Reducing clinical variance is a key part of a successful pay-for-performance program that has helped one Montana medical center beat the inflation rate-while improving care for patients.

In 2007, Kalispell Regional Medical Center (KRMC), Kalispell, Mont., launched a pay-for-performance program with Blue Cross Blue Shield of Montana (BCBSMT). The program promised shared cost savings for the hospital and financial incentives to physicians if certain quality and cost goals were met. Built upon a clinical quality improvement project that was already underway at KRMC, the program was so successful that it helped reduce everything from morbidity rates to variance and resource consumption in the hospital's most strategic clinical services.

KRMC had a history of collaborating with physicians since its 2003 creation of a joint venture hospital--60 percent owned by the hospital, 40 percent owned by physicians. Hospital leaders believe the joint venture has helped create the collaborative hospital-physician culture required to engage the medical staff for this one-year pay-for-performance project.

This is a sample article from HFMA's subscription newsletter, Healthcare Cost Containment, which helps healthcare financial leaders identify and implement strategies for reducing costs and improving efficiency without harming quality.

Learn more and subscribe to Healthcare Cost Containment.  

Pay-for-Performance Prep Work

Launching the pay-for-performance project took several years of prep work. One sticky point was getting the hospital's and BCBSMT's claims data to match, says Charlie Pearce, KRMC's chief financial and information officer. As KRMC was the first hospital in Montana to implement a true cost accounting system, administrators took pride in their data and were frustrated when it didn't sync with BCBSMT's data, Pearce says. "If you want to take full advantage of a pay-for-performance opportunity with carriers, you have to have better data than the carrier does so you understand what is happening to those patients and can change practice patterns to create more efficient patient care," Pearce says.

"Even though we didn't take on risk in that contract with Blue Cross, we needed to be able to validate their data with our system and define the baseline. That is what seemed to take forever during our negotiations," Pearce says.

In some cases, KRMC and the payer looked at data differences patient by patient. "Some of the differences were because their patients were on an older Medicare Grouper than we were on, which made a difference in how DRGs were assigned," he says. This became less of an issue once BCBSMT converted to a new financial system.

Another thorny issue was defining which type of patient would be counted within the pay-for-performance parameters. KRMC initially thought all Blue Cross patients would be considered. But the payer only wanted to include fully insured, in-state, BCBSMT patients-not those for which Blue Cross was just a third-party administrator. "The box we could put around patients to include in the pay-for-performance program was smaller than we thought, and that took some time to understand," Pearce says.

5 Key Measures

Under the pay-for-performance project, BCBSMT agreed to reward the hospital if its actual costs fell below anticipated costs. The rewards would be based on five key metrics of cost and quality at KRMC:

  • Resource consumption, based on a baseline determined from four years of total hospital charge inflation rates
  • Reductions in variation, based on the 10 most resource-intensive DRGs, which were trended over a three-year period
  • Severity-adjusted morbidity rates, based on 38 complications as well as readmission rates
  • Severity-adjusted mortality rates, which includes nine risk-adjusted inpatient mortality rates
  • Performance on national hospital quality measures, which includes 34 metrics that most acute care hospitals are required to submit to Medicare

Bottom-Line Results

At the end of the project, resource consumption and variance were reduced in KRMC's top clinical services, including orthopedics, cardiology, and obstetrics. Morbidity improved 26 percent, and the readmissions rate was more than 10 percent lower than the national average. KRMC was also able to curb its 2007 inflation rate to 2.8 percent, down from 14.3 percent just six years earlier.

The initiative also demonstrated the axiom that improving medical quality conserves resources. KRMC's increased efficiencies, when compared with five other acute care hospitals in Montana, saved employers and patients about $4.6 million per year for five years. The hospital also conserved $1 million in Medicaid costs and $4.8 million in Medicare costs each year for five years.

The BCBSMT resources were conserved by the physicians as shared savings, which translated to more than $44,000 in physician bonuses. Qualifying physicians received checks ranging from $505 to more than $4,000 for their efforts, depending on their specialty group and how much they improved.

The Key: Reducing Variance

Hospital leaders largely credit their efforts at reducing variance for the pay-for-performance project's success. Such positive results are why variance continues to be a main focus of the hospital's efforts to improve quality and reduce costs, Pearce says.

In the process to reduce variance, KRMC works with a consultant to extract data from the uniform hospital discharge data set, measure patient severity and risk-adjust the data, and review mortality and morbidity rates of all clinical services. Charges are used as a surrogate for costs, as well as the number of resources used to care for a patient. The variance data are then presented to each clinical group at KRMC.

Three years worth of data is presented in a four-quadrant graph that plots the differences in charges and length of stay between the most and least efficient physicians. For example, for adult pneumonia, it's not uncommon for a typical hospital to report as much as $40,000 variance per patient case. Different prescribing patterns, such as the use of a more expensive brand-name antibiotic, often account for variances. For heart failure and shock patients, the variance at a typical hospital might be as much as $60,000 per patient.


In addition to improving quality, reducing variance can have a significant impact on the bottom line. This savings can add up, particularly as hospitals move to fixed payment strategies, Pearce says.

After identifying problem areas, physicians at KRMC then implemented best practices to reduce variance. To date, KRMC has had the most success with reducing variance in two DRGs: degenerative nervous system disorders and Cesarean sections without complications or comorbidities.

The problem is that when physicians try to develop clinical pathways "in a vacuum," they tend to focus on the sickest patients, which often drives up utilization and makes variances worse, says William Mohlenbrock, MD, FACS, chairman and CMO at Verras, which assists KRMC with this process. Instead, hospitals should develop clinical pathways that concentrate on a homogenous group of patients, rather than the entire DRG.

Hospitals should create two pathways per DRG: one that addresses less sick patients-those on the variance graphs with acuity index method (AIM) scores of 1, 2, and 3-and another for sicker patients-those shown on the graphs with AIM scores of 4 and 5, Mohlenbrock says.


What's Next

Under the direction of its new chief medical officer, KRMC plans to step up its efforts to reduce variance within physician practices in specific disease states. Some of the DRGs that KRMC will focus on include:

  • 460: Spinal fusion except cervical without a major complication or comorbidity
  • 247: Percutaneous cardiac procedures with drug-eluting stents, without a major complication or comorbidity
  • 329: Major small and large bowel procedures with a major complication or comorbidity

These represent DRGs at KRMC where either variations have not been reduced, variations have increased, or the outcomes have not improved.  

 Interviewed for this article:

Charlie Pearce is chief financial and information officer, Kalispell Regional Medical Center, Kalispell, Mont. (

William Mohlenbrock, MD, FACS, is chairman and CMO, Verras, Las Vegas (  

This article is based in part on a presentation by Velinda Stevens, president and CEO at KRMC; A. Craig Eddy, MD, JD, CMO at KRMC; and William Mohlenbrock, MD, FACS, chairman and CMO at Verras, at the American College of Health Executives' 2011 Congress on Health Leadership, March 22, 2011,

 How to Talk to Physicians Successfully engaging physicians to reduce variance requires:

  • Presenting physicians with risk-adjusted data over a certain time period, say three years
  • Sharing variance data in an educational, non-threatening forum that is directed by fellow clinicians, such as the CMO or other clinical leaders
  • Maintaining anonymity when sharing variance data with a group of physicians
  • Drilling down data to the level where physicians can determine how they are different, such as how many X-rays they order or which antibiotics they prescribe for certain patients
  • Including nurses, pharmacists, and discharge planners in efforts to reduce variance
  • Keeping CFOs behind the scenes most of the time to stay focused on quality, not finance  

Publication Date: Thursday, August 04, 2011