March 2012
Feature Story

Anne Paxton

When Arlo Guthrie sang about a notorious eatery where “you can get anything you want,” chances are good he wasn’t thinking about laboratory testing. But until recently, when it came to the test menu of the clinical laboratories at the University of Rochester Medical Center (URMC), “We were basically operating in the mode of Alice’s Restaurant,” says Peter J. Sims, MD, PhD, professor of pathology and laboratory medicine. “Our doctors could write anything on a piece of paper and we would find the test and where to send it out.” When he came to URMC in 2006, send-out testing was increasing exponentially. “Literally,” he says. “In the period of 2006 to 2009 send-out test volume and costs were nearly doubling on a yearly basis.”

No more, however. URMC’s laboratory, like others CAP TODAY interviewed, has found that a well-formulated strategy for curbing send-out testing can bring large savings to the hospital without negative impact—and, in Dr. Sims’ view, with even a net positive impact on the quality of patient care.

Not long ago, URMC had plasma thiamine or B1 vitamin test on its menu as a send-out to ARUP Laboratories. “In its analysis of tests ordered, ARUP pointed out that our doctors have ordered annually between 1,500 to 2,000 of these tests. And it’s the wrong test for evaluating thiamine blood deficiency. The test that should be ordered is whole blood thiamine. Because most of the body’s stores of vitamin B1 are intracellular, and most of the blood thiamine is carried in red cells, the plasma level has a much lower diagnostic specificity and sensitivity. So the chance of error of diagnosis based upon plasma thiamine is quite high.”

“After that information came to light, we changed our test menu immediately,” Dr. Sims says. “We don’t offer plasma thiamine anymore.” He credits ARUP with uncovering the problem through its laboratory analytics program ATOP (Analysis of Test Ordering Patterns). “They’ve alerted us to a considerable number of send-out tests that were ordered through our labs that had no clinical value or were purely of research interest.” In reviewing ATOP analyses of institutional ordering patterns, Dr. Sims has found: “ARUP has in virtually all cases correctly identified those tests that were misused and/or overused by our physicians.”

The ATOP approach has meshed nicely with URMC’s stepped-up focus on cost consciousness and quality of care. “The labs are owned by the hospital, which is a nonprofit component of the university. The health of the entire medical center and university is dependent on the fiscal health of the hospital. When we looked at the numbers, there was a significant drain of dollars attributable to unnecessary send-out testing ordered through the clinical laboratories.”

Gathering and communicating useful data have been enormously helpful in bringing best practices to bear on send-outs, Dr. Sims says. One of the obvious factors: The hospital’s interns and residents were ordering a large number of send-out tests, particularly on hospital inpatients. “That’s typical, and anyone who has been through a residency experience knows that one of the high motivators is to show how smart you are, and cover yourself when the attending physician comes through and makes rounds, to show you’ve thought of all the issues. So house staff are motivated to look for every possible diagnostic decision tree to go down and order the corresponding laboratory tests that might be appropriate for each hypothetical disorder conceivably affecting the patient.” When he and others looked at the orders house staff placed, “it became clear that very often the attending physician would not have ordered the tests. And the turnaround times for many of the tests were such that the results would have no impact on diagnosis or clinical management of the patient during his or her stay in the hospital.”

In 2010, for the first time, URMC created a laboratory diagnostics committee to oversee testing and went live with a new policy.

“First, we restricted send-out orders placed on inpatients at Strong Memorial Hospital by requiring the attending physician to approve the order,” Dr. Sims says. “Originally, this was a paper form that the attending physician had to fill out and sign to approve a test order; it included a statement that send-out tests are performed by third-party labs outside of URMC and the turnaround time might be anywhere from several days to many weeks.” The attending had to confirm the test was required for diagnosis or medical management of the patient during the current admission.

Simply the requirement to sign this form reduced the volume of send-out tests by about 30 percent right away, Dr. Sims says. “There was very little pushback. I would say most of the complaints had to do with the paper form, but not with the requirement that the attending had to certify the medical necessity of the test.” To reduce the impact of this new restriction, the top two dozen send-out tests by volume were excluded from the requirement. “We figured the less we had to bombard the attendings with forms to sign, the more they would accept the policy.”

