The Power of a Shared Vision for Data-Driven Improvement: Reflections on our 10-year Partnership with Allina Health

By Dan Burton, CEO, Health Catalyst

For the past decade, it has been our honor at Health Catalyst to partner with Allina Health, a 13-hospital, 90-clinic healthcare system based in Minneapolis, to pursue the shared overarching objective of data-driven improvement. Through the leadership of extraordinary individuals from Allina, including Penny Wheeler, Duncan Gallagher, Tim Sielaff, Craig Strauss and Jonathan Shoemaker, Allina has systematically improved patient care and patient outcomes, while also providing meaningful and measurable financial returns. We at Health Catalyst have provided technology and expertise, particularly in areas of data, analytics and improvement, to support hundreds of these initiatives over the past ten years. We were thrilled and honored earlier this week when our partnership earned the College of Healthcare Information Management Executives (CHIME) 2018 Collaboration Award.

What follows is a description of the partnership and the improvements that have been achieved through this decade-long collaboration. Health Catalyst's role accounts for only a small part of the success that Allina has realized over the past decade. Multidisciplinary improvement teams and a new project management infrastructure have been essential to Allina’s transformation. Allina has improved outcomes by reducing variations in care and encouraging adherence to best-practice guidelines. Yet this effort would have not scaled without the integrated data and the advanced analytics that have supplied an enabling infrastructure to support the pursuit of healthcare excellence.

The most fundamental contributor to Allina's success, in my opinion, has been the pervasive and consistent dedication to improvement from our colleagues at Allina, exhibited by thousands of their employees, and embodied in their CEO, Penny Wheeler, who continues to be a personal hero of mine, a dear friend, and an extraordinary example of the power of caring, patient-centered leadership.

BACKGROUND

Recognizing that its healthcare delivery was inefficient and that its patient outcomes were suboptimal, Allina Health set out in the 2000s to achieve the Triple Aim of improving quality, lowering cost and enhancing the patient experience. The HCO made quality a board-level priority and changed its approach to quality improvement. In 2008, Allina became the first customer of Health Catalyst, which had been established by a group of data warehouse experts formerly with Intermountain Healthcare in Salt Lake City. Health Catalyst’s main product at the time was a “late-binding” data warehouse that allowed HCOs to create ad hoc reports and pull data together as needed, without having to do costly programming rewrites. Health Catalyst’s data warehouse and its associated analytics provided much-needed visibility into patient care trends and helped Allina jump-start its clinical improvement program.

THE PARTNERSHIP

Under their expanded 2015 partnership agreement, Allina partnered with Health Catalyst to provide its data warehousing, analytics and performance improvement technology and content, along with dedicated personnel, to accelerate the health system’s transformation. Dozens of Allina employees working in these areas became Health Catalyst team members working onsite at Allina.

Simultaneously, Allina gained access to Health Catalyst’s full technology, content and deployment expertise to help Allina achieve faster improvements in outcomes at lower cost. The partnership’s executive governance committee was charged with identifying a prioritized list of improvement projects each year. These projects are designed to provide measurable care improvement and financial value to Allina, and the partners share in the economic benefits of that success.

The technologies to which Allina has gained access include Health Catalyst’s Data Operating System (DOS). This is a software engineering approach that combines the features of data warehousing, clinical data repositories and health information exchanges in a single technology platform. Among the attributes of DOS are reusable clinical and business logic, near-real-time streaming data, the integration of structured and unstructured data, the ability to funnel insights back into EHR workflows, a microservices architecture to enable the integration of third party applications, native machine learning, and a vendor-agnostic data lake.

RESULTS

The combination of Allina’s analytics platform and its project management structure provided the foundation for a variety of clinical, operational and financial improvement projects. For example, in a recent year, these projects saved a total of $125 million in the following areas:

·     Productivity improvements – $43 million

·     Clinical documentation – $14 million

·     Site specific initiatives – $25 million

·     Care integration – $10 million

·     Length of stay – $3 million

·     Reducing clinical variations – $30 million

The reduction in care variations resulted from a range of improvement projects, including

·     Lowering heart failure readmissions

·     Lowering all preventable readmissions

·     Improving stroke care 

·     Improvement the prevention and treatment of VTE

·     Improving outcomes for spinal conditions

·     Improving cancer outcomes 

The continuing decline in care variations accounted for between 30% and 50% of the $100 million that Allina saved in 2015 as a result of its quality improvement program. For example, data analysis showed that 14% of the time, Allina’s ob/gyns were inducing labor before 39 weeks of pregnancy, which increased labor time and the risk of newborns going to neonatal ICUs. By making the doctors aware this was happening and persuading them to agree not to allow it, Allina improved outcomes and saved a great deal of money at the same time.

What follows are some brief case studies on how Allina, in collaboration with Health Catalyst, has improved its performance in key areas of clinical care.

CONGESTIVE HEART FAILURE

Congestive heart failure (CHF), apart from its high prevalence, morbidity, mortality and cost, represents a major financial challenge for many hospitals. Even if health systems do not hold value-based contracts, they can lose a significant piece of their Medicare reimbursement because of high readmission rates. 

