Vanderbilt Lab Uses Predictive Medicine to Improve Care

Clinical Lab 2.0 transition uses pharmacogenetics, formularies, and diagnostic management teams

> CEO SUMMARY: The integrative diagnostics lab at the Vanderbilt University Medical Center aims to use sophisticated diagnostics to advance the use of precision medicine testing to improve patient care and to do so at an affordable cost. As part of these efforts, the lab staff seeks to predict how patients will metabolize medications and then provides that information in the form of clinical decision support to their treating physicians. By integrating clinical decision support into physicians’ regular workflow, the laboratory is helping to improve patient outcomes.

As the healthcare system moves away from the fragmented, fee-for-service  method  of  payment, clinical laboratories and anatomic pathology groups are paid for delivering value defined by improving patient outcomes and controlling costs.

In the United States, only a few clinical labs and pathology groups are operating in integrated healthcare systems under value-based payment arrangements. One of those is the lab at the Vanderbilt University Medical Center in Nashville. At VUMC, the clinical laboratory serves an integrated health network and delivers added value through a three-part clinical strategy.

The initial elements of this multi-part strategy were added in 2010, and since then, the team at VUMC has added innovative clinical services in a step-wise fashion, according to pathologist Mary M. Zutter, MD, VUMC’s Vice President for Integrative Diagnostics. Zutter explained how the VUMC lab delivers added value during a presentation at The Dark Report’s Precision Medicine Institute in New Orleans last spring.

The first of these initiatives was the predictive medicine and pharmacogenetics testing program that VUMC implemented 10 years ago to demonstrate the utility of prospective genotyping. In this program, the laboratory uses prescribing data from prospective genotyping as the basis for decision support for prescribing physicians. The lab staff adds the results of its prospective genotyping tests into decision-support tools built into the electronic health record (EHR) system.

The clinical decision support was developed by a team of clinicians, laboratorians, and informaticians. Prescribing physicians use those decision-support data to guide the selection and dosing of prescription medications for individual patients.

The second of these programs is diagnostic    management    teams.    Working in DMTs, the lab staff—including pathologists and lab professionals—collaborate with treating physicians and other providers to improve diagnostic accuracy while cutting the average time physicians spend on diagnosis.

The third program is VUMC’s laboratory test utilization program  in  which  the lab uses a test formulary to eliminate unneeded tests. In an impressive example  of the effectiveness of the formulary, the lab published data on how the formulary reduced orders for costly vitamin D2/D3 fractionated tests.

After introducing the formulary in 2014, the number of these test orders dropped from almost 600 each month in 2014 to zero by January 2016. Labs in every hospital and health system could use this same approach to generate substantial savings.

While each of these three lab initiatives is distinct, they are part of an integrated strategy to improve patient outcomes and control costs.

Note, for example, that the decision support system built into the EHR also supports the diagnostic management teams, and each DMT is useful in eliminating unnecessary tests. In turn, all three of the strategies support the lab’s efforts to work with all clinicians to manage lab test utilization effectively.

In this way, the lab’s strategies are consistent with the principles of the Clinical Lab 2.0 model. In this model of lab management, pathologists, lab directors, and clinical lab scientists use laboratory data to produce clinical insights to improve patient outcomes and support care delivered in value-based healthcare systems.

“Our laboratory’s vision is to provide effective precision medicine at an affordable cost, and that vision transcends any specific disease or clinical domain,” Zutter said. “That’s the direction our clinical laboratory and our physicians are moving toward, and it’s a powerful and unifying theme across all the different disciplines at VUMC. This also supports fuller integration of clinical care.

“Our laboratory has a team approach to care that begins by making a diagnosis of a patient’s condition,” she explained. “Then our laboratory and the treating clinicians follow that patient through therapy all the way to an assessment of that treatment.”