When Strong Memorial’s EMR was implemented, a “hard stop” in the online ordering component of the EMR replaced the required paperwork. “If the provider who is logged in is not the attending physician of record for that patient, the order is never seen by the laboratory or the phlebotomist. It automatically bounces into the inbox of the attending for their approval or disapproval, by electronic certification, of the medical necessity of the test.” Once the EMR ordering was in place, the laboratory moved the two dozen excluded high-volume send-out tests into the required approval process. “And nobody even noticed. There was zero complaint about it, and we’re now down about 50 percent in total volume of send-out orders for inpatients” at Strong Memorial.

URMC’s lab diagnostics committee (which consists of the chief medical and operating officers and many of the chairs of URMC’s clinical departments) launched a new strategy Dec. 1: splitting the test menu for all send-out tests, inpatient and ambulatory, into three tiers. “In essence, we modeled ourselves after the hospital’s pharmacy and therapeutics committee in setting tiers of restriction in its formulary of drugs,” says Dr. Sims. Tier one lab tests are those that the committee has determined are of universal medical necessity—for example, thiamine, plasma metanephrines—that any practicing provider should be able to order. Most of their send-outs fall into this category. In tier two are restricted tests, defined as tests so nuanced as to require the provider to be subspecialty trained and qualified and to hold a staff appointment at one of the hospitals.

Tier three tests: “off-formulary and not offered in our test menu,” Dr. Sims says. The gastroenterology subcommittee decided, for example, that panels for inflammatory bowel disease are off-formulary, and the neurology subcommittee decided the same for complete gene panels to diagnose neurologic disorders. “We do permit physicians to request tier three tests on a one-time, patient-specific basis, but in this request they must spell out the medical necessity for the test, and before the order is processed, the request must be reviewed and approved through the laboratory diagnostics committee process.” A relatively small number of exceptional requests for tier three tests are approved.

But back to tier one. The appropriate test to rule out pheochromocytoma is plasma metanephrines. ARUP had alerted Dr. Sims and colleagues that providers were ordering a large number of inappropriate, presumably screening, tests for pheochromocytoma, among them fractionated catecholamines or vanillylmandelic acid. “These are tests that are quite nuanced as to when it’s appropriate to order them. In fact, according to ARUP data, the vast majority of tests ordered were likely to be the wrong ones, so our endocrinology group decided these latter tests should be restricted to subspecialty-trained providers.”

HCV RIBA testing is another category of test that the ARUP Laboratories’ ATOP program has found is ordered more than it needs to be, says Edward R. Ashwood, MD, president and CEO of ARUP Laboratories. “HCV RIBA is a supplemental test that shows you a spectrum of antibodies to hepatitis C, and if you have the right pattern, you can determine that the person has had hepatitis C in the past. But it doesn’t tell you if you have an infection now, which is really the question the doctor wants answered.” ARUP has found that the vast majority of HCV RIBA testing is not needed. “The doctor can start with a simple screening test for HCV antibodies, and most of our clients do that test themselves. If that’s positive, you shouldn’t reflex to hepatitis C RIBA. Instead, do a hepatitis C RNA.”

Based on his experience, inpatient testing is not as big a factor in runaway reference testing costs as is outpatient testing, says Mark Lifshitz, MD, director of clinical laboratories for New York University Langone Medical Center. “Inpatient testing accounts for only 25 percent of reference lab testing both by volume and by cost. The lion’s share of the action is on the outpatient side, and the dynamics there are quite different, and so are the strategies.” For example, he points out, if it’s an outpatient test, “I may not be able to tell the doctor I’m not doing the test or it’s not appropriate, because the doctor may then decide in any case that he or she will take all the other work somewhere else that will do the test.”

Dr. Lifshitz’s laboratory has for many years reviewed reference volume and cost for inpatient tests in general. “We have a detailed list of every single test—exactly what we’ve ordered in the last 12 months and the associated costs—so that gives us a baseline and we can monitor trends. If we see one test increasing in volume, then we can pull out the statistics for that test and take a more granular look at who’s ordering it and why. It may need to be increasing, or we may need to rethink our guidelines for usage.”

The basic concept the lab employs to control inpatient send-out testing is simple: Use your information system to optimize appropriate utilization. “You can’t order a test if it’s not coded in the system,” Dr. Lifshitz says. All tests above a certain cost are filtered out and referred to him or one of the other pathologists. “Most of our inpatient reference lab costs are tests that have clear indications for why they’re being done, but there are some new genetic tests and neurology tests that are usually not coded in the system because they’re sporadically ordered or new. So if I’m a clinician, I’ll need to call the lab because I won’t see the test in the test catalog. That’s the trigger for how we’ve established a connection between the physician and the lab. And someone’s going to follow up with the physician to explain this test isn’t available at our lab, it has to be referred out, and it costs X amount of money, you have to get clearance from Dr. So and So, and so on. That’s how we’ve been able to filter some tests out.”

