Feature Story

Want to trim $1,727,250?
Here’s how one lab did

Diarrhea Algorithm (PDF, 376 KB)
HCV Algorithm (PDF, 376 KB)

June 2004
Karen Southwick

Seemingly small, incremental changes in laboratory procedures can add up to big savings if the changes are followed consistently.

St. Luke’s Regional Laboratories, which serves 300 clients within a 100-mile radius of Kansas City, Mo.—600 physicians, urban and rural hospitals, other labs, nursing homes, and more—discovered 10 years ago that its laboratory tests were being overused.

"We were subscribing to a benchmark report that compared us to a peer group of 35 other teaching hospitals taking care of complex patients," said Fred V. Plapp, MD, PhD, medical director of St. Luke’s Regional Laboratories. St. Luke’s was performing 44 tests on each patient (as measured by inpatient tests per discharge), and its peers were performing about 35. "So we set out to change that," said Dr. Plapp, who shared the specific steps his lab took in a presentation in April at The Dark Report’s 2004 Executive War College.

First, the laboratory staff reviewed existing tests. "We had a high test volume—2 million per year—and an extensive test menu of 300 tests," Dr. Plapp said. "We didn’t think any single project would be very effective."

Through trial and error, St. Luke’s learned that multiple approaches were needed to change test-ordering behavior. This included behavioral efforts directed at physicians, environmental changes (such as redesigning computer screens to emphasize preferred tests), policy changes, and educational interventions.

"Strategies that will not work by themselves are trying to build consensus among your medical staff or establishing guidelines," Dr. Plapp said. These must be accompanied by physician education that includes presenting meta-analysis data on lab use and specific data on your lab’s costs. "Show physicians their own data on utilization—that has the greatest impact," he said.

In making changes to testing guidelines, Dr. Plapp said wooing key physicians is extremely important. "If that key physician agreed, they [other physicians] never questioned it," he said. "Otherwise we got fallout."

Environmental changes can result in dramatic drops in lab use, Dr. Plapp said. "Don’t offer large panels that can be shotgunned," he said. Improving turnaround times reduces the number of duplicate orders. Giving physicians immediate and easy online access to inpatient and outpatient results also leads to decreases in test ordering. "By merging inpatient and outpatient results you can reduce the number of tests in the first 48 hours after admission by 25 percent," said Dr. Plapp.

Policy changes consist of requiring a pathologist to approve expensive esoteric tests and giving physicians a decision-support system that looks at the outcomes of lab test ordering. "Control how often a test can be ordered," he said. Another important step: Provide financial feedback on how physician ordering affects costs.

An educational intervention might be to develop clinical pathways, with physician input. "We publish a Clinical Lab Letter to keep doctors up to date on old and new tests and new clinical pathways," Dr. Plapp said. It’s also useful to provide timely pathology consults on tests.

In the past several years, using a combination of these approaches, St. Luke’s undertook 65 projects to reduce excessive testing. The result: The number of tests per patient dropped, as did laboratory errors, and turnaround times improved.

So, just what did St. Luke’s do? Here are 10 examples:

1. It targeted excessive testing by looking, for example, at the monitoring of vancomycin levels. "Doctors were routinely ordering peak and trough levels on every patient," Dr. Plapp said. First, the lab worked with the pharmacy to educate the medical staff about what was important in monitoring and what wasn’t. "Eventually we deleted the peak level from the computer screen so it couldn’t even be ordered," he said. In time, vancomycin levels were cut in half, for an annual savings of $45,000 to payers.
   


2. Nurses and physicians in each specialty created clinical pathways, and pathologists reviewed the tests included in these pathways. The number of tests per case for 44 high-volume diagnosis-related groups dropped from 50 to 44. That amounted to 17,000 fewer tests per case over a year, and a four percent decrease in the lab’s workload. Today, St. Luke’s has 200 clinical pathways in place. "All new protocols and any revisions are routed through the laboratory for our input," he said.
   


