Do you want to be the 'best practice'? Look at your own data first

Use internal benchmarks to reduce variations in care

You can search the world over for best practices, but the greatest examples of excellence and opportunities for change may lie right in your own backyard.

Mount Plant Mease Health Care in Dunedin, FL, is a living example of this precept. The hospital uses internal data to target processes that need to be improved, which often means reducing clinical variation.

The result for 33 DRGs studied: "I can document over $60 million in reduced charges over three years and .5% decrease in the overall mortality rate of those patients," says John Babka, MD, FACP, FACPE, FACHE, vice president of medical affairs. "For some specific DRGs, I can show reduction of about half in mortality rates."

Another case in point: A group of obstetricians at a hospital wanted to lower their cesarean rate from 22% to a nationally recognized "best practice" of 12%. When the physicians examined their individual rates, they were surprised to discover variation ranging from 9.6% to 45%.

The physician with the low rate trained the others in the techniques she used to avoid cesareans. The obstetricians collectively lowered their average rate below 12%.

"Because there was so much internal variation, they rallied around reducing it," says Chip Caldwell, FACHE, senior vice president of Premier Performance Services in Charlotte, NC, who worked with the hospital and physicians. "They realized it was an opportunity."

Reducing variation will increase your efficiency, which means you save money and boost quality at the same time, says Janet Niles, RN, director of utilization and quality management at Carilion Health Plans in Roanoke, VA.

But there is another reason why physicians should become more attuned to this issue: Payers are developing profiles and comparing physician practices. "Physicians may not be looking at their own internal variation, but someone else is," says Niles.

Wouldn't you like to know what they know - and make sure the comparisons are based on valid data?

Your search for internal variation will hinge on your ability to review your current practice. Obtaining good, clean data may be your greatest challenge.

Premier, the nation's largest health care alliance, maintains a national database that encompasses records from more than 250 academic medical centers, community hospitals, and health systems. In addition, it offers another software product, Prospective Plus, that calculates internal variation and national benchmarks. Value View looks for variation in clinical outcomes and service quality across the continuum of care. (See editor's note for contact information, below right.)

Carilion uses the Provider Insight software product from HBO & Co., a health care software firm based in Atlanta, which can adjust the data for severity of illness and other factors and can calculate the number of "unexpected services" rendered to patients.

Morton Plant Mease Health Care plans to install Prospective Plus. But the quality teams have been able to detect and reduce variation in care simply by manually reviewing records, says Babka. Team members may review computer-based claims data, or they may visit physician offices or the emergency room to collect data from a sample of medical records. Because the data aren't used to rate or compare individual physicians, a smaller number of records can be used for analysis, he says.

The health system also has severity-adjusted APR-DRG (all-patient refined DRG) data that can be compared to norms in the state of Florida for length of stay, mortality, and charges. "Almost invariably, before we even start, we're better than the state," says Babka. "We don't look for problems. We look for opportunities."

Look deep for cause of variations

Niles takes a more statistical approach and looks initially for wide variations in cost as a marker of different patterns of care.

Yet that is just a starting point, she says. When you examine your data closely, you may discover that much of your variation is due to improper coding, she says.

Cleaning your data of errors is a vital step. After all, those same coding errors will skew the data that payers are reviewing as they look at variations in practice patterns. You may look like the high cost provider when you're really not.

Data help you identify outliers, but you'll then want to review their cases to better understand the variation. And it should be done in the context of a quality improvement process and in the spirit of educating, not punishing.

That distinction between improvement and judgment is vital to success, says Caldwell.

"Those organizations that use comparisons for learning will become world class," says Caldwell. "Those that view comparative data as judgment will fail. That's the non-delegable role of leadership, to ensure that data are viewed as an opportunity for learning."

[Editor's note: For more information on the Prospective Plus and Value View products, contact Chip Caldwell, Premier Performance Services, 4501 Charlotte Park Dr., Charlotte, NC 28217. Telephone: (704) 679-5062. E-mail:]