When Allina decided to improve CHF care, its average 30-day readmission rate for CHF patients was running above 20%. Its quality improvement managers believed that better care coordination could bring down this rate. But because CHF patients had so many different caregivers, it was difficult to coordinate their care. In addition, hospital, post-acute and ambulatory care functioned in separate data silos. 

The cardiologists in one Allina hospital organized a multidisciplinary improvement team that put processes in place to ensure that the care of CHF patients was coordinated. As a result, readmissions among this group dropped from 24% to 14%. 

Allina later rolled out a systemwide CHF management program, supported by Health Catalyst’s data analytics platform. The late-binding data warehouse pulled data from across the enterprise, aggregating and normalizing it in near real time. The analytics platform supplied information to CHF analytics dashboards that enabled care providers to monitor, analyze and manage diverse CHF cases and measure their own performance. In addition, Allina assigned nurse care coordinators to go into hospitals, explain care plans to patients and make sure they were implemented after discharge.

As a result of this care coordination program, Allina’s 30-day CHF readmission rates in its metropolitan hospitals dropped from 19.7% to 16.7%. Moreover, the percentage of CHF patients who had doctor appointments within five days after discharge jumped from 39.8% to 58.2%.

INTEGRATED SPINE CARE

In 2010, Allina noticed that public quality reports gave a poor grade to the spine procedures performed at some of its facilities. Moreover, the data in Allina’s data warehouse showed substantial variations in the outcomes of spine surgery, including length of stay, post-operative complication rates, readmissions and costs. However, physicians lacked visibility into these variations and their own performance compared to that of their peers.

Allina launched a spine surgery improvement initiative. The project started with the formation of multidisciplinary teams, including surgeons, physiatrists, nurses, physical therapists, other medical spine specialists and social workers. These teams identified areas of focus to improve care, following evidence-based, coordinated models. To reduce variations in care, Allina adopted standardized guidelines and educated clinicians about the importance of sticking with them whenever possible.

In addition, Allina used its data warehouse and the Health Catalyst Data Operating System to aggregate, normalize and analyze data from across the continuum of care. The HCO utilized a specialized dashboard on spine surgery and lower back pain to feed back performance data to providers. Physicians were shown their data relative to their peers to help inform them and engage them in the improvement initiative.

The key results included a 16% reduction in length of stay, from 4.26 days in 2012 to 3.5 days in the first quarter of 2015. The percentage of patients with post-operative complications declined from 9% in 2014 to 5.8% in 2015. And by persuading its surgeons to standardize supplies, Allina saved an estimated $2.7 million.

The availability of cost and outcomes data played an important role in driving consensus on this decision. In addition, Allina’s data-driven culture has led to the standardization of the care model for spine surgery. At Twin Cities Spine Center, for example, surgeons developed a “fusion risk score” that provides a simple, reliable method of assessing perioperative risk. Also, by adopting consistent standards in spine care, Allina physicians have changed their approach to clinical decision making. As a result, more patients with lower back pain now receive medical care instead of surgery.

VTE CARE AND PREVENTION

Venous thromboembolism (VTE) is a common cause of hospital-related deaths. To reduce these deaths in its facilities, Allina embarked on a three-year journey to prevent VTE and improve the care of patients who had this condition. 

At the outset, Allina had no systemwide standard practices in place to provide VTE prophylaxis or VTE follow-up care for the general inpatient population. In addition, Allina discovered that risk assessments for VTE existed in several forms and that clinicians did not apply them consistently across patients. Allina also lacked the high-quality data it needed to improve VTE care and meet CMS core measures. And when patients developed VTE, the system had no way to track whether best practices were being used in follow-up care.

The first step in Allina’s improvement initiative was to establish a multidisciplinary, physician-sponsored VTE workgroup, which included doctors, nurses, pharmacists, and health IT experts. It also created a standard VTE risk-assessment tool and an order group that incorporated both general admission and postoperative order sets. The HCO mandated the provision of VTE prophylaxis and, if it was not provided, required clinicians to document why. It also improved compliance with the educational requirements in discharge instructions for VTE patients discharged on warfarin. And it improved care bundle performance for VTE.

As in the CHF and spine surgery projects, the health system implemented a VTE analytics dashboard to make the data in its data warehouse actionable at the point of care. This dashboard enabled the VTE improvement team to identify target areas for improvement; it also  guided their decisions in selecting and evaluating VTE prophylaxis and treatment tools. The dashboard also helped the team drive behavior change, and it served as a case-finding tool, enabling the team to identify high-risk patients who needed more focused interventions.

These efforts resulted in a significant improvement in VTE prevention and care, as follows:

·     Increased VTE bundle compliance rate by 19%

·     Integrated systemwide approved VTE order group in 308 order sets

·     Achieved 97.9% compliance with VTE prophylaxis

·     Increased compliance with VTE warfarin therapy discharge instructions by 63%.

CONCLUSION

None of these quality improvement projects are simple or easy. They require the ability to aggregate and analyze data from many different sources; to deliver actionable insights into the clinical workflow; and to systemically work to change behaviors based on the data and analytics, requiring dedication, commitment and patience. Our colleagues at Allina are extraordinary examples of this level of dedication and determination. We feel honored to be their partner, and look forward to a continued shared dedication to improvement in the decade ahead!