Data Analytics: Paths to Value-based Care

It’s becoming increasingly clear that absent analytics, the movement toward value-based care is a nonstarter.

Without data analytics to identify gaps in care, improve efficiencies and tailor treatment plans, healthcare organizations are blind. “We wouldn’t know where to begin without data analytics,” said Bob Sehring, chief ministry services officer at OSF Healthcare in Peoria, Ill. “You can’t begin to build programs without it.”

Large organizations committed to accountable care all have their own approach to health analytics. Some are building internal tools to analyze data within their extensive network and others are outsourcing their analytics operations to vendors or health information exchanges.

As the federal government ramps up its movement toward pay-for-performance, lessons from these approaches are becoming increasingly valuable.

Outsourcing Analytics

Allina Health, a Minnesota health system that encompasses 12 hospitals and about 100 outpatient clinics, is investing big bucks in analytics. But instead of developing tools in-house, the health system announced in early 2015 that it is outsourcing its whole analytics operation—including its data warehousing, analytics and performance improvement technology and content—to the vendor Health Catalyst in a 10-year, $108 million deal.

The move comes as the health system, which covers about 13,000 of its population under its Pioneer accountable care organization (ACO), is trying to drive the market away from fee-for-service to value-based care, says CEO Penny Wheeler.

“The real reason we did this was to learn, quite frankly. We wanted to learn about our performance, our interventions and whether they are making care better and more affordable for our patients,” she says.

Allina had a previous relationship with Health Catalyst, which helped build its data warehouse that not only stored data, but equipped users with advanced tools to make them actionable. “We were growing in terms of dashboards, predictive models and engagement and caregivers. Their evolving tools and services and our evolving clinical engagement drove us back together.”

Shifting Accountability

The agreement is unprecedented as 20 percent of the vendor’s compensation is tied to how well Health Catalyst moves the dial on lowering Allina’s costs and improving quality and outcomes.

That means the organizations are now deeply intertwined. In fact, 60 of Allina’s employees now  work under the Health Catalyst umbrella to help the organization achieve its goals of better care, better health and more affordability, Wheeler says.

A joint governance committee is integral to clinical improvement and outcomes work. Around the table are healthcare executives and representatives from different clinical service lines who determine key processes to measure. “We wanted to really carefully look at key process indicators to make sure all services would be maintained,” she says.

The system has had some success with its analytics tools. For example, it was able to improve its breast exam program to ensure that those with concerning mammogram results received a diagnosis test and biopsy within seven working days. In the area of heart failure, it built a predictive model incorporating hundreds of factors to determine who is at the highest risk of being readmitted, triggering care coordination. That has enabled the system to drop its readmission rate to under 25 percent for this population.

Another example is Allina’s ability to look at elective inductions of labor before 39 weeks. Analytics staff shared data with physicians showing that this took place 14 percent of the time; and that insight spurred them to change their practices—and bringing this number to under 1 percent, she says. As a result, 250 fewer women had C-sections.

“As healthcare is being squeezed, it’s an alternative way to perform in the value-based system we’re evolving in. These are incredible tools for data to make us better, make others better and exist in a financially sustainable way to advance our mission,” says Wheeler.

Analytics at EMHS

Eastern Maine Healthcare Systems (EMHS) is utilizing data analytics resources from both outside and within its organization.

Committed to accountable care, EMHS was able to achieve the triple aim with high patient satisfaction during its first year as a Pioneer ACO, according to Iyad Sabbagh, MD, the organization’s quality officer.

In 2013, EMHS doubled its Pioneer Medicare population, and patient satisfaction and quality continued to be some of the highest in the U.S., although the system did not reduce the overall cost of care. Results are not yet available for year three. “Being a part of the Pioneer program, we were able to offer our Medicare patients quicker options so they could receive the right care, at the right time, in the right place,” he says.

Data analytics played a role in this success, as well as in its larger operation. EMHS utilizes an integrated EMR which allows embedded nurse care coordinators to access real-time data, including test results and diagnoses, so they can more proactively engage patients, he says.

Data also inform care coordinators on whether patients are being treated in the emergency department, walk-in-clinic or hospital so they can intervene appropriately with follow-up calls, medication reconciliation or serve as a resource.

“Risk stratification is the key to our success. When you have access to real-time data, you can make a difference in the lives of patients each day,” Sabbagh says. Risk stratification allows EMHS primary care teams to support patients where they are, which means through education and support of their basic needs so they can work on becoming their healthiest, says Sabbagh.

“Risk stratification tells us who is most at risk and allows us to do something about it before it becomes an emergency situation,” he says. “Having data and understanding them really is the basis for us to personalize and tailor our care to meet the needs of our community. 

Partnering with an HIE

Like Allina Health, EMHS found some benefit from outside partnerships.

The system actively collaborates with the state’s health information exchange, HealthInfoNet (HIN), which contains data from every hospital and most provider practices in Maine. The exchange leverages clinical data to help predict readmission and other indicators for patients.

“They look at clinical data over time, such as utilization of services, emergency department visits and doctor visits, and use an analytics formula to predict risk which helps us move our efforts from reactive to proactive. We are now able to intervene early,” Sabbagh says. “The difference between relying on claims data and HIN means we are better able to care for our patients early in their disease process. This way we can get involved in the beginning of a health condition.”   

When HealthInfoNet unveiled real-time analytics tools early this year, EMHS quickly jumped on board. “Our biggest limitation as an ACO is getting the data quickly and having them standardized across the state. HealthInfoNet is really the industry standard and brings together 100 percent of a patient’s health information from across our state. This means we can support the continuity of care for our patients no matter where they receive care.”

