Scott L. Teffeteller
Thomas M. Kish

A five-step process for identifying ways to decrease clinical costs led to big gains for one southern Indiana hospital, particularly related to sepsis prevention and treatment.

At a Glance

Hospitals and health systems should undertake the following steps in pinpointing areas for clinical cost reduction:

  • Identify potential areas of opportunity through an analysis of top discharges.
  • Use severity-adjusted data to review variability by case.
  • Review length of stay and resource consumption at a high level.
  • Examine granular charge data and practice patterns.
  • Determine action steps for improvement.

A critical success factor for hospitals in an era of reform will be the ability to aggressively manage costs while providing high-quality care and service.

Declining payment from both government and private payers, juxtaposed with increasing demand for healthcare services from an aging population, requires innovative approaches to providing greater value. But how can hospitals enhance quality of care while reducing clinical costs?

In October 2011, leaders at Union Hospital in Terre Haute, Ind., undertook an initiative to identify areas where clinical costs could be reduced while maintaining or enhancing quality of care for sepsis, a life-threatening blood infection. They knew the initiative would require a focused process with clear action steps that physicians and staff throughout the hospital would both understand and support. Success also would depend on engaged physician leaders who would willingly participate in initiatives that could improve value for consumers and payers while protecting the long-term health of the hospital.

From the outset, Union Hospital's executives understood that physicians would buy into clinical cost improvement initiatives only if the initiatives were supported by data the physicians could trust. An important first step was to identify numerous opportunities to achieve substantial savings through the implementation of physician-directed best practices. Other essential steps included using severity-adjusted data to assess variability by case, analyzing performance at the department level, examining granular charge data, and determining action steps for improvement.

Identify Opportunities for Improvement

Union Hospital's first step was to examine three years of internal data that could help in accurately assessing clinical trends for the hospital and identifying areas for improvement. Use of internal, risk-adjusted data is critical to engaging physicians in such an initiative because physicians believe strongly that a review of clinical metrics and trends within a hospital can be valid only if it takes into account the unique forces at work in the hospital. Union Hospital was therefore convinced that it could best gain physicians' trust and willingness to engage in the initiative by using its own data to identify variation and trends at the hospital, and by basing the analysis on at least three years of data (because year-to-year comparisons can be misleading).

The initial aim was to identify DRGs with the highest level of charges, which are not necessarily those with the largest case volumes. These areas generally provide the greatest opportunities to reduce clinical costs. DRGs are identified by highest level of charges because charges are consistent for all patients and reflect the resources consumed, and charge data are easily derived from patient bills.

The exhibit below shows Union Hospital's top 30 DRGs based on charges for the six-month period ending Dec. 31, 2011. A quick glance shows that DRG 871-Septicemia (sepsis) is the third-leading DRG based on charges, with $7.5 million in charges from 210 patients. Union Hospital was surprised by this finding, because it had previously undertaken successful efforts to improve care for this high-cost condition.

Exhibit 1


In 2007, Union Hospital had been recognized nationally for its efforts to reduce the incidence of septicemia (sepsis). In just one year, the hospital had decreased mortality related to sepsis by more than 10 percent, reduced LOS for patients with sepsis by 2.5 days (with a 1.5-day decrease in the intensive care unit), and reduced ventilator days for sepsis patients by 1.6 days (Chait, J., "Health Matters: City Hospital Garners National Attention," Tribune-Star, Oct. 10, 2007).

Nonetheless, Union Hospital's analysis disclosed that significant opportunities remained to reduce clinical costs for treatment of sepsis and improve quality of care and outcomes related to this condition.

Nationally, the United States spends more than $15 billion annually on treatment of sepsis; it is the single most expensive condition to treat in hospitals, the sixth most common reason for hospitalization, and the leading cause of death in hospitalized patients (Statistical Brief No.122: Septicemia in U.S. Hospitals, 2009, Agency for Healthcare Research and Quality, October 2011).

Exhibit 2


Use Severity Adjusted Data to Assess Variability

Opportunities to significantly reduce costs and improve quality of care depend primarily on reducing variability in clinical practice. But to effectively engage physicians in discussions regarding variation in clinical practices, apples-to-apples comparisons must be presented. The data therefore must be adjusted for severity so that physicians will understand that variation is in no way determined by the severity of their patients' conditions. Union Hospital used a severity-adjustment methodology that subdivides the Medicare severity-adjusted DRGs (MS-DRGs) into five severity levels.

After adjusting the data for severity, Union Hospital identified highest- and lowest-performing cases in terms of cost and LOS. Union Hospital's chief medical officer and other physicians engaged in the process presented the findings to the hospital's physicians using a series of four-quadrant charts. The exhibit on page 82 offers an example of a chart pertaining to current sepsis cases treated at Union Hospital from FY09 to FY11. The numerals in the exhibit (one through five) indicate the severity level of each patient; level five is the highest level of severity. The more efficiently treated cases-that is, cases where both the LOS and charges are lower than the risk-adjusted regional mean-are positioned in the upper right quadrant (shaded in green). Cases positioned within the oval are within two standard deviations of the mean.

As shown in the exhibit, in many of the best- performing cases for treatment of sepsis, charges were as much as $50,000 less than the mean, while LOS was as much as nine days fewer than the mean. Meanwhile, for cases within two standard deviations of the mean (inside the oval), charges were as much as $50,000 higher than the mean, while LOS was as much as 10 days higher than the mean.

Exhibit 3


Data were presented in an educational and non-accusatory manner. After reviewing the data, the physicians acknowledged there was tremendous variation in the treatment of sepsis. They agreed that it made sense to dive into the data to determine the underlying causes of variation.