Lab Initiative One
Predictive Medicine and Pharmacogentic Testing

One of the most significant of the VUMC lab’s three-legged strategies is  the  use  of predictive medicine under  a  program called Pharmacogenomic Resource for Enhanced Decisions in Care and Treatment, or PREDICT. The members of the VUMC faculty who developed and implemented this program included Dan Roden, MD; Gordon R. Bernard, MD; and Josh F. Peterson, MD.

The aim of the PREDICT program is to assist clinicians in diagnosing a patient’s condition, selecting an appropriate treatment for managing the patient’s care, and then assessing the patient’s response to treatment, explained Zutter.

“In our predictive medicine program, we use  sophisticated  diagnostics to predict how a patient will metabolize medications,” she added. “Then we provide clinical decision support to allow  the treating physicians to do what is best for the patient without the need to spend much time analyzing all of the possible options for each patient.

“Early on, we realized that one opportunity in diagnosis was for our lab team  to better integrate our clinical decision support into the physician’s regular workflow to help produce better outcomes for patients,” she commented.

Prospective Genotyping

In 2010, when VUMC began its predictive medicine program, pharmacogenetics testing was not widely accepted, noted Zutter, who is also the Director of the Tumor-Host Interaction Program at the Vanderbilt-Ingram Cancer Center and the Louise B. McGavock Professor of Pathology, Microbiology, and Immunology.

“In September 2010, our PREDICT program launched its first initiative, and the VUMC team selected certain medications for at-risk heart patients,” she said. “Our lab tests predicted how each patient would metabolize drugs and helped improve physicians’ drug-dose selections.

“From the start of the PREDICT program, our lab was doing true predic-tive testing and the institution was not being reimbursed for this testing—in part because this program was an institutional initiative,” she added. “Today, our lab does reactive, or indication, testing and most of our insurers reimburse for that work.

Medical Home Model of Care

“Since that launch back in 2010, we have followed more than 52,000 patients in a medical home model of care, and many  of them receive a number of the drugs for which we test in our pharmacogenetics panels,” Zutter added. “We use prospective genotyping to identify those patients in the high-risk group.

“When we do this testing, we don’t test for a single drug-gene interaction (DGI),” she explained. “Instead, our lab does predictive testing with all of the drugs on our panel. That means we test for five DGIs at once to predict genetic risk.”

The five drugs and the corresponding genes being assessed are clopidogrel (for CYP2C19), simvastatin (for SLCO1B1), warfarin (for VKORC1 and CYP2C9), thiopurines (for TPMT), and tacrolimus (for CYP3A5).

“We found that 91% of these patients will have at least one of these risks,” she reported. “In other words, almost all of those patients will take one of these drugs and have a drug-gene indication showing that clinicians would change how they treat those patients.

“Once we had data demonstrating  that the pharmacogenomic test results indicated a need to  change  the  course  of treatment for those patients and that those patients had better outcomes, we decided over the past two years to expand the PREDICT program beyond those five drug-gene interactions,” she added.

Vanderbilt Integrated Diagnostics two

“Back when we launched the program we had PREDICT 1.0, which included only a few drug-gene pairs,” Zutter said. “Now we have expanded that effort to the PREDICT 2.0 program, which is a rapid expansion to 10, and soon, 16 drug-gene interactions.

“Our lab is expanding—both to serve the interests of providers and customers, but also to identify more drug-gene interactions,” she noted.

Adding Decision Support

“Our goal is to provide better clinical care, and we do that with clinical decision support, because if a patient is given instructions on a piece of paper and expected to take that paper the next time they go to their doctor, that system is destined to fail,” Zutter explained.

“Instead, it’s necessary to embed this information completely in the healthcare system—meaning in each patient’s EHR—so that all physicians can see those instructions for their patients and follow them at the point of care.

“Part of that effort involves not just giving clinical decision support about which tests should be ordered, but also how to report that information back to providers and to patients,” she said.

PREDICT Team Meetings

To expand the program, the PREDICT team members—including lab, clinical experts, pharmacologists, and informaticians—began meeting once or twice each month with large teams of clinicians, including pediatricians, cardiologists, and scientists.

