In-depth data produce results worth having

Benchmark analysis jump-starts best practices

Be forewarned: Physicians insist on reliable, data-driven processes, says Dale Sargent, MD, executive vice president of medical affairs for Wellmont HealthSystem in Bristol, TN.

"That's why you have to have line-item data to be able to identify areas where physicians are overusing resources, as well as hospital charging patterns," he says.

In a major initiative at Bristol Regional Medical Center, Sargent relied on such a data analysis by HCIA Inc. in Baltimore to compare those proverbial apples to apples.

"Yes, we could have figured this out with extensive chart reviews," he explains, "but it would have taken an FTE about three to six months; meanwhile we still would have been continuing in the same direction."

Benchmarking analysis should serve to jump-start, not circumvent, the best practice processes. "It allows you to find out where the issues are, but it does not substitute for the hard work of quality improvement," he says.

Rhodes Moxley, MPA, director of consulting services in HCIA's Carpinteria, CA, office, explains how critical issues can be ferreted out quickly through sophisticated analysis.

First, Bristol's charge description master was "mapped" to the codes in the International Classification of Clinical Services (ICCS), he explains. "Unlike the charge description master of most hospitals, the ICCS codes are hierarchical, with each numerical digit storing specific information about the service," Moxley says. (See Table 1, below.)

Convert data to common measures

For example, Hospital A and Hospital B may have different names for a simple anticonvulsant, so direct comparison in cost and usage would be difficult. However, by converting all charge codes to an ICCS code, the common language enables the line-item comparison physicians love. (See Table 2, p. 66.)

"Every conceivable service a patient could receive while in the hospital is contained in this system," Moxley says. "From the patient billing tapes, we can retrieve the units of service provided and what the patient was charged per unit. This provides the QI team with a detailed picture of the ordering habits of the ordering physicians."

The areas mapped include:

· laboratory;

· imaging;

· pharmacy;

· surgery;

· medical and surgical supplies;

· nursing;

· specialty clinical services;

· nonclinical services.

Mapping enabled the teams to adhere to the first benchmarking basic: Know thyself. For example, they were able to identify and accurately map the crucial components of their charge structure:

· resource mix (which unit of service was used);

· frequency (how often it was used);

· distribution (the percentage of patients who received the unit of service);

· pricing (identifying the charge to the patient for the unit of service);

"That type of data could have been retrieved from chart reviews, but then it would have been very time-consuming," Moxley says.

Each component then was compared with the same component in HCIA's Clinical Pathway Data Base, a proprietary database that represents 6% of all U.S. discharges at the line-item level by diagnosis and procedure.

"By comparing Bristol's charge structure to the best demonstrated practices within the database for the same diagnoses and procedures, the team was able to establish benchmarks," he explains.

The data also were adjusted for differences in patient age, sex, and severity of illness and for hospital region and bed size. Cost-of-living adjustments were made using the Health Care Financing Administration's wage index.

"After you adjust for severity of illness and cost of living differences, what remains is the manageable portion because it reflects the hospital's pricing policies and physicians' resource use," Moxley points out.

For example, the "manageable portion," or opportunity for improvement, for DRG 89 was $982,827. (See Table 3, p. 66.) To further analyze that figure, HCIA identified the subsets within the DRG and determined the average dollar variance per case:

· patients with no substantial or moderate complications or comorbidities;

· patients with major complications or comorbidities;

· all other pneumonia patients.

Based on the fact that the first category contained an estimated $807,023 of the manageable portion, the team decided to select it for further evaluation. Then the team analyzed revenue centers to find out how much each center contributed to the manageable portion. (See Table 4, p. 67.)

Merge clinical, financial data

Next, for each revenue center and associated component, the team calculated a per-case total difference between the peer group and the benchmark. For example, pricing was the most signifi cant component for pharmacy. The category "Fluids/lytes/nutrients" represented an average charge per case of $985, or 40% of total pharmacy charges. (See Table 5, p 67.) The average charge for fluids was 123% higher than the peer group and 148% higher than the best practice.

Finally, the team had to find out if this variation was caused by Bristol's pricing policies or physician resource use. For this particular intravenous solution, the higher average charge was due to both. (See Table 6, p. 67.)

"At this point, it's the merging of clinical and financial data that allows the team to get the level of detail they need," Moxley says.

Such observation is important because it reinforces the need for the physician/administration partnership. "The data show that physicians should not be expected to shoulder the full burden of cost containment. Administration also has to play its part."

(Editor's note: A copy of Bristol Regional Medical Center's pneumonia clinical pathway appears on p. 68.)