Laboratory-Based Patient Care Project Moving From Theory to Practice
December 11, 2019
Laboratory-Based Patient Care Project Moving From Theory to Practice Dec 06, 2019 | Adam Bonislawski NEW YORK – With government and private payors putting the squeeze on pricing, and patients seeing more exposure to testing costs, there is not unreasonable concern within the lab industry about how long traditional fee-for-service business models will remain viable. To succeed in coming years, one common line of argument goes, labs will have to become more than just providers of test results. They will need to leverage the vast quantities of patient data they are sitting on to impact healthcare at a system and population level. Broadly speaking, this notion has come to be known as Clinical Laboratory 2.0, a term initially coined by members of the Project Santa Fe, a group formed in 2016 by five health system-based labs with the goal of demonstrating how labs can carve out new roles within the healthcare system and use lab data to proactively impact patient outcomes. While this vision of the lab's role is still in its infancy, the industry is broadly supportive of this shift in direction, said David Nichols, president and founder of lab services consulting firm Nichols Management Group. "No one in the industry would argue against it," he said. The extent to which that general support translates into concrete action remains to be seen, but examples of how Lab 2.0 ideas might be put into practice can be seen in a number of labs and lab information firms around the country. Albuquerque, New Mexico-based TriCore Reference Laboratories is one of the companies at the forefront of the move toward the Lab 2.0 idea. When Rick VanNess, director of product development for The Rhodes Group, TriCore’s technology subsidiary, started at the company five years ago he began contacting New Mexico insurers asking how the lab might help them better manage their members. "Their response was usually, well, you're the lab, you guys can't really do anything for us except, please, lower your cost per test," VanNess said. "It took us a little while to change that perception of us," he said. "To understand that, hey, I'm not here to talk about cost per test, I'm here to talk about creating products and services that help you in your contractual obligations, in your mission as a payor." VanNess cited the example of caring for patients with hepatitis C, which several payors he spoke to highlighted as a challenge. It was particularly a challenge for payors covering New Mexico's Medicaid population, he said, noting that Medicaid provides bonuses for these payors if they manage to deliver treatment to a certain number of hepatitis C patients.
"If they didn't treat that certain number, the state would claw back a percent of that bonus,"
he said. "We realized the payors have to treat people with hep C. What we want to do is tell them who is positive for hep C and also tell them who they want to treat first. Because if someone has cirrhosis happening and are in a high-risk state where they have diabetes or HIV or things that would indicate their hep C is more likely to result in a poor outcome, we want to let them know, you want to treat these patients now because that is a $250,000 liver
transplant" if they don't get treatment. Using lab data, The Rhodes Group was able to identify patients with hep C and then stratify
them by whether they showed care gaps or additional risk factors.
"We would go to the payors and say, OK, here are 30 patients with hep C and here is where they lie in that stratification grid," VanNess said. "And 90 percent of the time they would come back to us and say, we didn't know about any of these. We want to start using this information. How much do you want for it?" He said the company determined pricing by figuring out how much of their Medicaid bonus insurers stood to lose if they didn't hit their hepatitis C treatment numbers and then pricing at around a tenth to a twentieth of that figure. "We believed they would want a 10- or 20-to-1 return, so we priced that product right in there," he said. "That was our first kickoff toward getting paid differently." Since then The Rhodes Group has launched similar products, which is calls targeted intervention modules, or TIMS, for prenatal care, diabetes, and chronic kidney disease. VanNess said the company has signed TIMS deals with two of the state's three Medicaid
payors and is in negotiations with the third and has now begun working on deals with commercial and Medicare payors.
The Rhodes Group has also begun piloting efforts with health systems, many of which are financially incentivized by payors to hit various treatment targets.
"It's the same scenario we went through with payors where we sent them information and see if it made sense, if they liked it, and then their feedback would allow us to create products that they would want," VanNess said.
Health information firm Prognos Health is similarly leveraging lab data to help payors identify care gaps for patients with conditions like diabetes.
"We're able to find gaps in care for patients and alert the payor so that they can intervene to make sure the patient or the member is getting the care they need," said Mark Reis, vice president of diagnostics and data markets at Prognos.
The New York City-based firm was founded in 2010 with an initial emphasis on using lab data to help drug makers, particularly those developing targeted cancer therapies, to identify and reach potential patients.
Prognos started out nine years ago knowing that targeted therapies, especially in oncology, based on lab testing were beginning to hit the market. "That is where we started with our first solution and that paradigm has only grown," Reis said, adding currently, between 80 and 90 percent of oncology therapies are targeted. The company's move into using lab data to support payors began about three years ago, he said, noting that this decision was driven by the fact that with the recent launch of the Affordable Care Act's insurance exchanges, payors found themselves without longitudinal and timely data on many of their members. "Payors always use claims" data to follow their members, Reis said. "Our theory was that lab data is more timely, more influential, more accurate, more targeted. And for the types of
value-based structures payors are dealing with and with risk optimization and identifying patients that need care, lab data is going to be much more accessible and efficient for helping payors do this."
VanNess likewise argued for the superiority of lab data compared to the claims data conventionally used by payors and health systems for their population health analyses. ICD-10 and CPT codes can be very inaccurate because providers can see their patients for as little as 12 minutes. Health systems realize that if they focus on an ICD-10 code from that
patient visit to run its population health and care coordination, it's going to be inaccurate "a certain percent of the time. So, we're going to go to the lab, because the lab is going to tell us that patient does have rheumatoid arthritis, that patient does have anemia. Our sweet spot is translating that valuable QA'd lab result." Prognos has also turned to artificial intelligence to help it leverage the lab data it has acquired, which comprises more than 10 billion records on 160 million patients. Reis said Prognos typically structures its deals with labs as revenue sharing arrangements where it uses its capabilities to help a lab utilize its data in a way Prognos can then sell to a
pharma or payor customer. "Any data solution that we create and sell to a pharma or a payor, [the lab] shares in that," he said. The company works with a range of outfits ranging from hospital labs and national reference labs to specialty genetic testing firms.
