College of American Pathologists
Printable Version

  Feature Story





cap today

Not Mayo? No matter—smaller labs go molecular
What St. Luke’s did well
What St. Luke’s might have done better July 2004
Karen Lusky

Community hospitals that want to ride the wave of the molecular testing future without busting their budgets might do well to take lessons from St. Luke’s and Hartford hospitals’ decade-old molecular pathology labs, both of which had tenuous starts and steep learning curves.

That was a key message delivered in separate presentations at the Executive War College in April by Gregory Tsongalis, PhD, former director of the molecular pathology laboratory at Hartford (Conn.) Hospital, and Jane Rachel, MT(ASCP), MA, who manages molecular diagnostics and flow cytometry at St. Luke’s Hospital, Kansas City, Mo.

Community hospital labs that feel daunted by the prospect of putting molecular testing into place might take heart from their tales. Once St. Luke’s gave the go-ahead for a molecular lab in 1994 and $100,000 in startup funding, Rachel spent a week at Mayo looking at its molecular lab and "another week being depressed that we weren’t Mayo," she quipped in her presentation. Then the laboratory team got down to business, figuring out how to mirror Mayo as best it could.

Today the lab performs about 24,000 molecular tests a year, up from 119 in 1995. The test menu consists of infectious disease and genomic assays, and the lab has four full-time equivalent employees who perform the tests during the day shift only. "We are doing 13 different assays and have a number in development along with a list of ’wannabe’ assays," as Rachel puts it.

At Hartford, the startup money for the molecular lab vanished soon after Dr. Tsongalis signed on there in 1994, right before the hospital suffered an unexpected budget cut. Forced to make do with a broom closet initially, Dr. Tsongalis began informal quid pro quo negotiations for funding with the directors of other departments by saying, "This is what a molecular lab can do for you ... and by the way, do you have $9,000 for a thermocycler?"

Today Hartford Hospital’s molecular pathology laboratory is doing well to the point it has outstripped its resource allocation and bumped up against the fiscal constraints of the core laboratory. In that sense, "the molecular lab may be a victim of its own success ... but that’s also the nature of the field, which is exploding with new technologies and demand for the testing," says Dr.Tsongalis, who moved in February of this year to Dartmouth-Hitchcock Medical Center in New Hampshire to develop its molecular pathology laboratory.

The bargaining chip of what a molecular pathology lab can do for a hospital continues to gain value. "Cost savings will vary considerably from one hospital to another," Rachel told CAP TODAY in a followup interview, "but they are derived from shorter turnaround times, higher reimbursement for molecular tests, and reduced send-out costs."

For example, enterovirus detection using an amplified method can tell whether a patient with meningitis has a viral form of the condition and can be sent home safely from the emergency room—or needs inpatient care and intravenous antibiotics. St. Luke’s, which began offering such testing in January, has calculated that "the average length of stay for someone with suspected meningitis is 2.5 days at a cost of $16,000," says Fred Plapp, MD, PhD, medical director of St. Luke’s Regional Laboratories.

"Yet by performing viral PCR testing, you can identify patients with viral meningitis who can be sent home with pain medication," he says, whereas bacterial culture of spinal fluid takes about three days. At that point, the patient has already received unnecessary IV antibiotic therapy in the hospital and been sent home on oral antibiotics. Viral culture takes 10 days to ante up the answer.

To balance the cost-benefit equation, molecular pathology laboratories have to adopt what Dr. Tsongalis calls his "pizza business model," gleaned from years of watching how his family’s pizzeria turns a profit.

"Pizzerias offer the super deluxe pizzas with all the toppings, but that’s not where they make their money," he told War College attendees. They make money on the cheese pizzas because they’re cheaper to produce and sell in volume, and the same holds true in the molecular lab. "You offer your high-volume, basic tests to help offset the labor-intensive, highly technical tests that you only do a few of but need to do in order to support physicians at the institution," he says.

Hartford Hospital’s molecular pathology laboratory was able to pay for the "deluxe" tests by providing a core menu of volume molecular testing, such as HIV, hepatitis C, chlamydia, and gonorrhea. "For the most part, even though the reimbursement for CPT codes [for molecular diagnostic testing] is not as much as labs would like to receive," Dr. Tsongalis says, "we are getting to the point where we can start to cover costs based on the volume of tests." As a result, he adds, molecular labs can generate revenues for some institutions.

