College of American Pathologists
Printable Version

  Order patrol—how to rein in test


cap today

January 2006
Feature Story

Karen Lusky

George Rodgers, MD, PhD, director of the coagulation laboratory at ARUP Laboratories in Salt Lake City, recounts how he once got a call from a surgeon in Flagstaff, Ariz., reporting an "outbreak" of protein S deficiency. Intrigued, Dr. Rodgers queried the surgeon further.

Turns out all of the patients involved had samples for testing drawn when they had just been diagnosed with an acute clot. That bit of critical information ended the discussion and the hunt for the cause of the mystery outbreak. "Since protein S levels are reduced by acute-phase response," explains Dr. Rodgers, "the surgeon’s test results would be expected and do not indicate true protein S deficiency. His results were really due to inappropriate specimen timing."

Dr. Rodgers’ story underscores just how off course inappropriate lab testing can take even an experienced physician. Whether testing is ill-timed or over- or underused—or just wrong for a patient’s condition—the results are the same: Delayed or missed diagnoses and a cascade of additional testing and costs.

That’s where managing the use of lab tests enters the picture. Having shed yesteryear’s image as a purely cost-driven exercise, the effort has been reshaped as a way to improve patient outcomes and lower costs. "The goal is for the physician to order the right test at the right time to provide the right answer so the physician can take care of the patient in the best way," says Priscilla Cherry, MBA, MT (ASCP), director of Laboratory Consulting for Premier, Charlotte, NC. Cherry co-presented on lab test utilization management with Edward Ashwood, MD, director of ARUP Laboratories, at the Washington G-2 Lab Institute annual conference in October.

Cherry noted that labs are good at turning tests around quickly. But the question is, "Should you even be doing [some tests] in the first place?" she asked the attendees. In some cases, labs get test requisitions "where you don’t have the foggiest idea of what is wrong with the patient," she says. "It’s like the physician takes darts and throws them in every direction to try to figure out the best direction to pursue."

As part of utilization management, laboratories and health care organizations should ask a number of questions, according to Dr. Ashwood:

  • Are physicians using the most cost-effective protocol to establish a specific diagnosis?
  • Are physicians ordering a newly developed test that can diagnose a common disease?
  • After screening, are there indications for a more expensive test?
  • Are physicians ordering an expensive test when another is equivalent and less expensive?

Physicians may believe that failing to order every test in the book opens them up to liability, Dr. Ashwood says, "but it’s not about how much you order but what you order." Say an obstetrician who doesn’t believe in "newfangled" tests refuses to order the new molecular test for cystic fibrosis to detect that both parents are carriers and the baby is affected, he postulates. "That doctor could be sued when a plaintiff attorney shows that the American College of Obstetricians and Gynecologists recommends obstetricians use the test," he says.

Hospitals that want to "start down the path of test utilization should tackle their referred tests first," Dr. Ashwood suggests, because referred tests have high incremental costs and very low fixed costs.

They also tend to fall into an esoteric category and thus "have vague indications in many cases, especially some of the new ones," says Brian Jackson, MD, ARUP’s director of medical informatics. "And when the indications for doing the testing are vague," he adds, "you can make a legitimate argument to say the tests are not of clear-cut benefit."

ARUP has found that its clients typically can save five to 10 percent of their send-out costs, Dr. Ashwood says. He bases that estimate on findings from ARUP’s ATOP program (analyzing test ordering patterns). The program analyzes ARUP clients’ testing volume and patterns to look for opportunities for them to adopt more-effective testing algorithms, reduce esoteric testing costs, use new technologies cost-effectively, and shorten the time to diagnosis, among other goals.

To do an ATOP review, an ARUP team probes the lab’s database to retrieve test orders and analyze them using Microsoft Excel. "The tests are organized not only by test volume but also by total test costs," Dr. Ashwood explains. Next the ATOP team scrutinizes where the hospital is spending 80 percent of its money. And it looks for common themes among clients where ARUP already has standardized advice.

In that regard, hepatitis C testing ranks at the top of the list. Dr. Ashwood reports that at least half the labs that ARUP analyzes through its ATOP program could save about $30,000 a year by changing how they follow up on HCV. "So that’s our No. 1 mantra," he told attendees.

Hospitals can follow an algorithm for doing HCV testing that cuts to the chase in confirming infection and saves money by avoiding unnecessary RIBA (recombinant immunoblot assay) testing.

"If you get a strong positive on the screen, which occurs 90 percent of the time, the next test you should do is HCV RNA—not a RIBA," Dr. Ashwood tells CAP TODAY. "People will usually [show a] strong positive on the screen because most people with HCV are chronically infected if they have had hepatitis C at any time in their lives. The RNA test confirms they are infected. And the level of the RNA also helps guide treatment to some extent."

