March 2008
Feature Story

Karen Titus

With apologies to John Donne, let it be said that no lab is an island.

Group Health Cooperative, a large managed health care system in Washington State, appears to come close, at least on maps and satellite shots. With 20 clinics in Seattle (and five more in the eastern part of the state), it's naturally encircled by plenty of nearby water, including Puget Sound and Lake Washington, which you've heard of; Salmon, Portage and Union bays and lakes Union and Green, which you probably haven't; and the West and East waterways, with which even GHC employees may not be familiar.

In terms of practice, however, Kim Riddell, MD, service line chief of the clinical laboratory service at GHC, nonetheless makes it clear that some of the most important work she does-mending lab test utilization-depends largely on what those outside the laboratory do, say, and think. The lab is not entire of itself.

"I'm one clinical pathologist. I can't do everything," says Dr. Riddell.

• figured out how to use clinician report cards effectively.
• reined in wellness testing, after finding that a significant percentage of tests ordered as part of wellness assessments were inappropriate.
• discouraged the routine ordering of AST tests for patients on low-dose statins.
• reduced the numbers of orders for automatic urinalyses and cultures for working up uncomplicated dys­u­ria in young women.

These successes all involved outpatient testing, an area where many fear to tread.

One reason labs hesitate to take on outpatient test utilization is the lack of financial stakeholders, says Mark Lifshitz, MD, director of clinical laboratories, New York University Medical Center. "Does anyone really care that the test is being ordered? The patient's not paying for it; the physician's not paying for it. 'No one' is paying for the test, right?" He likens it to shopping in a supermarket-buyers won't fill their carts if they have to pay for every item in them. But if they're told they can take what they want for free, they'll load up. "To some degree, that's the case with outpatient tests," he says. Patients aren't likely to question the need for tests unless they're required to pay for a large percentage of the costs.

Wellness testing in particular can be even harder to control, he continues, because patients expect their physicians to do a thorough evaluation-and in patients' minds, that often translates into more tests. "I'm not suggesting that physicians order tests just to make patients happy," Dr. Lifshitz says. But thanks to information that's spread on the Web and in medical advertising, "You have a more informed patient who comes in and will often say to the physician, 'Oh, I heard about this test,' or 'I want this test done and that test done.'" None of which is a reason for doctors to order an inappropriate test. But physicians sometimes do capitulate, Dr. Lifshitz says, bowing to competitive pressures on the physician-patient relationship. "There's a tendency to order more tests, or tests that might not typically be necessary for the patient, especially when it comes to screening."

The so-called convenience factor can also boost test orders. It's much easier to rule out all the diagnostic possibilities at once, in one visit, rather than doing it sequentially based on an algorithm that might require followup visits. A good example is screening for activated protein C resistance. If the initial test is positive, it's usually followed by a test for factor V mutation. But because this requires a fresh sample if one hasn't been drawn alongside the initial test, there's a tendency to go either straight to the mutation test or order both tests simultaneously, even though most APCR tests will be negative.

Dr. Lifshitz says he doesn't have an answer to this dilemma. On the one hand, algorithmic testing is pervasive. Yet it's easy to argue that the additional tests being done are inexpensive. "Maybe it makes sense to do all the tests upfront. But the downside is you may end up with false-positives." Ultimately, the matter slides into a different question: What perspective should be considered? "From the medical side it doesn't make sense," says Dr. Lifshitz. "But from a convenience side, both for the physician and the patient, it does."

It can also make sense for patients cost-wise, Dr. Riddell chimes in. She has worked with several oncologists to develop a more streamlined monitoring panel for patients undergoing chemotherapy. "But the patient ends up paying something like $156 for 10 tests, as opposed to, say, $40 for a 14-test comprehensive metabolic panel." Dr. Riddell's own efforts began with wellness testing. It's a worthy target, given that many screening tests might be overrated.

Both Dr. Riddell and Dr. Lifshitz point to the U.S. Preventive Services Task Force, which, as part of its push to link outcomes and performance pay, has tried to identify which screening tests are useful and which aren't, based on scientific evidence. Though these efforts are not specific to laboratory testing, "What's interesting is that they recommend against certain types of tests that in the community are very frequently ordered as screening tests," Dr. Lifshitz says. "For example, one could argue that potentially most of the urinalyses ordered for screening purposes are unnecessary when one tries to apply the available scientific evidence." Other common screening tests that may be suspect include thyroid testing in adults, triglycerides as part of lipid panels, and PSA and blood glucose screening, he says.