To make a molecular lab work, you have to debunk the "false truth," Rachel says, that a lab shouldn’t do unprofitable tests as part of a larger plan to become profitable. "You might talk about doing a BK virus in-house for $200 when you only get $50 reimbursement for it, because it costs $400 to send it out," she says. "So losing less money can be a worthwhile goal." St. Luke’s molecular lab this year began offering enterovirus testing off season as a loss leader to garner business during the part of the year when meningitis cases occur most frequently.

For the lab to grow, new tests have to be added to the menu. At St. Luke’s, tests end up on the "wannabe" list in a number of ways. For example, Dr. Plapp and other pathologists have frequent informal conversations with physician clients about their needs for diagnostic testing. Or St. Luke’s lab will add a test to the menu to comply with standard of care or a new set of practice guidelines, such as that for cystic fibrosis mutation screening. Sometimes competitive pressure forces the lab to consider offering a test—for example, if a client threatens to move all of its testing to a lab that can offer a molecular test St. Luke’s isn’t doing.

The molecular laboratory also monitors the volume of testing sent to an outside reference lab to look for a spike in testing that would signal a change in practice. "Or we might see a gradual accumulation of test volume that indicates a test is now worth looking at in terms of bringing it in," she adds. Scientific journals, professional meetings, and the New York Times and Wall Street Journal are other sources of ideas.

While offering a new test may sound prudent initially, the proposition has to pass the "exhaustive gauntlet" of the St. Luke’s molecular laboratory team members, who bring to bear their scientific, clinical pathology, business, and marketing perspectives. "Anyone on the team has veto power," Rachel says.

For example, St. Luke’s molecular pathology laboratory has found in some cases that a test that makes sense from a technological perspective may not be used by clinicians. An example: the first amplified assays for Chlamydia trachomatis (CT) and Neisseria gonorrhoeae (NG), which became available in 1998. Rachel recalls the lab was excited about the new technology until the marketing person said, "This is great, guys, but how am I going to tell docs that we can offer them testing that is more expensive and not a whole lot more sensitive, but, hey, it’s way cool?"

There was another "stop" point for CT and NG: The lab would have had to keep its nonamplified assay to avoid forcing a clinician into a more expensive test. And offering both would have posed a logistical nightmare, Rachel says.

The lab did implement CT and NG amplified testing in 2000 because the cost of the test had come down. "We were at a point where we could tell our clinicians that we are upgrading testing to an amplified probe technology at the same cost. The Clinicians were really excited about that, so we went live with the testing," Rachel says.

Yet the molecular laboratory team decided not to take advantage of new technology this year that would have allowed it to do CT/NG testing from ThinPrep Pap specimens. For one, there were specimen integrity issues: The molecular lab was worried about potential contamination of the specimen on the ThinPrep instrument. They could remedy that by pulling off an aliquot of the specimen before it underwent cytology testing, but then someone pointed out that specimens might come up short if they required human papillomavirus testing. And the lab knew the clinician would not be pleased at having to call the patient back in for another Pap test.

As an alternative, the molecular lab recommended its clinician clients obtain a swab at the time of the ThinPrep Pap to provide a good, clean sample for DNA testing for CT/NG. "It’s not going to inconvenience the woman to get a swab sample because she’s already in the office for a Pap smear," Rachel says. "Our thinking was that we wanted to have separate samples for CT/NG—and not use the ThinPrep specimen—for a lot of reasons, and one was that we didn’t want to link HPV and CT/NG testing. We are talking about different groups that should be screened—under age 30 for CT/NG and over 30 for HPV with abnormal Pap smears."

Dr. Tsongalis’ approach to bringing in a new test is similar. "There are about a dozen molecular tests that everyone [offering such testing] does," he says. They are chlamydia, HPV, HIV, HCV, gonorrhea, Factor V, Fragile X, hemochromatosis, cystic fibrosis, B and T cell gene arrangement studies for lymphoma, Bcr/abl (for leukemia), and HER2 (for breast cancer), among others. "Beyond that, each hospital’s lab has to come up with specific tests to address the physicians’ particular needs," Dr. Tson gal is says. "That becomes important if you want them to buy into the use of the lab."