Patients who have a low-positive result on the HCV screening should receive RIBA testing, which detects the pattern of HCV antibodies, Dr. Ashwood says. "In that setting, it is cost-effective to do a RIBA before you move on to the RNA testing," he says. If the patient’s RIBA is negative, you don’t need to do an RNA because the screen was a false-positive and you can stop the testing. "And the majority of low positives will be negative on the RIBA test," he says.

If a hospital wanted to make it simple and do RNA testing on all patients with low-positive screens, that would be a better strategy than testing all the strong positives with a RIBA test, Dr. Ashwood says. "It would save you $100 for every case of HCV diagnosed," he says. "If you do the RIBAs on the low-positive screens, you can save a bit more but not a lot."

Here are other areas where ARUP has found that hospitals can put an end to inappropriate testing:

  • Free or unbound prostate-specific antigen. Clinicians tend to over order this test, in ARUP’s experience. The only time the test should be ordered is when the total PSA is higher than 2.5-4 and below 10, Dr. Jackson says. "In that gray area, evidence shows that free PSA will help push you toward or away from doing a biopsy," he says. "But that’s the only indication for the test."

ARUP also sees clinicians order PSAs on men older than 75 though men at that age generally have non-aggressive tumors, Dr. Ashwood says. In such a case, a patient is more likely to become ill from the therapy than the cancer, he says.

  • HPV testing to identify low-risk genotypes. Though there are about 100 genotypes for HPV, only 15 pose a high cervical cancer risk for a woman, Dr. Ashwood says. "Most gynecology specimens we get [at ARUP]," he told attendees, "want you to do a reflex to HPV when [the cytology] shows AS-CUS (atypical squamous cells of undetermined significance). And [while there are] molecular tests for both high and low risk, do we need to know if the patient is at low risk?" Dr. Ashwood asks. "[ARUP] feels so passionate about it that we took [the low-risk tests] out of our catalog. You can still get it if you order the panel, but we don’t offer it by itself anymore."
  • Chlamydia/gonorrhea DNA probe testing in lieu of the amplified DNA screen. The ATOP program recommends use of the amplified testing even though it’s more expensive. "The molecular probe is half the cost of the amplified test, but it’s very clear that the probe misses a lot of low-level tests" representing subclinical infection. "ARUP could take the probe test off its menu," he says, "but if we don’t do it, people will send it on" to a reference laboratory that does.
  • HIV viral load. Like many other reference labs, ARUP offers two forms of testing for HIV viral loads. One has routine sensitivity and reports results between 400 and 750,000 copies of HIV RNA per mL of blood. The other one, an ultra-sensitive test, can detect a range between 50 and 100,000 copies per mL. If a physician wants a number and the patient with HIV has been well cared for, Dr. Ashwood advises, it would be better to order the ultra first and then reflex to routine when the ultra can’t measure more than 100,000.
  • Inherited thrombotic disorders. Dr. Ashwood finds that clinicians tend to order everything in a shotgun approach to figure out the underlying disease predisposing the person to clot formation. "They tend to order everything but the most common cause of thrombophilia, which is factor V Leiden," he says. "Or they do functional and antigen tests for proteins C and S when they could rely on functional tests alone."

Dr. Ashwood says laboratories could save money by ordering the functional test for factor V Leiden rather than the DNA test. "The functional test costs one-third that of a DNA test," he says.

Dr. Jackson did a study recently using ARUP data that suggests clinicians don’t know when to order total antigen versus activity tests for protein C, protein S, and antithrombin, which means they aren’t following the CAP guidelines for thrombophilia published in 2002. "Clinicians should order activity tests—or alternatively the free antigen in the case of protein S—for initial assessment," he says. "Total antigen tests are at best useful as an adjunct in patients already found to be deficient for the protein in question."

Identifying improper testing patterns in a hospital or other health care setting is a first step. But figuring out who is generating the testing and why requires additional data mining and analysis.

Hospital labs can use their own computer systems to find out who is ordering what, says Priscilla Cherry of Premier. Look at the test volume and results and where the money is going, she advises. "For example, if you are sending out testing for factor V Leiden, the majority should be positive—that is, if the clinicians have done a good family history and identified a pattern of people with clotting disorders and done some other coagulation tests."

Hospital labs can also stratify the ordering physicians to identify the big users. For example, orthopedists tend to order a lot of blood, Cherry says. "So you might want to look at blood utilization to see how much blood each orthopedic surgeon is wasting," she suggests.