One team would identify DGI pairs to add to the EHR and another would identify ways to expand the program to more clinicians throughout VUMC. “We wanted these teams of clinicians to identify their priorities for decision support,” Zutter said.

“At this time, we have focused on delivering a shorter turnaround time and lowering costs,” she commented. “The VUMC HealthIT team has expanded clinical decision support so that it’s available to every provider in our Epic EHR system, and also to every patient through the VUMC MyHealth portal.

“Therefore, treating physicians have the information and patients can take it back to their own providers,” she added. “Having that information, patients can talk to their physicians about what they need and what they want.

“To accommodate testing for that number of new prescription drug starts, we scale up our lab processes so that we can get the right turnaround time,” she explained. The goal is to reduce TAT from about five days to 24 hours.”

Lab Initiative Two       
Diagnostic Management and Decision Support

Not only does the lab need to manage more tests more quickly, it also needs to manage the vast volume of data these tests generate and deliver that data in a useful format to clinicians.

“As our laboratory starts putting data on functional and structural genetics, proteomics, and other effects of molecules into the decision-making matrix for every individual patient, whenever a provider needs to see that information, the amount of data can be overwhelming,” Zutter commented.

“Because we can’t overwhelm them with lab data, our lab team has made a huge effort to give them clinical decision support, which we define as an effort at integration.

“By that I mean, we aim to bring all of that information together, including information on the patient’s symptoms, the results of the physical exam, the lab test data, and the drug therapy results,” she explained. “Then, we put it all together in a summation developed by a team approach to care. In that way, our laboratory contributes to providing the best care to patients.

“This is where all of our efforts are leading us, and those efforts mimic what the diagnostic management teams do,” noted Zutter.

Traditional Test Order Model

“For most lab testing, the traditional model of ordering tests and getting results assumes the clinician is the expert,” she added. “In this model, the clinician orders the individual tests, the lab sends back results, and the clinician interprets those results. Then, the clinician may order more tests or prescribe drug therapy.

“This method is inefficient, costly, and yields limited quality results because clinicians cannot keep up-to-date on increasingly specialized tests and often are not ordering based on current evidence or guidelines,” Zutter explained. “Getting advice from pathologists and lab specialists takes time and that is not feasible for clinicians in today’s rapid-throughput patient care settings.

“At VUMC, the diagnostics management team is a collaborative effort among pathologists, clinicians, and biomedical informatics,” she added. “Under this approach, we use standard test ordering algorithms to develop the correct pattern of diagnostic testing for each patient.

Decision Support Core

“The DMTs have system-supported care with decision support that ensures clinicians follow evidence-based practice because decision support comes on the front end before the lab runs any tests,” she said. “Diagnostic experts in the decision-support core work between the clinician and the lab to ensure that all—and only—appropriate tests are ordered.

“In this front-end decision support approach, unnecessary tests are deleted, and essential tests are added, but only if needed,” Zutter reported. “Then, the lab runs the tests and the decision-support core intervenes again when it  receives the individual lab test results, interprets them, and issues comprehensive guidance to the clinician.

“Our approach to decision support helps the clinician and the patient under- stand the results of the testing,” she said. “The benefits are increased quality of care for the patient, because the patient gets improved decision making, a faster test result, and reduced costs because we have eliminated unnecessary tests and avoided hospital admissions or shortened  length of stay.

“In addition, we create a single, evidence-based and comprehensive report of integrated diagnostic data to guide therapy and disease monitoring,” she added. “That process allows us to improve the algorithms iteratively as evidence-based practices evolve.”

Lab Initiative Three
Lab Test Utilization

One of the goals of DMTs is to develop standard ordering practices that can be applied to a wide variety of tests, including next-generation sequencing for cancer. Thus, at VUMC, the lab can incorporate these guidelines to support effective laboratory test utilization.