It's not that the tests are unnecessary per se; rather, they've been classified in a category for which there is insufficient evidence to make a recommendation either for or against their use in screening, Dr. Lifshitz explains. "Yet that's how most of them are being used.

"That's pretty amazing," he continues. "You think, with these high-volume tests, there's lots of scientific evidence to suggest they're useful." But that may not be the case. As other organizations-Dr. Riddell points to guidelines from the American Academy of Family Physicians as another good source of information-join the discussion and rely more on evidence-based medicine approaches, the broad screening approach may fade away, replaced by more targeted uses based on detailed guidelines that take into account age, individual clinical indications, risk factors, and the like. Like Dr. Riddell, Dr. Lifshitz is a big fan of guidelines, either from regulatory or professional groups. Without them, it's hard to talk to physicians about what's appropriate and what isn't. "It's much easier to explain to them, 'Listen, here's the guideline-why are you doing this when the guideline says that?'" Guidelines, especially from the physician's own specialty, can carry more weight than a pathologist's words, no matter how sage those words may be. "That's a lot better than my saying, 'I don't know why you keep ordering this test,' or, 'Do you really need this test?' or 'Another test might be better than this one.'"

Fortunately, he says, more guidelines are coming onto the scene. "There are more attempts to ask the question, 'Is what we've been doing for the past 40 years good medicine?' The more we look at that, the more we can head toward appropriate utilization of lab tests."

Dr. Riddell and her colleagues took a more recent view when they began investigating test utilization at GHC, which, along with two subsidiary health carriers, serves more than half a million members in Washington and Idaho. In monitoring utilization, the lab discovered its test volumes grew five percent annually from 1999-2001, despite declining enrollment. It also noticed a large variation in ordering patterns at its different clinics, despite a similar mix of patients at the various sites-from 300 tests to 550 tests per 100 patient visits (with common panels considered one test). They also found that more than a quarter of ordered tests were part of wellness assessments (such as comprehensive metabolic panels, CBCs, TSHs, and routine urinalyses), a figure deemed overly high based on evidence-based guidelines; nearly a quarter of tests were not linked to the clinical investigation; 15 percent of tests for disease monitoring were not done at the correct time intervals; and close to 20 percent of the panels could have been replaced by a single test.

The lab responded with individual physician report cards, issued quarterly. The first ones came out in the first quarter of 2003. Now, five years into the program, Dr. Riddell reports, "This is the only situation that I'm aware of where we've been able to show lasting changes with a medical staff over a long period of time in the U.S." (See testing trends at left.) She notes that a paper from the Netherlands (Verstappen WHJM, et al. JAMA. 2003;289:2407-2412) describes a health care system similar to GHC, which also used report cards to address use of wellness testing and achieved similar success, though the study covered only a six-month intervention period. (CAP TODAY was unable to reach the researchers involved in this trial.) Another report, also from the Netherlands, documented nine years of effective, continuous feedback to physicians (Winkens RAG, et al. BMJ. 1996;312:490-492).

The report cards solve the outlier problem that often pops up in inappropriate test use. Consistent peer pressure, provided by the report cards, can quickly steer those one or two physicians back on course. The report cards are made available on the GHC Web site; though the information is not available to outsiders, physicians can view results of all their colleagues in the system. That includes clinic chiefs. "They're unmasked too," Dr. Riddell says. "By making this open to the other docs, they can pretty much see for themselves where they all stand, and who the leaders are." The report cards are issued only to physicians in the Puget Sound clinics, though the lab plans to start issuing them to clinicians at its five Eastern Washington clinics this year.

Dr. Riddell emphasizes there was virtually no opposition to launching the report cards or making the results so transparent. The pharmacy department has a similar program, which suggests feedback is a strong part of the culture at GHC. "I haven't heard any complaints about this," Dr. Riddell says. "The fact that some 86 to 90 percent are compliant tells me this is accepted."