Some testing can help prevent unnecessary and invasive care, which means the hospital would want to see specialists ordering it when appropriate. "For example, fetal fibronectin can help detect women who are going to develop preeclampsia or go into premature labor," Cherry says. "And one expects to see a doctor who cares for a high percentage of women with high-risk pregnancies to be ordering the test. Or you may see some who never order it and instead hospitalize patients early or unnecessarily."

Hospital labs might also take a look at ordering patterns for sedimentation rates. "It’s a commonly ordered test but not particularly diagnostic, and while the test used to be fairly inexpensive, it’s now done on an automated instrument and more costly" to do, Cherry says.

"A sed rate is appropriate to identify temporal arteritis, but it’s not a good diagnostic tool for many other things," she says. "If someone has an infection, a C-reactive protein would give you more information. But you see doctors ordering sed rates and CRPs."

Once a hospital laboratory has mapped out problematic ordering patterns, it has to get clinicians to make changes, which can be tricky. "What worries laboratorians the most," Dr. Ashwood says, "is how they can change physician behavior without appearing to be a cop or bad guy."

He recommends a three-step process to enlist clinicians’ cooperation in managing test use. For one, the lab can alter test-requisition forms to steer clinicians in the right direction. That’s the most effective intervention to reduce inappropriate lab testing, Dr. Ashwood says.

One such option: an "out of sight, out of mind" approach in which certain tests simply don’t appear on the menu. Group Health Cooperative, a Seattle-based managed care network with 20 outpatient laboratories, achieved a 10 percent reduction in overall lab testing by simply reducing by half the number of tests on its paper requisitions, says Debbie Birkland, the network’s clinical services manager for laboratory services. "We went from the classic candy-store requisition to one with about 40 tests on it," she says.

Mark Wurster, MD, an assistant clinical professor of medicine and medical director of the anticoagulation program for the Ohio State University Primary Care Network, Columbus, suggests the emergency department should not have proteins C and S and antithrombin as part of order sets for hypercoagulability because these tests will be falsely low during an acute clot. "But the test menu could [allow the clinician] to check factor V Leiden and prothrombin gene mutation, lupus anticoagulation assay, and homocysteine" as potential causes of clotting, Dr. Wurster says.

A second way to promote appropriate lab testing would be to educate clinicians through grand rounds, newsletters, and CME lectures. And, third, labs and hospitals can reinforce positive changes by auditing clinicians’ use of new protocols and offering feedback.

All hospitals should implement lab utilization committees to help in overseeing testing, Dr. Jackson advises, just as they have pharmacy and therapeutics committees.

And the messages should be targeted to the physicians who most need to read or hear them to avoid overloading all clinicians. In this regard, hospitals can take lessons from drug companies who have mastered the technique of delivering brief, targeted messages to the right people, Dr. Jackson says.

"Drug companies have a different perspective than that of labs in that they aren’t trying to push clinical appropriateness of their products, as labs are," he adds. But the technique can work for labs all the same. First drug companies closely track pharmaceutical prescribing in the community. "They know every clinician’s ordering profile and the trends," Dr. Jackson says. "And they only visit clinicians whom they think they can influence."

To succeed in getting clinicians on board with test use management, labs need clinical torchbearers for their evidence-based protocols, and surely the support and endorsement of the executive physician management, Cherry says. "You want ... the guy at the top supporting you when the doctors start throwing darts at you—and they will," she says, "because sometimes they misunderstand what test utilization management is about."

A select group can help get other clinicians on board. Kaiser Permanente Northern California regional reference laboratory, which performs more than 20 million tests a year, has used clinical champions for endocrinology and infectious disease for a number of years with great success. The champions come to the lab each week for a half day and are paid for their time, says Gene Pawlick, MD, Kaiser Permanente laboratory director.

"The two clinician ambassadors have a major [role] in going back to the constituency of physicians and telling them how to appropriately use tests," Dr. Pawlick says. "The infectious disease physician, who happens to be the chair of the department, discusses the issues within his group and they will decide, for example, that only physicians practicing infectious disease or the gastroenterologists can order a certain test [such as HCV]. We put limits on who can order a test and then educate those people on how to use it effectively."

Helping physicians and even patients navigate the maze of laboratory medicine is part of promoting appropriate test use, Cherry says. "These days physicians have been trained to rely more on diagnostic testing and they know they have a plethora of things they can order," she says. "But they haven’t been taught to use them effectively."

To make matters worse, patients pressure doctors for tests and drugs. "So patient education is part of the equation in changing test utilization patterns," Cherry says.