“Using clinical decision support between the laboratory and the clinician, and by generating a comprehensive report that aggregates all of the data for clinicians and patients, we then wanted   to know if we could scale this program  to serve other clinicians in the medical center,” Zutter said. “For example,  can we apply the decision-support system to secondary testing standards for all hematologic malignancies? Right now we’re working with oncologists for breast and gastrointestinal cancer.

“We know from a report published in the American Journal of Clinical Pathology (AJCP) in 2013 by Adam C. Seegmiller, MD, PhD, and others, that the use of a DMT review makes it possible to identify the optimal number of tests for patients who need bone-marrow testing. The research showed that optimal testing based on an evidence-based, interdisciplinary team approach could save the U.S. health- care system about $500 million annually.” Zutter said.

The Vice Chair for Clinical Pathology, Seegmiller also is Professor of Pathology, Microbiology and Immunology, and Director of Laboratory Medicine and Hematopathology.

Working with other members of the clinical team at Vanderbilt, Seegmiller and Zutter did the research and wrote the AJCP article, “Optimizing personalized bone marrow testing using an evidence-based, interdisciplinary team approach.”

“From this work, we know we’re help- ing patients and providers, which is one of the keys to success for an effective lab test utilization program,” Zutter explained.  “If your laboratory wants to get provider buy-in for what it’s doing, your lab team must educate providers about how they and their patients will benefit.

“Anything that puts an additional burden on providers will fail. They can’t handle any more work, because they already have too much to do,” she commented. “That’s why we show them how using the DMT system could save about 10 minutes per-patient-per-provider at the front end, and an additional five minutes per-patient-per-provider at the back end.

Right Diagnosis, More Time

“What’s more, the providers then trust that they’ll  get  the  right  diagnosis  and  have much more time either with their patients or more time with their families,” she added.

“Another way we’ve improved lab test ordering is by developing a laboratory formulary committee,” Zutter said. “This idea was approved in 2014 to protect patients from the costs and possible consequences of inappropriate or unnecessary laboratory testing, particularly for the many new and expensive clinical laboratory tests where the clinical utility is not clear-cut.

“The committee decided that a laboratory test should be ordered only when  it is both medically necessary and likely to alter the diagnosis or treatment plan,” she added. “Using this process, we’ve been able to save about $2 million a year for the Vanderbilt healthcare system.”

Vanderbilt’s lab test formulary has three categories. The first category is unrestricted tests which are automatically run or sent out. The second category is restricted tests that are conducted or sent out if certain prerequisites are met. The third category is non-formulary tests that are for research or investigational-use only. For these tests, the lab will offer an alternative formulary test or ask the provider to cancel the order and discard the sample.

Lab Tests with Clinical Utility

The tests VUMC targeted for lab-formulary review included all tests with no clinical utility, all outdated tests, all high utilization and high-cost reference tests, all high utilization in-house lab tests, and all overused tests.

Under the lab’s inpatient medical-necessity review, the staff limited inpatient ordering of tests when the TAT was expected to exceed the patient’s length of stay because testing performed on inpatients should affect hospital care. Any  test with a TAT of four to seven days requires an attestation of necessity (called a soft stop), and any test with a TAT of more than seven days requires the medical director’s approval (a hard stop).

“In 2016, hard stops on 1,031 tests saved almost $260,000, and in that same year, soft stops on 4,567 tests saved almost $70,000,” Zutter reported. Among the tests that are commonly canceled are those for platelet factor four,  Sezary preparation, urine hemosiderin, and red blood cell folate.

Vanderbilt Integrated Diagnostics three

Many lab directors have found it challenging—and, at times, frustrating—to change physicians’ lab test ordering patterns, particularly for vitamin D tests. VUMC’s lab overcame those challenges after it introduced the formulary program late in 2014 and the number of such tests ordered each month dropped from almost 600 to zero by January 2016.

“We simply took the more expensive vitamin D tests off of the lab-ordering menu,” Zutter explained. “We told the ordering physicians that we were planning to do so, and we thought when we did it that the endocrinologists and some of private primary care providers would complain. But no one ever complained.