Most clinicians, in fact, don’t have any idea these tests can cost in the thousands of dollars. “When I’m talking to them about cost, the No. 1 thing I find is they think the cost of the test is reimbursed; if it costs $1,000, they think the hospital is getting that, so what’s the downside of ordering it? When you explain there is no reimbursement for these tests, the test that seems so important becomes a bit less important in their minds, and most are fairly open to reviewing the appropriateness or need at that point.”

In addition, despite labs generally being able to negotiate discounts with reference labs if they commit to referring enough tests, for some specialized tests the net gain is small. “If someone’s discounting a $1,000 test to $800, it still costs me $800.” So there’s not that much benefit to the discounts, Dr. Lifshitz says. “On the other hand, some reference labs will give 20 percent off everything and you can pick 20 tests on which you’ll get 40 percent off.” Would he pick the more expensive genetic tests for the 20-test list? “I might or I might not, because at the end of the day, if I’m sending out a high volume of a certain test, the total savings there might be far greater than a discount on a low-volume genetic test.”

Dr. Lifshitz counts spreadsheet analysis as one of the most useful tools to help make such decisions. “You can talk whatever utilization strategies you want, but at the end of the day, the No. 1 thing someone needs to look at is the data.” It’s a task he performs monthly with two aims: “One is to identify trends in what tests are being ordered with greater frequency. And we also decide which tests have increased in volume to where we want to consider bringing them in-house.” In-sourcing even a relatively inexpensive test such as vitamin D, which the medical center is now considering, can reap savings because it is one of the high-volume reference tests.

But reference lab testing is not a static environment, he notes. “It’s dynamic and you have to constantly keep on top of it. Let’s say you are doing a hemoglobin evaluation test, and the volume was originally a hundred a month and for whatever reason it dropped to 20 a month. Perhaps a particular physician who was the primary orderer of the test left the medical center. You may decide the volume no longer supports performing the test in-house, provided you don’t need a result the same day for clinical reasons.”

On the other hand, the indirect costs of sending tests out must also be taken into account, Dr. Lifshitz says. “There’s a certain level of busywork and a lot of costs are hidden. You have to separate the sample when it’s being sent out, package it, and track testing status with your reference lab, though generally you have an interface to report results back. The cost of all that can be significant. It’s easier to run tests as they come into the lab, especially if they are available on existing chemistry and immunodiagnostic testing platforms.” Similarly, though many think they can reduce the cost of doing testing in-house by holding tests and batching them a couple of times a week, “you might be better off by sending it to a reference lab that’s going to do the test every single day.”

One of the considerations his lab faces regularly is whether to purchase a new analyzer to bring a test in-house. “Sometimes we’ll hold off on bringing a test in-house, because if we do it right away it means we have to purchase a new analyzer, but if we wait a little we might be able to run it on an existing analyzer.” This is because of the natural evolution of some specialty tests, which may be available initially only on one vendor’s analyzer but will later be featured on other vendors’ equipment, as is happening now with vitamin D.

“That’s unlikely to occur with genetic tests, however. There are a fair number of molecular tests that are coming into the microbiology section of the lab, but I don’t see a lot of genetic tests coming into the clinical lab anytime soon,” Dr. Lifshitz says.

Despite these causes for hesitation, ARUP Laboratories has been seeing robust in-sourcing recently, especially for tests like vitamin D, Dr. Ashwood says. “There are a lot of tests coming on the market right now that are available to our clients’ platforms, and they’re anxious to have those tests in-house. But we also do some very simple tests. Aldolase, for example, anybody can do, but it’s a huge test for us as a reference lab because hospitals do not have enough clinical demand for the test to justify having it up and running in their lab.”

Robert G. Gurdak, MD, medical director of the Department of Pathology at Trumbull Memorial Hospital (TMH), Warren, Ohio, says financial necessity was the driving factor in his laboratory’s move to modify the hospital’s reference testing strategy. TMH was part of the Forum Health system in northeast Ohio, which included two acute care hospitals and a rehabilitation hospital. All of the hospitals emerged from bankruptcy in October 2010 after being purchased by a 130-hospital chain, Community Health Systems.