The exploration of Lab 2.0-type models is driven not just by the immediate pressures the lab industry faces but also by the understanding that US healthcare generally is moving from a fee-for-service model to a value-based model, and that the lab business needs to make space for itself in this new world. "In a value-based model, the more value you offer, the better off you are," said Jill Warrington, chief medical officer at South Burlington, Vermont-based drug testing firm Aspenti Health. "I think we're gearing up for that."
In May, Aspenti was awarded the Clinical Lab 2.0 Innovation Award by the Project Santa Fe's Clinical Laboratory 2.0 group, which gave the award in recognition of the company's work using lab data and social determinants of health to improve care of patients at risk of substance abuse.
Using lab data to identify patients using both opioids and benzodiazepines — the co-use of which is associated with an increased risk of opioid overdoses and increased use of acute medical care — the Aspenti team then looked at the social determinants of health predicted co-use and found that the provider ordering the lab testing was the most predictive factor of co-use, with increasing age and patient geography also proving predictive. The researchers published their results in an October paper in Academic Pathology. The company is currently in the process of rolling out a software tool that provides doctors with this data for their patients in real time, Warrington said, noting that it just finished beta testing the software with 20 or so clients across Vermont and that it plans to begin offering the data as a commercial product in early 2020. She said Aspenti was evaluating other areas where its lab data could prove useful, citing as
one example monitoring patients prescribed buprenorphine to help overcome opioid addictions to make sure they are staying on that medication.
At New York's Northwell Health, Tarush Kothari, medical director of informatics at Northwell Health Laboratories, led a project aimed at better identifying hospital patients at risk of acute kidney injury. Tracking patient creatinine measurements, he and his colleagues found that incidence of AKI was significantly underreported. The impetus for the project came from the observation by the chief medical officer at the system's Long Island Jewish Forest Hills Hospital in Queens who noticed that several patients a week were having their stays extended due to AKI caused by the contrast used in
imaging studies. "They were looking for a way that these cases could be detected in a more timely fashion so that preventive measures could be taken to prevent the disease from progression," Kothari said.
Looking into the literature on the subject, the researchers found that the UK's National Health Service had developed a program using lab data to improve AKI detection. Using that program as a model, they built a system using lab creatinine measurements to identify patients suffering from AKI.
Tracking every creatinine result that went through Northwell's lab information system, they ran the diagnostic algorithm at the Forest Hills hospital to determine how many patients had AKI.
"And what we found was actually astonishing," Kothari said. "It was not just two or three patients every day. It was more than 15 to 20 patients on any given day that had that diagnosis."
While AKI is a very common condition, most internists and frontline providers in the ER and medical floors — the ones who actually make the diagnosis — may not know the diagnostic criteria for the ailment, he said. "If you talk to a nephrologist, they are very well versed in the diagnostic criteria, but they are not the ones who are picking up the early phases of [AKI] in the hospital setting." After Kothari and his team presented their findings, Forest Hills Hospital decided to pilot the program. Over the course of a year, documented AKI cases at the hospital went from around 5 percent to around 12 percent, he said. Rolling the project out across all of Northwell's hospitals, they similarly boosted documented AKI cases from around 6 to 7
percent to 12 to 14 percent. "That was a big achievement," Kothari said.
He added, though, that applying the lab data was only the first step in the process. Also key was making frontline physicians aware of the problem and paying attention to it. "We partnered with the chief medical officers at every site," he said. "They were aware this project was going on, and they conducted education sessions. There was a lot of behavioral
change that was involved."
Another key consideration was in how to present doctors with the alerts, Kothari said, noting that presenting them as active alerts in patients' electronic medical records might lead doctors to ignore them.
Instead, AKI alerts were put into a daily report for each hospital unit, giving doctors within each unit a list of patients receiving a diagnosis of AKI within the last 24 hours. "A lot of coordination between the lab, the chief medical officers, the revenue cycle teams, had to go into this project to make it successful," Kothari said. He recalled showing the discrepancy between AKI cases as represented by Northwell's claims data versus its lab data. "It got the attention of the CMOs, even the chief operating officer," he said. "And they said, we are really underdiagnosing this condition, what can we do to improve this? And once we got their buy-in, [implementation] became progressively easier. There was support from the finance folks who were willing to share their claims data with us. There was support from the analytics team with other data." Kothari said that while this sort of cooperation is key to making Lab 2.0 ideas work, it isn't always present. He and his team have worked to implement their AKI system at other health systems but have on occasional run into organizational or institutional barriers that have limited progress. "As some institutions, the nephrologists have not been very welcoming of the idea," he said. "For instance, at one institution, the nephrology department told the lab, look, we have a grant for this project and we don't want labs to be leading it. There are those kinds of issues at different hospitals." He mentioned an effort at Northwell to develop a lab utilization system based on one developed at Henry Ford Health System in Detroit. "It was not for lack of trying, but we just didn't find enthusiasm for that project, and it really didn't take off," he said. Technical limitations also come into play. "The [AKI] algorithm would seem very simple to implement, but a lot of electronic medical records really can't compute thousands and millions of creatinine values, so a lot of hospitals told us that they just didn't have the bandwidth to conduct this kind of project," Kothari said. "What we have learned through these projects is that healthcare microenvironments are very different from each other, and we talk a lot about our successes and failures and try to learn from them," he said.