College of American Pathologists

  Pulling back the reins on superfluous testing


CAP Today




September 2010
Feature Story

Karen Lusky

What laboratory topic encompasses money, psychology, patient safety, and a nifty statistical concept that also applies to baseball rookies?

The answer, according to presenters in a popular symposium at the July AACC meeting, is lab testing utilization management, which many view as the ticket to improving health care outcomes and cost savings to boot.

“If you had a nickel for every time someone ordered a lab test they didn’t need, you’d probably have a very large booth” at the exposition, pathologist Geoffrey Baird, MD, PhD, of the University of Washington, told AACC conferees in the session, “How to Get Clinicians to Stop Ordering Tests They Don’t Need.”

“Conventional wisdom,” said co-presenter Kent Lewandrowski, MD, associate chief of pathology at Massachusetts General Hospital, holds that about 20 to 50 percent of the lab testing that is done is unnecessary, especially high-volume automated testing. To drive down that number, you can apply “Psych 101”: You make it easy for clinicians to do the right thing and “a nuisance to do the wrong thing.” You set up systems that relieve busy clinicians of having to make decisions about certain tests. That tactic “can be devastatingly effective,” Dr. Lewandrowski said.

For example, by removing a test from the order sheet, Dr. Baird said, you “do the Jedi mind trick” on clinicians. Yes, they can still write it in, but if people don’t see the checkbox for the test, they’re less likely to order it.

Removing lactate dehydrogenase from Massachusetts General’s “quick pick” computerized order menu of routine tests caused monthly volumes to drop by half (from more than 2,000 to 1,000). “LDH is abnormal in patients with hepatocellular disease and a variety of other conditions,” Dr. Lewandrowski told CAP TODAY, “but usually other tests such as liver enzymes and bilirubin are more specific.” LDH is therefore redundant and usually provides no additional information except in selected cases such as hemolysis and in some hematological disorders, he adds.

A laboratory will not ring up major savings by reducing high-volume test use, where costs are generally fixed except for reagents, which aren’t that expensive, Dr. Lewandrowski told the audience. But you will eliminate the “operational sludge of unnecessary testing,” which he likened to “digging holes and filling them in.

“In the end, if technologists are busy doing tests that are not useful, this detracts from their attention on doing tasks that are useful to patient care,” he said.

As a case in point, Dr. Baird shared how his institution was doing huge amounts of serial inpatient hospital testing for ionized calcium, which tends to run low in patients who are very ill. It’s unclear whether raising the calcium level in that setting is potentially harmful, Dr. Baird noted, pointing to an animal sepsis model study that suggested that could be the case. A Cochrane Review also says “there’s no good evidence to support replacing calcium in critically ill patients unless the person has some specific calcium-related disorder,” he adds (

Yet when clinicians at Dr. Baird’s institution saw patients had low ionized calcium levels, they would often order intravenous calcium gluconate to correct them. And the patients’ ionized calcium levels would go up “a little bit,” he says, perpetuating the cycle of testing and medicating.

But was it really the calcium gluconate doing the trick?

Based on an analysis of more than 58,000 iCA tests and related IV calcium administration, the answer was “pretty much no,” Dr. Baird said. He and his colleagues found that the iCA levels would have gone up anyway in most cases because of a statistical phenomenon called regression toward the mean. Put simply, when doing serial testing, if you have one calcium test result that’s low, the chances are it will be closer to the mean the next time you measure it, he said.

“It’s the same reason why a rookie baseball player does great his first year but not so great the second year,” which is known as the “sophomore slump,” Dr. Baird explained. What’s probably really going on is that the rookie is a normal person who is performing closer to the mean during his second season.

To stop the iCA over-ordering and related calcium gluconate administration (which was requiring a full-time pharmacy aide to manage), the lab instituted a voluntary reflexive testing algorithm whereby clinicians ordered iCA only for patients with a total calcium level of less than 8 mg/dL. While iCA is actually the better test for a patient’s calcium status, testing for total calcium is less costly and much easier to do than iCA, Dr. Baird pointed out.