Dr. Riddell continues to monitor lab reports for outliers in wellness testing. "They usually fall into one of two categories. One is the brand-new doc who has just joined our group and is not familiar with the program." These physicians receive "what we call a new doc letter-we usually see a fairly immediate response." The other outliers tend to be more experienced physicians who've returned from a sabbatical, vacation, or locum tenens. They, too, receive a letter from the lab, which tends to nip the problem in the bud.

The lab has zeroed in on other targets over the years. AST testing has been another success story, specifically, curbing the routine ordering of ASTs for patients on low-dose statins. "This used to be a HEDIS measure, and we used to have automatic reminders on our clinical workstations telling physicians their patients were due for their AST." But it's no longer a HEDIS score, and the lab is working with the IT department to remove the pop-ups. "There's a fair amount of literature that shows this is no longer necessary," she says.

Another early success story involved dropping automatic UAs and cultures for uncomplicated dysuria in young women. "The evidence is rather powerful, in that the chief complaint is more predictable of a good response to antibiotic therapy than a positive culture," Dr. Riddell notes. Women phone in their symptoms, and, if they do not meet the exclusionary criteria (see box, page 52), receive a prescription for an antibiotic. "Our urine cultures dropped dramatically when this guideline was implemented," she says (Saint S, et al. Am J Med. 1999;106:636-641).

How does Dr. Riddell pick her targets?

Sometimes her technical managers get the ball rolling, letting her know when they suspect tests are being ordered too often.

That's an approach other labs have used as well. At Massachusetts General Hospital, the laboratory technologists will note when a prohibitively expensive molecular send-out test has been ordered and bring it to the attention of the laboratory director, says Kent Lewandrowski, MD, associate chief of pathology and director of clinical services (anatomic and clinical), MGH, and associate professor, Harvard Medical School. "And in other areas they'll come up and say, 'You know, blank inpatient unit seems to be ordering blanks with their blanks,'" says Dr. Lewandrowski, giving an excellent though probably unintentional impersonation of "Match Game" host Gene Rayburn. Just as valuable, he says, is when technologists confirm that something isn't a problem.

Fellow pathologists can also raise the red flag. A GHC pathologist colleague in Eastern Washington noted what appeared to be irregular HPV ordering patterns by primary care physicians in his clinic. When he asked Dr. Riddell if she encountered similar problems among her clinicians, she turned to her microbiology manager, who reported a handful of clinicians did indeed seem to be befuddled about HPV testing. "So we'll be looking at this too," Dr. Riddell says.

Dr. Riddell also looks to the literature. "That's how I found out about AST," she says. "I serendipitously ran across a paper in the Archives of Internal Medicine [Smith CC, et al. 2003;163:688-692]." The test was one of the top 10 at GHC, so the paper gave her pause, to say the least. "I first went to our clinical practice and improvement committee and asked them, 'What do you think of this? And why is this still a HEDIS measure?'" The clinicians then did a little digging and discovered that Kaiser Permanente had urged its removal as a quality measure, based on its own study. A GHC epidemiologist wrote a summary article on the topic that was presented to the heart care committee and the pharmacy and therapeutics committee. Not long after that, the lab and clinicians decided to halt routine AST orders.

She relies heavily on current recommendations. When she began investigating wellness testing, "I knew I was addressing a primary care audience, so I went to the primary care literature to get references to back me up."

She also builds plenty of bridges to clinicians. She keeps in close touch with an epidemiologist who keeps her posted on the latest issues in stool testing studies. This lets Dr. Riddell know what areas might be worth focusing on down the road, using studies that "are a lot more rigorous than anything I could do."

"There are a lot of things I leave to other experts who have researched these things far deeper than I have," she continues. "Cervical cancer screening is left to the cervical cancer screening committee, and we have a pathologist who sits on that." H. pylori is covered by the gastroesopha­geal reflux disease committee. A technical manager sits on the deep venous thrombosis committee. "There are a fair number of resources here that I can draw on to help me with this work. Plus, I can't be a control freak or micromanager, and I have to pick and choose my battles very carefully."

Interestingly, she points out that the 2003 Dutch study on report cards was brought to her attention by the GHC medical director. "When docs are with you, they can really be a big help," she says. "Sometimes they point you to things that you haven't even thought of. It helps so much to involve as many docs as you can."

All of which has kept the lab connected, away from island status. "I'm the only clinical pathologist, but on the other hand, I'm not exactly alone," says Dr. Riddell.