inventory of cases that need to be worked—how do you decide what comes first? Traditional wisdom says to prioritize the biggest cases and the oldest cases. Today, we have the ability to build models that look at the pool of inventory each day and make that determination based on more sophisticated observations in the context of that specific inventory. We may decide to look at when a case is going to court and prioritize that differently only as the court date draws near.
  • Work assignment: Looking at an inventory of cases, we have to decide who gets which cases to work. Individual experiences cause people to have different performance on the same case. There is an opportunity to look at the collective history of employees to determine their strengths and weaknesses and then make sure they are assigned as much of what they are good at as possible.

3. Subrogation recovery

Here, we look to optimize the same four areas that we did in the investigation and resolution step. The same types of analytics that drive investigation and resolution are applicable for the recovery work, where we also need to figure out how to optimize how we assign work and in what order. As an example, we may still be doing outreach to the hospital during recovery except that it’s a different part of the hospital and we’re seeking different information—financial versus clinical. We still have to figure out the best approach and time to contact.

This applies across prioritizing and assigning work as well. The work may be slightly different, but ultimately we should be able to leverage the work done on investigative models for what we’re doing in recovery.

For most organizations, leveraging analytics to drive improvements in payment integrity is more of an evolution that a revolution. I suggest starting small with a very specific problem that you believe analytic models can help you solve. From there, learn from your failures and build upon your successes.

For more information about how to get started on building analytics into your organization, read our blog post, “5 key roles your company needs for data analytics success.”

 

 

 

Steve ForcashShould health plans expect more from their subrogation efforts? It’s worth a look!

Related Posts