Long story short: Implementing the reflexive panel at two academic medical centers reined in calcium gluconate administration by 45 percent to 81 percent; iCA testing decreased by 72 percent to 76 percent (Baird GS, et al. Clin Chem. 2009;55:533–540).

Diagnoses of hypocalcemia in inpatients also became less frequent, probably because the diagnoses were actually “laboratory diagnoses that were not clinically consequential,” Dr. Baird told CAP TODAY. The number of cases of seizures and tetany, which are symptoms of hypocalcemia, actually decreased as well, though Dr. Baird isn’t sure why. “We didn’t observe significantly more deaths or cardiac arrests,” he said. And the pharmacy aide was assigned to other duties.

Reducing high-volume, daily testing also helps to lower the risk of blood transfusions required to treat iatrogenic anemia. In his talk, Dr. Baird presented the case of a hospital patient who had received more than 160 consecutive ionized calcium tests and greater than 80 metabolic panels. As a result, the patient likely had to receive an extra unit of blood.

ICU physicians have many times approached Dr. Baird with concerns about iatrogenic anemia caused by repeated testing. As a solution to the problem, one physician suggested using pediatric tubes to reduce the blood volume. “But that’s missing the point,” Dr. Baird says, “because what you should be doing is looking at what tests are ordered, with the goal of avoiding unnecessary tests.”

Moreover, “small pediatric tubes don’t fit on automated chemistry machines, and if we used the [tubes], we probably would not have a sufficient specimen volume to be able to repeat problematic tests.”

To eliminate the excesses of daily lab tests, Massachusetts General initially educated house staff, a strategy that worked wonders until the educational effort stopped. The test ordering pattern returned to where it was pre-education. Further, fully a third of medicine house staff turn over every year, thus “depreciating the effectiveness of one-time educational efforts,” Dr. Lewandrowski said.

So, beginning this fall, house staff will no longer be allowed to order daily lab tests—with one exception: “Daily prothrombin times for patients on Coumadin,” he said, “because we know that sooner or later they are going to forget to do that” and cause someone harm. Aside from that, they will have to choose what tests to order each day.

“Our medical attendings are very supportive of banning,” Dr. Lewandrowski told his AACC audience. “They say, ‘We’re sick of looking at all these labs that [the house staff] order on a daily basis.’”

Also on the hospital’s banishment list starting this fall is CK-MB, a cardiac necrosis marker that, evidence-wise, has been supplanted by troponin for assessing patients with signs and symptoms of acute myocardial infarction. The hospital will no longer have the test as part of its electronic rule-out MI template, which spells out the recommended testing, medications, and so on for acute MI.

For other hospitals in various areas of the country, going solo with troponin has been a win-some-lose-some proposition.

In talking to hospital labs about their routine testing, Brian Jackson, MD, director of informatics at ARUP Laboratories in Salt Lake City, finds that a lot of laboratories still run total CKs on their chest pain patients. “They will have a standard cardiac panel of CK, CK-MB, and troponin run at the same time, which is really wasteful.”

Dr. Baird’s institution has tried to reduce CK-MB testing by implementing a reflexive panel that offers troponin I first, followed by CK-MB if the troponin is moderately elevated. Initially, the intervention cut CK-MB use by about 50 percent. But CK-MB use “has crept back up over the years, as various other trainees have come through and begun to circumvent the reflexive algorithm with up-front orders,” Dr. Baird says.

In some cases, confronting clinicians with patient data can help them let go of an unnecessary test like CK-MB, as clinicians did with iCA at the University of Washington.

For example, at Hennepin County Medical Center in Minneapolis, the last holdout for using CK-MB was a group of physicians who were performing percutaneous coronary interventions. The cardiologists were “adamant” about using the test based on old literature supporting its use, says Fred Apple, PhD, medical director of clinical laboratories. “Last year, we met with that group again and performed a 50-patient comparison of cardiac troponin and CK-MB results, and agreed CK-MB was no longer needed.”

Now CK-MB is a send-out test at Hennepin County Medical Center. “We took it off our formulary so if anyone wants it, it has to be approved by the on-call laboratory medical staff,” Dr. Apple says. To date it’s never been approved as a send-out.