“The bankruptcy forced us to scrutinize all of our costs, from test tubes to our reference lab testing bill,” Dr Gurdak says. Before filing for bankruptcy, the hospital system brought in a major turnaround firm. “The consultants put us through a rigorous process resulting in specific recommendations, such as consolidating certain tests in one of the two hospital laboratories, as well as what reference laboratories should be used for our send-outs.” Choices and recommendations were on the table that under ideal circumstances his laboratory might not have made, he says.

Now, after the reorganization, the laboratory continues to monitor the volumes of reference lab testing from inpatients and outreach clients to determine whether to reference it out or perform it in-house, says Dr. Gurdak. Vitamin D assays recently were part of this evaluation process. “Molecular testing also continues a steady rise, but in some instances instead of ordering a test right out of the gate—in other words, as soon as we see a tumor in the patient—we order it down the road. Typically we’ve allowed the oncologists to help us make those decisions.”

This runs somewhat counter to the idea of positioning pathologists front and center with the health care team and making them more proactive, he says. “But the problem with being proactive is that in the inpatient setting, this testing may get wrapped into a prospective [DRG] payment, and that would just drive up the cost of analyzing the tumor. For breast cancer patients, most of whom are outpatients, our pathologists order estrogen, progesterone, and HER2 studies at the time of diagnosis. For lung cancer patients the molecular tests don’t have to be done right away, at least in our community. The patients are recovering from surgery or undergoing metastatic workups, and the oncologists are probably not going to make a decision on chemotherapy or other treatment for several weeks.”

The splitting of the technical component from the professional component has been a good compromise to assist in the make-buy decision for some testing, Dr. Gurdak points out. “We do a fair amount of breast cancer testing in our community each year, but it’s not hundreds of cases, and after looking at the HER2/neu validation requirements, I didn’t feel it was worth the time and effort to do the technical component in-house. I don’t have the personnel in place to do these tests that are really borderline numbers, so we allow our partner laboratory Clarient to do all the validating and monitoring of the assays, and they communicate when something drifts or goes awry. We’re kept abreast of that. But, except for HER2 by FISH, our pathologists still maintain the professional component.”

How TMH handles utilization on the anatomic pathology side is not unique, Dr. Gurdak says. “If we get a specific test for, say, an Oncotype breast cancer test, we look to make sure it’s an appropriate order for what we know about the patient. Every once in a while we find that the oncologist might not order the test correctly, so we intercede. The pathologists sign off on every one of those tests before they’re sent out, and that’s one of the ways we manage utilization.”

Hospital practices on reference testing range all over the map, notes ARUP’s Dr. Ashwood. “Every hospital send-out area can be either very complicated or streamlined. I’ve talked with clients that work with 100 different labs and clients that work with just us.” From his point of view, of course, the streamlined arrangement is preferable—and not only because it’s simpler. When ARUP Laboratories needs to refer tests out, “that tells us there’s demand for a test we don’t do, and we very frequently have our medical directors look at our top 30 send-outs to see if there is something there that looks like good medicine and whether, if it has intellectual property tied to it, we can get a license to perform it.” Predicting demand for tests is always a tricky proposition, but recently the strongest candidates have been in two main areas: molecular genetics and molecular oncology. “There’s been much more robust growth in those areas than in everything else,” he says.

Clinicians don’t always understand how reimbursement of lab tests works, Dr. Ashwood agrees. “There are a lot of physicians who don’t know what things cost, who confuse wholesale with retail, and who don’t understand that the list price isn’t what hospitals get, that payers negotiate discounts off list. The typical doctor will not be able to tell a patient what the insurance company will pay for a test, and that leads to a lot of misinformation and misunderstandings.”

ARUP Laboratories is trying to use information technology to help hospitals with their utilization strategies. “Our informatics director is working with his hospital IT counterparts to put decision support into the hospital physician order entry systems,” Dr. Ashwood explains. They’re working with Epic Care and Cerner PowerChart, and including short utilization messages, where appropriate, combined with hyperlinks to online order guidance. “So if you want to drill down and get more information, you can. We’re trying to build these for all the hospitals’ tests, not just the send-out tests; we’re not distinguishing between send-out tests and in-house.”

As the nation gears up for broader coverage of patients under the Affordable Care Act, the public and clinicians need to ally with laboratories in keeping reference testing under control, Dr. Sims emphasizes. “The public needs better understanding that misuse and overuse of diagnostics are not only expensive to the health care system but also bad medicine,” he says. “And that has risks associated with it. Most important, the engagement of clinical leadership and clinicians in the process of checking utilization is essential. If you’re going to be successful, the users must be the direct participants in defining best practices in laboratory diagnostics.”