Clinicians also tend to become interested in new tests, some of which are expensive and promoted heavily by manufacturers’ public relations arm, says Ohio State’s Dr. Wurster. But more costly care doesn’t necessarily mean better care. Dr. Wurster says he’s been haranguing house staff for years to name four antibiotics that cost less than $10—and rarely can they answer. "Why? Because they only hear about the real expensive ones. It’s the same with testing for hypercoagulability. The prothrombin gene is very sexy. Factor V Leiden is famous."

Yet not every patient who develops a clot needs testing for a genetic defect, Dr. Wurster notes. "If you have a patient with a first-time deep vein thrombosis or pulmonary embolism, you treat [with anticoagulant therapy] for a number of months ranging from three to nine," he says. "If the person has a second DVT, you probably need to look for the underlying genetic problem. But you might do that after a single episode if you have an atypical presentation where the clot occurs at a young age and appears unprovoked [no surgery, immobility, etc.] or if the clot is in an [unusual] site. Or you might order the testing earlier for someone with a positive family history with affected first-degree relatives."

There’s also a group of patients for whom you want to identify risks because they have a family history—for example, before initiating hormonal birth control, Dr. Wurster says. "If the woman has siblings or first-degree relatives with DVTs at a young age, the clinician could consider screening her for protein C, S, antithrombin-3, and factor V Leiden before starting the birth control."

Clinical risk assessment and stratification can go a long way toward avoiding the false-positive results that put clinicians and patients into an expensive and worrisome loop of additional rule-out testing and potentially wrong diagnoses.

One example is HLA B-27 testing for ankylosing spondylitis in patients with back pain. Ordering this test on every Caucasian patient who presents with back pain will lead to an eight percent false-positive rate for that population, Dr. Ashwood says. But if the clinician can narrow the probability to about 50/50 that someone may have the condition, then HLA B-27 testing is an effective test to order in terms of helping the clinician make a definitive diagnosis.

"Clinical evaluation that determines a patient has restricted mobility [he can’t touch his toes] raises the level of suspicion that the patient may have ankylosing spondylitis," says Dr. Ashwood. "The physician can then do imaging and, if that’s equivocal, order lab testing" for HLA B-27.

Physicians should also avoid ordering ANA testing for systemic lupus erythematosus for patients who have a low pretest probability of having the condition, says Mark Wener, MD, professor in the Department of Laboratory Medicine and director of the Immunology Division at the University of Washington, Seattle. That’s because even when ANA testing is done by the best of labs, five percent of the normal population will test positive, says Dr. Wener, who co-presented with Michael Astion, MD, PhD, in a Dec. 7, 2005 AACC audiconference, "Autoimmune Disease: New Strategies to Improve Testing and Management."

To illustrate when not to order an ANA, Dr. Astion presented a clinical scenario of a male patient who visits the doctor complaining of a burning feeling when urinating. The patient says his girlfriend was recently diagnosed with "lupus." And he’s read about lab tests for lupus online and wants the doctor to order an ANA test. In this case, "there’s a 99 percent chance that a positive ANA test would be a false-positive result," says Dr. Astion, director of reference laboratory services, Department of Laboratory Medicine, University of Washington.

"In the setting of SLE, testing is useful, however, when there is a reasonable chance that the patient has SLE, even though the likelihood is that the patient doesn’t," Dr. Astion adds. Someone fitting that bill, he says, might be a 40-year-old woman who comes to the physician complaining of fatigue for a month and hair loss who describes a rash she had near her nose although it’s not present during the office visit.

To promote better clinical and lab practice nationwide, ARUP’s Dr. Ashwood suggests pathologists collaborate more with medical organizations to help develop clinical protocols that include lab testing. That way, pathologists and laboratories would help optimize the guidelines and pave the way for their use by standardizing guideline-recommended testing, he says.

"For example, when clinicians first began using hemoglobin A1c to monitor glycemic control in patients with diabetes, there was an amazing disparity among methods labs used for the test," Dr. Ashwood says. "Then laboratorians standardized the testing methodology so that when clinicians advise their patients to shoot for a goal of less than seven percent, as recommended by the American Diabetes Association, that value will be the same across labs," he says.

Could managing laboratory test use ultimately help improve outcomes and reduce costs nationwide? Though Dr. Jackson says there are "mixed incentives right now that muddy the waters for providing more cost-efficient care," it’s his "gut feeling" that managing lab test use is a good starting point for saving money downstream. "Lab testing drives so much of health care treatment," he says.

Or, as Cherry puts it, "It’s about saving lives, and as we do that, we can save money."

Karen Lusky is a writer in Brentwood, Tenn.