“Some cardiologists,” says Cleveland Clinic emergency department physician Frank Peacock, MD, “argue that you can use CK-MB to size an infarct,” because CK-MB rises and falls quickly. A lot of data exist on how that peak level correlates with the size of heart infarcts, he notes. By contrast, troponin remains elevated for seven days after an acute myocardial infarction, so “we don’t have the same database for troponin.”

But today, someone with an acute MI typically goes to the catheterization laboratory, Dr. Peacock notes. “The argument about sizing an infarct might have some merit, if you’re trying to decide whether to send a 90-year-old to the cath lab.” But what a physician is really doing when using CK-MB in that way is “waiting for someone to infarct bigger.”

There is a cautionary note, however, about using only troponin. “Without naming names, there are assays where troponin is pretty invisible at the lower levels,” Dr. Peacock says. “So you do have to look at the sensitivity of the troponin assay if you’re going to use it as the sole marker.” He also advocates performing creatinine along with a troponin for acute MI because troponin’s sensitivity for detecting acute MI is lower in renal failure (this is true, he adds, for troponins I and T).

It’s important, says David Grenache, PhD, medical director of ARUP’s Special Chemistry Laboratory, to pay attention to any test’s sensitivity and performance when the test is used in an algorithm where a yes or no (binary) answer takes you down different paths. In addition, Dr. Baird says, a lab should look at its own data to confirm the appropriateness of a particular cutoff for a test in a reflexive algorithm.

To tackle the more expensive, lower-volume tests, the best strategy is often the good old-fashioned face-to-face meeting. Dr. Lewandrowski recounted how a single meeting he had with the chief of urology at Massachusetts General saved about $100,000 a year by eliminating use of an $80 tumor marker, NMP-22, for bladder cancer. The chief of urology agreed to ban its use because the physicians weren’t using the marker to change patient care.

Labs looking for savings might take a look at tumor marker testing overall. It’s an area, says ARUP’s Dr. Jackson, where one finds inappropriate ordering or over-ordering. “For the right patient, it can be appropriate to order these markers, but in some places, doctors are ordering them for patients who have not yet been diagnosed with cancer,” a practice the evidence does not support.

Also, the American Society of Clinical Oncology guidelines for breast cancer say there’s no clear evidence to indicate whether use of CA 15-3 and CA 27.29, which are “roughly equivalent” tests, improves outcomes even in patients with biopsy-proven breast cancer, Dr. Jackson says.

ARUP has noticed that ordering volumes for ovarian cancer marker CA-125 are “highly variable from hospital to hospital,” he says. “Whenever you see testing that is not uniform like that, it’s a strong indicator of [it being] preference-driven,” either by patients or, more often, physicians. “This suggests that many of the test orders may not be leading to medical value for the patient.”

For Massachusetts General, a focus has been to come up with a more rational practice for ordering expensive genetic tests, which are “growing like dandelions on a New England lawn,” as Dr. Lewandrowski put it.

In fact, when the lab mapped its reference testing budget to ordering clinicians, they traced almost a third of the spending to 25 physicians in pediatric genetics and neurology. The pediatric geneticists won in the biggest users category, however. But “you can’t just yell at them,” says Anand Dighe, MD, PhD, director of the MGH core laboratory. “They are as invested as we are in creating guidelines to help families select the best course of action for their child.”

Over the past year, the lab has worked with pediatric geneticists to develop guidelines for ordering the genetic tests, a strategy that has reduced the testing to what Dr. Dighe views as a “sustainable level—at least until the next big wave of genetic tests.”

Next, the laboratory plans to develop a formulary for reference lab testing. Using a color-coded scheme, green tests will be ones that anyone can order, Dr. Lewandrowski said. Examples include aldosterone and hepatitis A serologies.

Yellow tests can be ordered only by specialists or with approval from a specialist. Examples include the Trofile assay for aiding drug selection in patients with HIV, and a complete hereditary ataxia panel for neurology.

Red tests (such as leptin to assess appetite and weight) aren’t recommended for clinical use. A pathologist acting as gatekeeper will have to approve all such orders.