Analytics at OSF Healthcare

OSF Healthcare, an integrated health system that includes nine acute care facilities and two colleges of nursing, is analytics-minded. A Pioneer ACO, the system covered 33,000 Medicare patients in 2014. The ACO did not realize savings in 2012 and 2013, but during the first six months of this past year savings were achieved, according to Sehring.

“We have seen significant improvement in quality metrics over the past three years,” he says, but the development of effective processes and procedures took time. The system had gaps that have since been filled, so now the system has in place a robust infrastructure that supports care management, transitions of care, a skilled nursing facility and medication management.

The system as a whole also has evolved in the area of data analytics.

Over two years ago, OSF Healthcare leadership identified analytics and population health as crucial to transforming its delivery of care. They invested millions of dollars into the creation of a new division of healthcare analytics—which includes data warehouses that contain clinical and claims data, as well as analytics tools and applications.

The healthcare system has harnessed analytics with some striking results. Its predictive readmission model, which generates daily reports for clinicians with scores of the most high-risk patients, has helped bring the all-cause readmission rate to under 10 percent, says Mark Hohulin, senior vice president of Ministry Healthcare Analytics.

The predictive model analyzes 70 variables, including current and past utilization along with social- economic factors. The model produces a list of patients at both medium and high risk of readmission. Of the over 55,000 annual readmissions, the medium- and high-risk patients account for 18 percent of admissions and 50 percent of readmissions, he says.

“It’s not just one thing you have to do different and better. You have to get them all right to move the needle,” says Sehring. But identifying high-risk patients can really have an impact. “It really starts to hardwire activities like follow-up visits with physicians, medication reconciliation and med refills.”

Clinicians embrace the ability to receive timely, focused information on their patients, especially when it’s real-time information. They have developed special, targeted alerts in the EMR so physicians are notified of a significant change in a patient’s status. They strive not to overwhelm clinicians with data, however, Sehring says.

Data analytics helps executives more smartly invest in precious resources, like care management and medical office visits. “We’re really beginning to understand through data and data analytics who we should be focused on. With this better understanding of those at the highest risk, we’re bringing in additional resources to the right population, with the best outcomes and results.”

Through the Clinical Looking Glass

Montefiore Health System, a Bronx, N.Y.-based academic health system that consists of seven hospitals and an extended care facility, a school of nursing and primary and specialty care at 150 locations, is another system investing heavily in analytics. The system is a Pioneer ACO, with 25,000 patients under that Medicare reimbursement umbrella. 

The system has been a frontrunner in health IT for some time, having implemented an inpatient and evolving outpatient EMR since 1997, according to Eran Bellin, MD, vice president of clinical IT research and development.

Montefiore invested in building its own tool, now-patented software called Clinical Looking Glass (CLG). The solution is quite cutting edge, and is guided under the philosophy that analytics should be something that physicians do themselves instead of relying solely on IT staff. “People on the ground working with patients have very valuable insights,” Bellin says.

CLG allows clinicians, medical analysts, quality assurance staff and executives to gain insight into the quality and efficacy of medical treatment. Through it, they can evaluate length of stay, patient characteristics and outcomes associated with hospitalizations for particular conditions and procedures. For clinicians and patients, CLG provides evidence about benefits, risks and results of treatments so that together they can make more informed decisions, according to Bellin.

For example, the system not only shows how many people have diabetes, but allows users to input a start time for tracking patients with certain conditions over a set period of time such as 6 or 12 months. This is valuable to measure the efficacy of an intervention. Unlike most systems, which are geared to looking at an individual patient at one point in time, CLG is tailored towards evaluating the trajectory to outcomes of multiple patients across time, Bellin says.

Having physicians themselves pose questions with the tool, as opposed to writing them down and giving it to the “priesthood,” (i.e., IT staff) allows for rapid-cycle research. It also brings about “the democratization of meaningful access to clinical data,” says Bellin.

So far, Montefiore has trained 800 employees, including internal residents and all new residents in pathology and cardiology fellows, on CLG. The end game is helping physicians look at their cohort and plan how to deliver care to those who continue to be unhealthy.

Use of CLG has had many successes. Utilizing the tool, one part of the organization in south Bronx that delivers care to homeless patients was able to achieve patient-centered medical home level 3 status by taking a longitudinal responsibility to partner with patients, share goals and identify subgroups and actively conduct outreach.

Bellin also cited use by a geriatrician who posed the question, on behalf of a patient, on whether there was a difference in survival rate for those receiving local rather than general anesthesia. “She went into the system and found no difference,” he says.

Another example was asking the system how mortality rates were affected when patients received rehabilitation medicine. “We looked at the population and found a significant difference in mortality.” With reimbursement rates flat due to federal policy, these data can help shape important policy issues, he says.

“When you have the tool, you have many people looking at a problem from many vantage points enhancing your understanding,” he says. As such, a deep level of commitment to transparency is paramount.

“It’s pretty gutsy for an organization to say a doc can look and see if we are treating ulcer patients right. Anyone can do it. You can’t look at names, but you can evaluate a problem,” he says. “We are going to be open with our process, learn where we improve. We’ll find things that are embarrassing, but we’ll become a much better organization.”


The Department of Health and Human Services is pushing accountable care like never before. Its plan revealed early this year to tie 50 percent of Medicare reimbursement to value-based care by 2018 means that organizations will be more incentivized than ever to drive strategies that result in the best, most cost-effective care.

With resources as precious as ever, data analytics is needed to shine a light on gaps in care, inefficiencies and ever-evolving, evidence-based practices. Whether it’s best to look outside for help, like Allina and EMHS, or invest resources of great magnitude within an organization, or both—time can only tell.