How Allina Health Used Data to Improve Quality and Reduce Cost

April 28, 2017 12:15 pm

A Minneapolis-based health system’s data-driven cost and quality improvements realized savings of $125 million in the first year.

The staff member, who worked at an Allina Health facility in Minneapolis, remarked that the hospital had experienced “a really good year” after a flu outbreak caused a flurry of patient activity. The fundamental flaw in the logic is obvious, but the perception that people falling ill could be a good thing reflects the challenges the industry faces as it transitions to a value-based system while remaining largely dependent on volume-based payment. For Allina’s leaders, the remark brought to mind the question “What is a health system’s primary mission in today’s fast-evolving environment?”

Allina’s executives determined that the best way to answer that question would be to use the health system’s data to identify problem areas, help determine solutions, and measure outcomes. In the first year after implementing a data-driven process improvement model, the health system achieved $125 million in savings on the health system’s overall annual costs amounting to $3.8 billion prior to the implementation, with about $100 million in savings obtained in subsequent years—all while improving patient outcomes.

Focus on the Triple Aim

In any industry, improving performance and accountability requires a shared goal that unites the interests and activities of all stakeholders. For Allina Health, a health system serving Minnesota and western Wisconsin, with 13-hospitals, 90 clinics, 5,000 employed and affiliated physicians, and 27,000 employees, that goal is to deliver high value to its patients in every service. To focus its efforts, the health system adopted the definition of value used by the Institute for Healthcare Improvement (IHI) in describing its Triple Aim—improving the health of populations while also improving patient experience and lowering the per capita cost of care.

Allina’s leaders recognized that delivering this value to patients and the community would require a realignment of the health system’s strategies, organizational structures, and management practices. It also would require a sharp shift in mindset: The health system needed to look at health care as a data-driven science rather than a proficiency-based art, using data-driven insights to deliver better outcomes, reduced variation, fewer readmissions, lower infection rates, and fewer medical errors.

A Data-Driven Improvement Strategy

Timely, accurate, and reliable data can support delivery of the best possible care at a lower cost in a healthcare environment with coexisting volume- and value-based payment models. The approach depends on deployment of a sound data infrastructure such as an enterprise data warehouse (EDW) and associated analytics applications. The EDW gathers data from virtually any IT system inside or outside of the health system and makes it available to analytics applications. Those applications, in turn, provide data in consumable form to clinicians and others at the point of decision, enabling them to focus on the right opportunities for improvement.

But data and technology must work in conjunction with a permanent project management structure to identify, oversee, and manage improvement projects. This services arm of the improvement strategy requires each project to have an improvement plan with clearly defined target outcomes and cost reduction goals for a specified period, as well as metrics to ensure visibility into performance.

Allina also developed a 10-step quality improvement process model that examines the current state, identifies opportunities for improvement, implements action steps, and measures the results. The process includes the following steps:

  • Determine what needs to be accomplished.
  • Identify stakeholders.
  • Assess the current process.
  • Set goals for process improvement.
  • Identify root causes and barriers to improvement.
  • Develop an improvement plan.
  • Test the improvement plan.
  • Monitor results and redesign tests.
  • Standardize the process.
  • Capture lessons learned.

Using that process, Allina teams identified and undertook several important clinical, operational, and financial quality improvement efforts focused on improving productivity, eliminating inappropriate clinical variation, reducing length of stay (LOS), and enhancing care integration.

The combination of Allina’s analytics platform and project management structure drove improvement projects focused on productivity improvements, clinical documentation, reduction of clinical variation, site-specific initiatives, care integration, and LOS that achieved a better experience for individuals, better health for populations, and lower per capita costs.

Reducing Readmissions for Heart Failure

Congestive heart failure sends more adults over 65 in the United States to the hospital than any other cause. The cost to Medicare alone for treating hospitalized patients with heart failure exceeds $17 billion annually, with readmissions significantly contributing to the issue. Not surprisingly, the Centers for Medicare & Medicaid Services (CMS) considers reducing readmissions from heart failure a prime area for pay-for-performance initiatives.

When CMS announced its Hospital Readmissions Reduction Program (which reduces payments to hospitals with excess readmissions) in 2009, Allina’s own 30-day readmission rates were generally running well above 20 percent. Allina’s leaders believed this rate could be improved with more coordinated care, but like most health systems structured to operate in a fee-for-service world, Allina was not well organized for coordinated care delivery. For example, although well monitored as inpatients, discharged patients were largely on their own until they were either readmitted or seen in a clinic.