Pathologist-driven gatekeeping measures might best come with a “don’t try this at home alone” warning. “You cannot do utilization management by sitting in your office and making edicts,” Dr. Lewandrowski stressed. If you try, he said, you might keep in mind the adage, “Friends may come and go, but enemies accumulate.” Put another way, “Heroes tend to die.”

Instead, “you have to have the protection of an interdisciplinary body that includes the medical staff as well as the laboratory staff” and administration, he counseled. Include nurses, too—they tend to be invested in eliminating blood draws and work associated with unnecessary lab testing.

Hospitals can make inroads in utilization management whether or not they have computerized physician order entry systems for lab tests, which Mass General continues to develop. (The University of Washington doesn’t have a CPOE system, Dr. Baird said, who describes his brand of CPOE interventions as changing order forms.) CPOE linked to the central lab makes it possible for the laboratory to interact with the ordering physicians when test-ordering decisions are being made and when audits, among other functions, are easier to do, Dr. Lewandrowski said.

Whatever a lab’s computerized capability, the lab has to analyze its data to flag a utilization problem and decide how best to manage it. Then you monitor how well an intervention is working, which ideally includes a look at the impact on patient outcomes, as Dr. Baird’s ionized calcium testing study did.

“You need to develop metrics,” Dr. Lew­an­drow­ski said, “to determine not just on an individual level [per test] whether you’re being successful, but also on an organizational level.” Massachusetts General measures the number of inpatient tests per hospital discharge, which dropped by 26 percent from 2002 to 2007. In fiscal 2007, the hospital moved to a system in which a chemistry panel was counted as one test, making the utilization management effort suddenly look wildly successful, Dr. Lewandrowski says. “Therefore, you need to understand your metrics and be willing to adapt them to changing conditions.”

Massachusetts General now also monitors the number of free-text electronic lab orders, which clinicians key in when a test isn’t on the CPOE menu. In 2008, when the lab started the implementation of laboratory CPOE, about 10 percent of orders were free text. With increased monitoring of free-text orders and improved house staff training, free text usage is now down to less than one percent. And that’s a good sign given that free-text orders “generally represent confusion and disorganization on the part of the house staff and physician,” Dr. Dighe says.

Today, clinicians who submit free-text lab orders get a boilerplate e-mail from the laboratory explaining the function’s downside and providing a tutorial on how to avoid using it.

Last but not least, lab utilization management, done right, involves identifying and promoting underused testing that is known to improve patient outcomes. That list, according to Dr. Lewandrowski, includes cholesterol screening, Pap tests, HIV testing, cystic fibrosis testing, and A1c for diabetes. Dr. Jackson adds chlamydia screening to the list, noting physicians don’t offer it like they should.

Massachusetts General’s Elizabeth M. Van Cott, MD, cites a couple of coagulation tests that, when underused, can take a toll on patient outcomes. One is heparin-induced thrombocytopenia testing, when clinically indicated. “HIT can have devastating consequences for the patient, and it is a very common cause of medical malpractice lawsuits,” says Dr. Van Cott, director of the coagulation laboratory and medical director of the core laboratory.

Another underused approach involves making sure that a positive lupus anticoagulant test isn’t due to a factor inhibitor (a specific antibody against a factor), Dr. Van Cott says. Figuring out if you have a false-positive is clinically important, she says, because a positive lupus anticoagulant increases the risk of clotting whereas factor inhibitors can promote dangerous bleeding. (For more on using lab utilization management strategies for coagulation testing, see In coagulation, reflexive testing algorithms do the trick.)

The bottom line, Dr. Lewandrowski says: Lab utilization management is about more than cutting costs and reducing the number of tests. “It’s also about appropriate utilization to produce efficient and effective episodes of care.” If laboratories just put “cost containment on the utilization management label,” he says, clinicians will always have some “degree of skepticism” about it. “They won’t be nearly as enthusiastic as they will if it is improving efficiency and the quality of care.”

Karen Lusky is a writer in Brentwood, Tenn.

Related Links Related Links