Data-Driven Healthcare That Works: a Physician Group Perspective

My Folder

For us, data-driven healthcare really started back in 1997, on the day our CEO at Crystal Run Healthcare returned to school to get his MBA. That’s when we decided as a practice that we needed both a mission statement and a vision. The answer? To improve the quality, availability of, and satisfaction with healthcare services in the communities we serve. We’d do this with data.

To set the story straight, this wasn’t our first foray into data. Crystal Run was a relatively early adopter of the electronic health record (EHR), but our practice’s vision quickly propelled us forward in the quest for data-driven care in our population health management efforts.

But our practice’s vision came before any of us heard of the Triple Aim. If you compare the Triple Aim today with our 1997 mission statement, you’ll see they align nicely. This worked out well for us at Crystal Run — we had unknowingly started to prepare for the Triple Aim by embracing the need to improve care and making it part of the fabric of our business. And we continue to see success through this approach today. Here’s how we make it all happen.

Population Management Is Critical for Future Viability and Population Health Improvement

Crystal Run has a lot of competition — both within the community with other physicians and also from hospitals and health plans that are buying and employing providers. Plus our coverage area is growing. For us to do what we need to do in the value-based care world — so we’re not just getting paid for doing things but instead getting paid for health outcomes — it’s important to have systems in place that enable population health management.

For anyone in healthcare to be able to manage their patient populations, larger is better. With lower numbers of patients, there’s a greater risk of rare, random events that really put a wrench in the works. Larger populations, however, behave in a more predictable way. That means Crystal Run needs to expand in size, scope, and geography. We also need to evolve and continue to build the skills sets of the members of the team, so Crystal Run becomes the practice of choice, not just for patients, but also for our employers and physicians. To accomplish these goals, here are a few examples of what we’re doing to implement data-driven, value-based care.

  1. Waste reduction. Crystal Run’s strategy for care management is to provide high-quality, value-based care that is data-driven and evidence-based. For example, if procedures and tests have not been shown to benefit the patient, then we try to eliminate them. As a result, clinicians review every test and procedure to ensure it will contribute to advancing the patient’s health.
  2. On-site care managers. Registered nurse care managers are located at each of our medical homes. Their role is to meet with patients and coordinate care as needed. There’s also a care manager in the hospital — paid for on Crystal Run’s dime — who coordinates the transition of care. In addition, a team visits with patients in their homes if they can’t make it to the office.
  3. Identification of high-risk patients. Identifying high-risk patients and those who need complex care management is very important. Being able to identify these patients, however, is only possible with systems in place to evaluate risk and then create lists of patient, also known as registries. Then the patients in those lists that need management are evaluated by providers or sent to care managers.
  4. Bluetooth-enabled devices. If a patient has multiple admissions or is a high-risk patient with certain conditions, they receive a Bluetooth enabled monitoring device, such as a scale, a pulse oximeter, or a blood pressure device, to take home with them. The Bluetooth device sends the patient’s results to a web service that sends it to a nurse if there are problematic changes, so the nurse can reach out to the patient and offer assistance. For example, if a patient with CHF has a sudden increase in their weight, a nurse can reach out and intervene.

Data-Driven Outcomes

We had been relying initially on a data warehouse that we developed ourselves. Applying the strategies mentioned above, we achieved some relatively positive results, including the following:

  • Reduction of admission rates by about 4 percent — and that’s an absolute 4 percent.  So it’s about 20 percent overall from where it was before.
  • Improvement of mammogram screening rates from 60-65 percent to greater than 75 percent.
  • Achievement of hemoglobin A1c rates of less than 9 percent. So that’s very good control.
  • Improvement of pressure control by more than 75 percent for hypertensive patients.
  • Reduction of avoidable admissions down to now less than 17 percent at this point. Overall definitely an improvement.

avoidable admissions

Crystal Run’s avoidable admissions are down to less than 17 percent at this point.

 mammographyCrystal Run improved mammogram screening rates for its patients from 60-65 percent to greater than 75 percent. They achieved these improvements by implementing a data-driven approach during the second quarter of 2012.

Inability to Keep Up with Accelerating Data Demands

While our results were solid, we wanted more. Our organization had depended on data for years, but, recently, we were noticing how it was becoming even more important in our quest to provide the best care for patients, advance our mission, and maintain our quality reporting requirements to outside entities. But to be able to provide these services and achieve the types of outcomes that are required, we wanted to be able to store and track ever-increasing amounts of data in a more automated fashion.

Our analysts had spent a lot of time with manual entry and extraction of the data, and it was simply not scalable. As a result, if we wanted to grow we’d have to keep hiring and hiring. As a lean organization, it’s important to have people really practicing at the top of their license with automated processes wherever possible, and then be able to present the data to physicians and others in the organization. Instead, we have had to decide how badly we need a particular measure versus the value that it provides and how much work it was to produce.

There’s also a need to marry claims data with clinical data from the EHR, so we can improve our knowledge of patient care provided outside our system. In addition, our financial reporting is becoming more and more complex and order tracking is necessary from both within and outside the organization.

So, we knew we needed to have an enterprise warehouse to put all of the data in and to be able to simplify, automate, and scale the reporting. We also needed dedicated analytic applications that would enable analysts with actuarial science and statistical expertise to help us sort through questions that were more difficult and grayer. These types of reports just aren’t possible with a more ad hoc system.

Going forward into the future to ensure our viability, as well as to be able to provide the best data-driven healthcare possible for our patients, Crystal Run decided to find a partner to develop a more modern enterprise data warehouse (EDW) to update the existing data warehouse.

Crystal Run EDW Requirements

We looked at different options because we had several requirements for the EDW: 1) The solution needed to hit the ground running and give us quick results, reportable data, and early returns; 2) We didn’t want to spend a year creating the perfect data warehouse after which we might start getting actionable data out of it; 3) There needed to be a library of measures of analytical applications that we could apply as needed; 4) The healthcare data model needed to be something we could see and evolve over time, not a “black box;” 5) We wanted to be taught how to fish for the data ourselves and not rely on running back to the vendor every time there was a need to make a change; 6) It was important to have a long-term relationship with our vendor. We partnered with Health Catalyst and have implemented a Late-Binding™ Data Warehouse.

 It’s All About the Data

The nature and sheer volume of healthcare data is much more complex in healthcare than in other industries. These days there’s a sense of urgency for physician groups like Crystal Run, payers, and providers to get access to their data to provide data-driven care. As a result, we need to take over ownership of managing, adding, and expanding our data sources. Yet when we’re dependent on a vendor to make changes, our ability to scale and find much-needed answers to our questions when we need them is limited.

The flexibility of the Health Catalyst architecture and the adaptable approach to mapping data to business rules and definitions when it’s most appropriate has given us a solution that can keep up with the complexity and rapid changes in healthcare data. From our perspective, data-driven healthcare provides the much-needed insight Crystal Run needs to achieve our mission as well as the goals of the Triple Aim.

Has your health system seen results using data-driven healthcare? If so, what tools and initiatives have helped you achieve your goals?

Powerpoint Slides

Would you like to use or share these concepts?  Download this Why Healthcare Data Warehouses Fail presentation highlighting the key main points.

Click Here to Download the Slides 

Loading next article...