Shortly after CMS announced its focus on readmissions, Allina’s Mercy Hospital in Coon Rapids, Minn., launched a multiyear effort to improve the care for patients with heart failure and reduce 30-day readmissions. As a result, readmissions at the hospital ultimately dropped from 24.2 percent in 2008 to 14.3 percent in 2013. Encouraged by Mercy’s success, Allina rolled out a systemwide heart failure management program, supported by the health system’s EDW and heart failure analytics dashboards, which would enable Allina providers to monitor, analyze, and manage a diverse set of metrics.

Key people and processes. The heart failure management program is designed to overcome persistent challenges with care coordination—particularly a lack of a clear ownership of the care management process. The program focuses on five major functional areas: nursing, care coordination, protocols and guidelines, measurement and reporting, and education. A multidisciplinary team was established for each functional area. The team was led by a cardiologist, and the committees focused on protocols and guidelines and on measurement and reporting were led by other cardiologists. The care coordination committee was co-led by a cardiologist and a primary care physician. All committees were led by roles close to the work, and membership included representation of all affected disciplines. Care is supported by the multidisciplinary teams, which include specialists, nurses, therapists, hospitalists, rehab and palliative care providers, and others, all of whom are tasked to follow evidence-based guidelines provided by national societies, such as the American College of Cardiology.

The program’s overarching goals are to ensure a consistent approach is used among all clinics and hospitals, to integrate appropriate services such as palliative and rehab care into the care process, and to coordinate care across the continuum. To help achieve these goals, the interactive analytics dashboards help the program’s multidisciplinary teams monitor and manage performance. For example, the leaders can track the percentage of patients coded with primary heart failure who were managed by a heart failure care coordinator while the patients were in the hospital or during the 30 days after discharge, or both. Allina also can track the percentage of patients who were seen by a cardiologist or primary care physician within five days after discharge. (To view online sample analytics dashboards, go to

Having easy access to the data also helps to drive behavior change and acts as a case-finding tool. High-risk patients are identified and monitored to ensure they receive optimal care coordination and follow-up. Care coordinators can review their high-risk patients, looking for trends in key outcomes indicator such as the number of hospital visits, readmissions, and, LOS.

The data also help in monitoring the impact of care coordination, and in supporting end-of-life management. Again, the interactive care coordination dashboard monitors the extent to which patients receive care coordination during their hospital stays and 30 days after discharge, and the care coordination data are correlated with key measures associated with the Triple Aim.

Results. The heart failure care coordination program has enabled Allina to reduce heart failure readmission rates in its metro hospitals from 19.7 percent to 16.7 percent. Moreover, as a result of meaningful improvements in identifying heart failure patients who would benefit from care coordination efforts, Allina has seen an increase in the percentage of post-hospital follow-up clinic appointments made within five days from 39.8 percent to 58.2 percent.

Another example of a successful data-driven approach to quality and performance improvement is Allina’s effort to reduce LOS. As one of the largest healthcare systems in the Upper Midwest, serving 41 communities, Allina would realize significant cost savings by reducing LOS, but it wasn’t until 2009 that Allina had sufficient analytic and technical capabilities to be able to embark on its in initial four-year performance improvement effort.

Before adopting a data-driven approach to reducing LOS, Allina had no effective means of identifying meaningful opportunities for improvement across the organization, adjusting clinical decision making during patient hospitalization (because of insufficient performance data), or estimating the financial impact of LOS opportunities.

Traditionally, LOS was calculated and reported by the finance team in whole days, from admission date to discharge date. However, this method did not consider that patients enter and leave the hospital at different times of the day. In any given day, whether a patient was discharged at 11 a.m. or at 10 p.m., LOS would be calculated as one day, even though an 11 a.m. discharge would open up hospital bed capacity for 11 additional hours. Use of the traditional method for calculating LOS meant that Allina could neither understand true LOS performance and variation nor identify meaningful opportunities for improvement.

On the clinical side, Allina’s providers had limited ability to adjust care to fit the needs of individual patients because they lacked key LOS information. The process for gathering and sharing LOS data was resource intensive and difficult to disseminate across the organization. When LOS data were available, it often arrived several weeks or months after discharge when it was too late to affect patient care.