3. A test algorithm was used to reduce overuse of diarrhea testing. "We found it questionable whether it made sense to do a complete O&P workup on inpatients with diarrhea," Dr. Plapp said. The laboratory reviewed 200 results and "found we didn’t detect one enteric pathogen after doing all that work," he said. For every patient stool samples had been sent to the laboratory on three consecutive days, so there were about 600 samples and $240,000 in charges billed to payers. Twenty percent of the exams were ordered on patients who had been in the hospital for three or more days, "so it was very unlikely this was due to a parasite." The new policy said that the lab test on patients who develop diarrhea after three days would be a C. difficile toxin instead of a complete O&P. For those who had diarrhea on admission, the lab would perform a giardia screen instead of a full O&P. The samples are held for seven days so the physician can ask for the full O&P if warranted, such as with an immunocompromised patient or one who has traveled recently. These changes resulted in annual savings of $400,000 for payers and $12,000 a year in reagents and labor savings for the laboratory.
   


4. Reference ranges were changed. The lab reported any ANA (antinuclear antibody) test greater than 1:40 as positive before 1995. The patients who tested positive were referred to a rheumatologist and generally nothing was wrong. "We found a Mayo Clinic study that showed with an ANA titer of 1:160 there’s less than a five percent chance of anything abnormal," said Dr. Plapp. So the cutoff was changed to 1:160. "We do an ANA qualitative and if it’s positive—above 1:160-we do the quantitative," he said. With this change, the positive rate decreased from 31 percent of patients tested to 17 percent. Payer charges decreased by $99,000 per year. The number of referrals to specialists declined, and the lab eliminated 500 manual tests.
   


5. It phased out obsolete tests, such as the rapid bacterial antigen test. The test was introduced in the 1980s to diagnose meningitis, and physicians continued to ask for it even though its clinical utility was questionable. "The medical literature was saying the test wasn’t sensitive enough to rule out bacterial origin and wasn’t specific enough to direct antibiotic therapy," Dr. Plapp said. Besides, with improved antibiotic therapy, "there’s no need for this information."

   Despite that, the laboratory continued to get requests. "Oftentimes we got a call at night to do the test and no one qualified was available," so someone had to be called in to perform it, said Dr. Plapp. A pathologist review of 22 cases in which the test was ordered found that 50 percent were ordered inappropriately—on patients with negative white counts or gram stains. "We published our data in our Clinical Laboratory Letter," and introduced guidelines for how the test should be used, said Dr. Plapp. "We monitored utilization for one year after introducing the guidelines, and the total number of orders had decreased by 75 percent." St. Luke’s discontinued the test altogether in October 2001.

   Another example of an obsolete test is bleeding time. "We found this to be a very poor predictor of whether a patient will bleed in surgery and a poor diagnostic test," Dr. Plapp said. An article published in 2001 showed that if the laboratory stops doing bleeding times, no adverse patient outcomes result.

   St. Luke’s thus devised a new algorithm to evaluate bleeding risk. If a patient has a positive history of bleeding, personal or familial, the algorithm calls for a prothrombin time, activated partial thromboplastin time, platelet count, and von Willebrand’s disease panel. If the coagulation tests are abnormal, a hematology consult is done. If the platelet counts are normal, "we do platelet aggregation," Dr. Plapp said. For patients with no history of bleeding, "we don’t do any testing."

   In 2003 St. Luke’s dropped the bleeding time manual test. This eliminated 425 manual tests per year, cutting lab time by 212 hours and resulting in labor savings in technologist and technician time of $31,875. Payer charges decreased by $108,375.
   
   

6. It made more efficient use of reagent kits, in this case for cytomegalovirus polymerase chain reaction quantitation. The lab started doing Cobas Amplicor CMV quantitative assays in October 2001. To meet the needs of the hospital’s bone marrow transplant program, "we would try to do them as frequently as possible," Dr. Plapp said. Initially, this meant on a Monday-Wednesday-Friday schedule. However, "often we didn’t have enough patients to use the complete kit and those reagents couldn’t be stored, so we had to discard the unused reagents." In January 2003 St. Luke’s introduced a new flexible schedule where "any time we had nine patients we would go ahead and do a run," he said. That dropped wastage of unused reagents from $5,500 a month to zero. Turnaround time on the as-needed testing schedule improved by 20 percent. "We’re able to turn out more results on the first day when they’re ordered," Dr. Plapp said.
   


7. It "in-sourced" tests. Dr. Plapp noted that as a result of vendors’ direct-to-physician marketing efforts, esoteric testing "has become an increasing problem for hospital labs." In particular, St. Luke’s was getting requests for HHV-6, FISH, NK cells, and cystic fibrosis. The lab reorganized in 2001 and replaced medical technicians with clinical laboratory scientists who could handle many of the esoteric tests. In addition, "the more expensive tests would be reviewed by a pathologist before being sent out," he said. These steps saved $200,000 annually. One test that St. Luke’s brought in-house was cystic fibrosis, after national guidelines adopted in March 2001 recommended the test be offered to all pregnant women. In 2002 St. Luke’s began performing the test itself rather than sending it out, for a savings of $80 per test or a total of $40,000 in 2003.
   