Analyze Performance at the Department Level

Union Hospital's clinicians, physicians, and leaders examined LOS and charges for sepsis at a high level, using analytical tools to identify the specific resources responsible for the variation. Such variation generally is found at the department level.

For example, exhibits 3 and 4  were used to analyze cost variation and LOS for sepsis. The first of these exhibits depicts significant variation in several departments, including laboratory, pharmacy, the critical care unit, and the intensive care unit (ICU). The best practice bars are the case averages for the best-demonstrated charges of all patients in the green quadrant by service area. The red bars represent the case averages from patients in the left lower quadrant (red), or those who were less efficiently managed by physicians at the hospital.

Exhibit 4


Findings were reviewed with hospital physicians by their peers collaboratively. At this point, the conversations revolved around the reasons for the higher LOS and heavier use of resources.

Examine Granular Charge Data

Next, Union Hospital examined granular charge data to drill down into charges for sepsis cases. This analysis disclosed significant variation between best practices and less efficient practices in areas such as general pharmacy, ICU, oncology room and board, pharmacy-IV solutions, private room and board, and therapeutic services. (A chart showing the findings of a detailed analysis of charges for patients with sepsis in 2011.)

This part of the analysis required strong chart audit capabilities to allow specific practice patterns used in best practice cases to be compared closely with practice patterns from other cases.

The analysis focused on five key questions:

  • What specific tests, drugs, therapies, and other items or services were ordered?
  • In what sequence were they ordered?
  • Which physician or physicians ordered the tests, drugs, and therapies (the primary physician or consultants)?
  • How often were tests, drugs, and therapies ordered?
  • When were patients moved to other units in the hospital (e.g., emergency department [ED], ICU, general floor)?

In performing this analysis for sepsis, Union Hospital found that much of the variation resulted from a failure to identify the sepsis diagnosis early in the hospitalization, so that the patient could receive the appropriate level of care sooner and begin early, aggressive treatment. For many patients with sepsis, aggressive treatment was not begun in time to minimize the chances of developing morbidities and even organ failure. This delay in necessary treatment compromised clinical outcomes, resulting in longer LOS and greater use of resources.

Early, goal-directed therapy for patients with sepsis, featuring many components of the sepsis clinical pathway to be completed in the first hours after patients with sepsis arrive at a facility, helps to promote optimal outcomes.

Reviewing these internal best practices at the granular (order) level assists the hospital in optimizing a more effective sepsis order sets that include guidelines for drawing the appropriate labs and blood cultures, beginning antibiotic therapy, initiating fluid resuscitation, beginning vasopressors, and inserting a central line for monitoring. The hospital's goal for these revised order sets should be 100 percent compliance by physicians and staff.

The findings of Union Hospital's analysis of charts for the hospital's sepsis patients over a three-year period disclosed that the hospital had fallen short of this goal: Physicians and clinicians had not once complied with the current sepsis order sets. Although agreed-upon order sets were in place, they had not been utilized, nor did the ED have sepsis protocols that flowed into the inpatient order sets. For example, established sets called for main central lines to be placed within six hours, but this time frame was not being met.

As hospital leaders and physicians drilled down into the reasons for noncompliance, it became apparent that a gap in physician and staff education existed, as well as numerous organizational deficiencies.

Determine Action Steps for Improvement

Having determined the causes of variation in sepsis care, Union Hospital needed to identify the steps to reduce variation. It also needed to work collaboratively with physicians to implement best practices.

Action steps undertaken by the hospital with the support of physicians included:

  • Putting monitors and criteria in place to track treatment order set compliance, mortality rates, and charges per discharge
  • Implementing ED protocols to complement inpatient order sets
  • Sharing data on outcomes with targeted hospitalists and appropriate physicians
  • Adding physician champions and ED leaders to the hospital's sepsis team
  • Establishing a work group to establish timelines for and ensure availability of central line placement
  • Initiating ongoing education at the physician, clinical team, and hospital operations levels


To date, Union Hospital has pursued many of the opportunities identified during its initiative to clinical costs while maintaining or enhancing quality of care. And it has succeeded in reducing clinical costs in a number of areas-while actually improving quality in most instances. Although Union Hospital's chief medical officer and quality review teams have just implemented the new order sets, process changes that occured during mining and analytics of the internal data have enabled the hospital to achieve:

  • A 1.2-day reduction in LOS
  • A 13 percent decrease in hospital charges for septicemia
  • A 3 percent decrease in mortality rates

The next step will be to educate outlying hospitals whose clinicians send patients to Union Hospital on these best practices, so that they begin these protocols before transferring sepsis patients. Then, the hospital will teach nursing home staff how to identify sepsis early, so that patients with sepsis may be sent to the hospital when they are in the first stages of the disease.

As reforms are implemented throughout the industry, hospitals and health systems will need to become much more efficient in care delivery to both reduce costs and improve outcomes. Identifying severity-adjusted internal data at a granular level, having systems in place to analyze the data to identify opportunities to reduce variance at a physician level, and working with physicians to implement improvements will be critical to protecting the ability of hospitals and health systems to meet the needs of the communities they serve.

Scott L. Teffeteller, FACHE, is president and CEO, Union Hospital, Inc., Terre Haute, Ind. (

Thomas M. Kish, FACHE, is president, Verras, Chicago, and a member of HFMA's Maryland Chapter (


Publication Date: Monday, December 03, 2012

Login Required

If you are an existing member, please log in below. Username and password are required.



Forgot User Name?
Forgot Password?

If you are not an HFMA member and would like to access portions of our content for 30 days, please fill out the following.

First Name:

Last Name:


   Become an HFMA member instead