Hartford Hospital, for example, serves large populations of Hispanics and Asiatic Jews, so the molecular lab had to tailor its genetic testing to those groups. In addition, the lab has to ask if a test meets both inpatient and outpatient needs. "Sometimes these can be like day and night ... in terms of what physicians want and are willing to accept," Dr. Tsongalis says. And finally: Is the test expandable or adaptable as new technologies become available, and how quickly can the lab accommodate that when it occurs?

Hartford Hospital learned a lot about the technology as its molecular lab matured. "One is that trendy science doesn’t necessarily bode well for good diagnostics," Dr. Tsongalis says. "You can read the greatest and latest in the New England Journal of Medicine and that may be suited for some institutions. But some of these technologies are almost impossible to bring into certain clinical labs." And just having technology available doesn’t mean the technology’s performance will meet the lab’s standards. "It takes a lot of evaluation work and validation to bring those technologies on board, so you have to be careful with that," Dr. Tsongalis cautions.

Molecular laboratories need to choose instruments that have "fantastic performance," in his view. You need to know: Is the equipment going to meet your needs and produce desired turnaround times? For example, when Dr. Tsongalis arrived at Dartmouth in February, the turnaround time for Southern blot analyses was two to three weeks. "With a simple change in chem is tries, we dropped that down to four days," he says. "The through put has to be there because no doctor in the world wants to wait two to three weeks for results. The stakes are too high with liability and patient-centered care."

To keep test results flowing, the instrument has to be a reliable workhorse, Dr. Tsongalis says. But vendor reliability is also important. "Will the vendor provide the support if the equipment crashes or you need another tech trained or have any issue?"

Labs also have to consider data output in terms of how easy it is to view and interpret. The user friendliness of the instruments and the software can be a deal-breaker. "A lot of older medical technologists [doing the testing] aren’t comfortable with new software packages," says Dr. Tsongalis, "and that can be a showstopper for a lab bringing in some type of technology."

The true cost per test result can be a showstopper, too. And in that regard, relying on the vendor’s projections, which provide only a ballpark figure, can put a lab out of the game. Rachel explains: "If you are going to buy a set of reagents that costs $1,000 and get 10 tests, it’s a logical assumption that the cost per test is $100 ... but that’s not really the case." Reagent cost is determined by the frequency of testing, number of samples, linear versus clinical range, assay reliability, and reagent packaging, stability, and lot variation.

"Some assays have a given number of vials of reagents and once you open one or reconstitute it, it has a shelf life of about 24 hours," Rachel says. "So if you have a vial of PCR master mix that will do 12 samples, and you’re doing only one run with five samples, you’ve wasted seven tests."

Labor costs are determined by frequency of testing, number of samples, and whether the test is automated. "When vendors talk about one-hour PCR or one-hour molecular testing they mean one hour from the point you put extracted DNA mixed with master mix on the instrument," Rachel says. "It’s not one hour from the time you get the specimen."

Reporting test results for clinical trials typically involves more handholding and personalized service than do other tests, which costs more. Rachel advises labs to figure out all of the factors that affect the cost per patient result ahead of time—not after the fact.

Batching, of course, can lower the molecular lab’s cost per patient result, if the lab has the volume to do it. In fact, St. Luke’s molecular lab found batching saved the day in its effort to provide quantitative cytomegalovirus viral load testing, which it began offering two years ago for the hospital’s cardiac, renal, and bone marrow transplant programs. Rachel says that experience shows that managers have to question conventional wisdom if they want to keep molecular testing in the black.

Here’s what happened: The molecular lab agreed initially to do the CMV testing on Mondays, Wednesdays, and Fridays to produce desired turnaround times for clinicians. "The lab decided its medical technologists wouldn’t start the assay until 1 PM, which would give physicians time to do their rounds and order the testing, and the lab could report results at 5 AM the next day," Rachel says.

But almost from the beginning the lab found it was wasting expensive reagents on short runs. In addition, samples would arrive at 1:30 PM accompanied by phone calls begging the lab to run the late test. Rachel found herself giving in to these pleas because she knew that if the test wasn’t run on Monday, the physicians would not get a result until Thursday morning.

In time, the lab realized it had a stalemate on its hands: It was losing money on the test and physicians weren’t happy with the service.

So the lab decided to raise the price to cover reagent wastage. But that still left the problem of extra runs and gaps between the days when the test was offered. The lab thus moved to plan B and abandoned the test schedule, even though that approach seemed to be counterintuitive in terms of meeting clinicians’ requests for consistent turnaround times.