Using the EDW and analytics platform, Allina deployed four key changes over one year to address the shortcomings of previous approaches to reducing LOS.

Accessing data. First, Allina developed an inpatient data “mart”—a subset of the data warehouse focused only on inpatient data—so relevant data would always be available for analysis. The inpatient data mart holds individual patient information from the electronic health record (EHR), including data on demographics, diagnoses, timestamps throughout care processes and pathways, and billing information. A new analytics dashboard gave providers and improvement team members easy access to clinically actionable data in near real time.

Accurately measuring performance. Allina implemented a new method for measuring performance that adjusted for the severity of a patient’s illness, ensuring the LOS of a very sick patient was not compared with that of a healthier patient.

Identifying meaningful opportunities for improvement. Allina created performance reports for individual sites, provider groups and individual providers, clinical service lines, nursing units and ancillary departments, quality departments, and leadership across the organization. Through the analytic capabilities made possible by the EDW’s inpatient data mart, Allina identified several opportunities for improvement, including discharge orders, practice pattern variation, and delayed discharge.

For example, through its performance reports, Allina discovered that hospitalists were writing their discharge orders first thing in the morning, but patients were rarely discharged before noon. The performance reports uncovered multiple reasons for this disconnect, including inconsistent or ineffective communication on the day of discharge among the hospitalists, nurses, social workers, and transportation services, and providers’ lack of prior knowledge of each patient’s anticipated discharge date.

Further, performance reports confirmed significant variations in practice among hospitalists across all clinical conditions. The ability to view specific performance data for individual care providers motivated the providers to strive for improved LOS performance.

Performance reports also disclosed that delayed discharges were more common on weekends when the availability of diagnostic procedures was limited and skilled nursing facilities were slower in accepting new patients. The reports also exposed daily delays in ordering medical equipment for patients’ homes, confirming rides for patients, and similar easily-corrected issues.

Understanding the financial impact of LOS improvements. The fourth key change in Allina’s approach to lowering LOS was quantifying the financial benefit of a hospital day saved. The most expensive portions of a hospital stay are the admission and discharge processes, regardless of LOS. With that in mind, Allina sought to quantify the overall cost of saving a day of hospitalization by analyzing the expenses associated with the third day of a four-day hospital stay in its medical or surgical units, as these expenses were most representative of the savings expected from reducing LOS. Using calculations looking only at direct supplies and partial labor, Allina arrived at a conservative estimate of $500 per day in cost savings, enabling the health system to quantify both opportunities and impact.

Improving LOS Performance

As a first step in its LOS performance improvement program, leaders engaged the health system’s hospitalists through frequent reporting of LOS performance data by provider. Making these data public led to a clinical transformation from a reduced variation in practice patterns, which in turn contributed to reducing LOS.

Allina’s hospitalists partnered with members of the LOS improvement team to develop a consistent discharge planning process and established communication expectations with all team members involved in the care of the hospitalists’ patients. Team members are now consistently informed of the anticipated discharge date and when discharge orders are complete.

Again, data have played a key role in Allina’s efforts to reduce LOS. First, the availability of data with high specificity, with the ability to adjust for acuity and applicable for benchmarking, helped leaders obtain provider buy-in. Providers championed efforts with multidisciplinary team members to dramatically improve performance.

Second, the inpatient dashboard created a financially viable means, free from bottlenecks, to get clinically relevant and actionable data into the hands of providers. Allina’s ability to use a dashboard rather than manually collecting, creating, and distributing information has provided the organization with transparent data that do not require additional resources.

As a direct result of its LOS performance improvement program, Allina realized dramatic improvements in LOS in the program’s first two years:

  • More than 26,000 inpatient days saved
  • $13.4 million saved (based on an estimated rate of $500 per day)
  • Hospital capacity (bed availability) created for more than 5,000 admissions, equivalent to adding 90 new beds

People and Data: The Key to Change

Clinical and operational improvements will yield clinically and financially beneficial results as long as efforts are focused on reaching specific goals and accountable to a business plan with measurable benchmarks. These qualifiers require engaged stakeholders and champions, who in turn require access to strong data. Indeed, the very effort to improve the health of communities depends on people acting together supported by such data.

Penny Wheeler, MD, 
is president and CEO of Allina Health, Minneapolis.


googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text1' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text2' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text3' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text4' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text5' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text6' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text7' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-leaderboard' ); } );