Health plan leaders are recognizing that they need new ways to address a member’s health sooner to prevent them from escalating into poor outcomes that drive great financial risk for the plan.

All health plans, regardless of size, location, and ownership, offer chronic care management programs, which identify eligible members from claims data and match them to interventions based on overall risk and specific care gaps. This is according to survey results published in the American Journal of Managed Care in 2015.1 So, it should be safe to assume that after implementing these programs for several years, health plans are hitting the quality and cost goals that they set out to achieve.

Not so fast. Health payers are realizing that this common modus operandi is falling short. In essence, leaders are recognizing that they need new ways to address a member’s health sooner to prevent them from escalating into poor outcomes that drive great financial risk for the plan.

The challenge for health plans is to get to patients faster. How do you get to a member before they become extremely expensive and are facing a negative health outcome? We know claims data has a lag in time, and it’s always a challenge when restricted to that data source. How can we look beyond that traditional model and find ways to identify patients who need help faster? The goal is to move beyond the current use of data to gain additional insights that could, for example, empower payers to provide care management to overweight members before they develop type 2 diabetes or offer shingles vaccines to at-risk members before they develop the disease. More timely insights can drive more personalized engagement, improving health outcomes and patient satisfaction.

To move in this direction, health plans need to push their data analytics efforts beyond the status quo by:

Bringing more crayons into the proverbial box. Health plans have traditionally only had claims data. They are now finding ways to partner with providers to bring more clinical data into the mix.

Data from admission, discharge and transfer (ADT) messages, for instance, can help payers identify when a member has used the emergency department or has been admitted to the hospital. Through this data, health plans can quickly identify members who are frequenting the ED when there are more appropriate settings of care or who are not managing their chronic conditions, for example. These are the members that health plans should target with early interventions.

Health plans also can partner with physicians to access more detailed clinical information that typically resides in electronic health records. Combined with claims, this data can help payers and providers measure and manage clinical quality together more efficiently.

In addition, social determinants of health (SDOH) data, which sheds light on economic and social conditions that influence health status, can help payers as they strive to improve outcomes. Such information is especially helpful when trying to overcome barriers to care management.

If you knew when someone was going in for a knee surgery that they didn’t have much support at home and that they didn’t drive, you would know that they’re going to struggle to get their medication. So, the health plan could connect them to neighborhood services that could help them manage their post-operative care. If you can help a member figure out how they are going to get their medications and how they are going to get to physical therapy before they go in for the procedure, then you can ensure that the care gap does not grow into something bigger and less manageable.

Dealing with the downsides of data. As more data flows into organizations, payers will need to address a myriad of analytic challenges, such as:

  • Spotty information. When working with one hospital or one physician network, you’re getting dribs and drabs of data. If you have too many holes in the data, it becomes very hard to draw good conclusions about the whole population.
  • Generalized data. SDOH data is often drawn from public sources and used to build profiles that are segmented at the zip code level. This data might indicate that someone living in a certain area is not likely to own a car, but the data does not definitely prove that the member doesn’t have a car. While it is possible to collect individual SDOH data, doing so is often labor-intensive and expensive.
  • Disparate data. It is difficult to build an insightful, granular database because data is collected by various entities in various formats. There are so many industry standards. Making sense of the information can be a complicated task.

Building the right models. As health payers move toward predictive analytics, artificial intelligence (AI) and machine learning, they need to build data models that determine what works and what doesn’t with specific patient populations. The models should show which interventions are likely to work with patient populations that have specific clinical and SDOH attributes.

Adopting a pragmatic approach. When implementing advanced analytics, AI and machine learning, it’s important to be very specific, and focus on trying to solve a specific clinical care or business problem. We live in a world of economic scarcity. There’s never enough money to do all the things you could do. So, it’s important to zero in on solving specific problems, be pragmatic about what you are trying to accomplish and drive change in the area that you can actually have an impact. It’s also important to go beyond “reports” and get the analytic insights into the hands of the people who actually take care of members. Because, at the end of the day, that is where analytics can have a real impact.

1 Soeren Mattke, MD, DSc; Aparna Higgins, MA; and Robert Brook, MD, ScD. Results from a National Survey on Chronic Care Management by Health Plans. American Journal of Managed Care, May 2015.

Subscribe to our blog

Don't forget to share this post!