"We figured out that we could do 11 tests using the same reagents without wasting any. So we decided that every time we received nine [samples], we’d do a run—whether it was 8AM or 1PM," she says. And the lab never ran more than 11 samples because that’s what it could do without wasting reagents.

The lab prioritized the samples and never bumped a test for an inpatient, a patient who’d had recent positive results, or a stat test. The strategy paid off in spades by decreasing turnaround times by 20 percent and reducing annual reagent costs by $68,370. Most dramatic of all: Lab staff can’t remember the last time they got a phone call with a plea to do a test or had to authorize an extra run.

Of course, if the lab’s turnaround times get too long because of batching, it will have to do short runs to meet patients’ needs. That happened recently when St. Luke’s clinicians began switching to HCV quantitative testing because it’s based on real-time PCR and is more sensitive than qualitative testing. "As a result of the change, the volume of HCV qualitative testing has decreased and the turnaround time waiting for the magic number [of samples] is increasing," Rachel says. The lab will have to raise the price of the HCV qualitative test to cover the cost of doing short runs, which will become a deterrent to clinicians ordering it.

Converting from conventional to real-time PCR provides a huge turnaround time advantage, Rachel says. "That’s another thing we did, and then analyzed what we had done to ourselves after the fact" in terms of the cost, she adds. The lab acquired two Roche LightCycler real-time PCR instruments (which take readings as they amplify, unlike conventional PCR). Much to its delight, the lab found that even though the reagent costs for the LightCylers are higher, the labor costs are lower and the turnaround times about four times faster than that of conventional PCR. "So there was only about 20 cents difference between assays," Rachel says.

The lab also found that the LightCycler picked up another mutation in the prothrombin gene (at position 20209 on the gene, first reported by the Cleveland Clinic). "And every patient in whom we have identified the genetic mutation has had a serious thrombotic episode," Rachel says, "so the clinical impact of that mutation appears to be significant." St. Luke’s is studying the prevalence of that mutation in its patient population.

Dr. Tsongalis agrees that the LightCycler and other companies’ instruments, such as Cepheid’s Smartcycler, make it possible for labs to do molecular testing on a nearly stat basis. Labs should weigh the benefits of the better turnaround times, he says, against the higher cost and data output issues. "They should also assess their preparedness for more automation than is typical, from validation to interpretation," he adds.

Molecular testing technology is actually headed in a simpler direction. "Down the road, instruments like the Nanosphere Verigene System and Third Wave Technologies don’t require any type of amplification in the assays because they are chemistry and not necessarily molecular. This makes assays easy to perform because we’re just talking about liquid handling." He adds, "With some of the nanotechnologies, we can get faster, cheaper, more precise, and simpler, which means a lot of medical technologists will be able to do them easily."

It all sounds leading edge, but St. Luke’s Health System and Hartford are community hospital systems with resources and referral bases that far exceed those of smaller community hospitals. And that raises the question: How feasible is molecular pathology for single community hospitals?

The notion that molecular testing is beyond the capability of community hospital laboratories ranked at the top of Rachel’s list of false truths. "Molecular testing is becoming so pervasive that before long, almost every hospital is going to have to have at least familiarity with the testing," she told CAP TODAY. Rachel predicts that most hospitals will be doing the testing "in not so many years."

Dr. Tsongalis sees two hurdles for community hospitals that want to develop a viable molecular lab. "One is the level of complexity of the technology, which is such that you might not have enough qualified people at a smaller hospital," he says. The lack of qualified clinical laboratory scientists is a huge problem for even the large urban hospitals.

The second is whether the hospital can justify doing molecular testing based on volume. Hartford Hospital, for one, had an extensive outreach program through its subsidiary core laboratory, and St. Luke’s Regional Laboratories has 300 active clients within a 100-mile radius.

Yet defining a threshold for the volume of testing required to make a molecular lab work depends on numerous factors. "These include whether the hospital does outreach testing, the technology selected, and the lab’s relationship with its vendors," Dr. Tsongalis says. The hospital also has to have pathologists willing to buy into the program and serve as the interface between the lab and clinicians in marketing the testing. "So the rules [for making it work] are not cast in stone," he says-and the management techniques absolutely can’t be."

Karen Lusky is a writer in Brentwood, Tenn.