$30,000 Saved in One Year

“In fact, no one seemed  to  notice  that we eliminated all that testing,  and  that we saved about $30,000 over a year just for the vitamin D work,” Zutter said. “Also, we replaced 23 reference tests with in-house alternatives.

“We continually review new tests  that come online,” she  said.  “This  part is important because new tests require providers to make a decision on each  one. Therefore, no new clinical laboratory tests are offered at VUMC without going through this formulary-review process.

“I should add that one key to the success of the lab test formulary is that it’s not a lab-based process,” she commented. “It’s an institution-based process.”

When implementing the formulary, Zutter said, two factors were critical to success. First, the committee’s nine voting members were clinical leaders from throughout the medical center and, second, they used evidence-based data to decide which tests to restrict and which tests to add. “Committee members look hard at clinical utility,” Zutter explained.

“Today, all genetic testing goes through the formulary-review process,” she added. “That may be one reason a large proportion of our  genetic  testing  is appropriate for our patients. Genetic counselors review all tests. And, for pediatric patients, we have a pediatric geneticist and a pediatric pathologist review those orders before we send them out.”

Genetic Test Order Reviews

In 2016, of the 1,667 genetic test orders that were reviewed, 77% were sent as ordered, 18% were revised, and 5% were cancelled, Zutter reported.

“The whole laboratory test formulary process allows us to better define which tests should be done based on strong, published evidence and which lab vendors provide quality results,” she said.

In conclusion, Zutter said that in all  of these programs, VUMC’s clinical laboratory uses the data it collects to improve care and to improve processes iteratively over time for all patients. “By pulling all of the data together, we know that the impact of our lab’s efforts has been huge on cost savings to our parent medical center,” she commented. “Next, we want to identify the effect we’ve had on clinical care.”

Looking Forward for Labs

Now is a time when many clinical lab- oratory administrators and pathologists are looking for ways to offset deep cuts  in lab test prices. For this reason, the multi-year experience of the clinical lab at Vanderbilt University Medical Center is  a useful example of how a lab can evolve and thrive during healthcare’s transition from disjointed care to fully-integrated clinical services.

As described in this intelligence briefing, although the VUMC lab developed three separate initatives that use lab testing to deliver more value, all three programs integrate seamlessly with VUMC’s operational and clinical objectives.

Predictive Genotyping Improves Patient Care

In 2020, the Clinical Laboratory team at the Vanderbilt University Medical Center started a pharmacogenomics (PGx) testing program to evaluate patients for five drug-gene interactions (DGIs). In that program, called Pharmacogenomic Resource for Enhanced Decisions in Care and Treatment (or PREDICT), the VUMC lab used PGx testing on patients taking certain drugs.

The drugs and the corresponding genes being assessed with pharmacogenetic testing were:

  • clopidogrel (for CYP2C19),
  • simvastatin (for SLCO1B1,
  • warfarin (for VKORC1 and CYP2C9),
  • thiopurines (for TPMT), and,
  • tacrolimus (for CYP3A5).

After the program began, the lab staff followed more than 52,000 patients who were covered under a medical home model of care. Many of those patients received a number of the drugs for which the lab was testing using pharmacogenetic panels, said Mary M. Zutter, MD, VUMC’s Vice President for Integrative Diagnostics.

The lab used prospective genotyping testing to identify patients in the group who were at high risk by testing for all of the drugs on its panel. In this way, the lab tested for five DGIs simultaneously.

Almost all (91%) of the patients had at least one DGI risk, meaning the physicians caring for those patients would change the patients’ treatment once they had the test results. By continuing to follow those patients, VUMC found that such testing improved patient outcomes.

Given that result, the VUMC lab has expanded the PREDICT program over the past two years by adding other drug-gene interactions, Zutter said. At first, the lab expanded testing to 10 DGIs, and soon that number will rise to 16 DGIs, she added.

Contact Mary Zutter at 615-322-5000 or

Vanderbilt Integrative Diagnostics one



Leave a Reply