8. It optimized a reflex test to improve cost efficiency. In 1995, St. Luke’s hepatitis C virus guideline recommended following a positive HCV antibody enzyme immunoassay with a recombinant immunoblot assay. If the RIBA was indeterminant or positive, the lab would perform a PCR qualitative test. "Most of the time the RIBA was indeterminant," said Dr. Plapp, "so we were doing two tests and we weren’t getting reimbursed for performing both." About two-thirds of the time, both RIBA and PCR were performed. In 1997, after consulting with the infectious disease physicians, St. Luke’s changed the algorithm. If the HCV EIA was positive, the lab went directly to a PCR qualitative test. "Only if the qual was negative would we do the RIBA test," he said, followed by a repeat HCV for an indeterminant result. "The qual is very specific and more sensitive than the RIBA, and this process eliminated indeterminant RIBAs," Dr. Plapp said. That change shaved lab costs by $39,000 annually and payer charges by $63,000. For a positive HCV PCR qualitative, the laboratory in 1997 would do a PCR quantitative and a genotype. However, "we found the PCR quant had limited range," said Dr. Plapp. "Approximately 25 percent of cases exceeded the upper limit of linearity and had to be repeated after dilution." To avoid that labor-intensive process, St. Luke’s adopted the TaqMan real-time PCR test. Because it offers a much wider dynamic range than the previous PCR quantitative test, that eliminated the need for a followup PCR qualitative assay. "We discontinued repeat testing on patients with high values," saving $23,000 per year. In addition, the current version of the HCV algorithm is much simpler for physicians to remember because it cuts out the PCR qualitative test. If the HCV EIA is positive and the signal-to-cutoff ratio is greater than 3.8, "we go directly to the quantitative TaqMan PCR and get a genotype off of that," he said. If the ratio is less than that, the physician can stop the testing or, for high-risk patients, order the PCR quantitative test.
   


9. It became active in the transfusion arena, where open-heart surgery accounts for about one-third of components transfused. "We had a pathologist track the data by surgeon-specific usage," Dr. Plapp said. Each of the four open-heart surgeons was shown the data, which were reviewed also by the cardiothoracic surgery team. "They give us their input, and we publish guidelines on blood usage and present to both the medical and house staff on an annual basis," he said. In 1990, when the process started, St. Luke’s average was 4.5 units of red blood cells per open-heart surgical case. By 1994 that had decreased to one unit per case. Even though inpatient surgeries have become more complex since then, the average has crept up to only 1.5 units per case. Since St. Luke’s does 1,000 open-heart surgeries annually, that’s a savings of about $600,000. Lower blood use also decreases transfusion reaction risk and the lab’s workload.
   


10. It made sure patients were properly matched with their lab results to improve safety and cost-effectiveness. St. Luke’s does about 12,000 point-of-care glucose tests each month and had a 9.7 percent average error rate in patient identification, amounting to about 450 unidentified results per month. Working with the nursing staff, the lab started a project in December 2002 using Roche’s Accu-Chek Inform System with Rals-Plus and bar-coded armbands. One year later the error rate had dropped to 0.6 percent, and the lab is realizing an additional $40,000 in billings.

In summary, Dr. Plapp said improving St. Luke’s standing in test use—from the 75th percentile of its peer group when it started to the 15th percentile in 2003-didn’t hurt physician satisfaction. The overall rating for physician satisfaction for all hospital-based services is 82 percent, but is 95 percent for anatomic pathology and 94 percent for clinical pathology.

"Why should the lab be concerned about excessive testing?" he asked. Not only because of the cost and unnecessary use of laboratory resources, but because of the potential harm to the patient. "The more tests, the more chance of a false-positive result," he noted, which can trigger more tests or invasive procedures. Meanwhile, needed procedures might be delayed while a false-positive is tracked down.

"Because of the way we set most of our reference ranges to include 95 percent of healthy individuals, any time you order one test there’s a five percent probability that test will be abnormal even though the patient’s perfectly healthy," Dr. Plapp said. "We want to spare the patient as much additional testing as possible."

Karen Southwick is a writer in San Francisco.