It’s become a well-worn truism, but is no less true for its repetition: the future of U.S. healthcare will be a data-driven one. What’s more, the unfolding of the global COVID-19 pandemic in the U.S. has heightened the awareness of and need for powerful data analytics in patient care organizations, including the use of artificial intelligence (AI) and machine learning tools, and predictive analytics, to help patient care organization leaders move forward effectively.
The leaders of all the progressive hospitals, medical groups, and health systems already know this—and are acting on that awareness. Take, for example, the leaders of CareMount Medical, a Mount Kisco-based multispecialty group that serves the Hudson Valley in New York state, and that recently expanded into New York City. Not only are the physicians (500 physicians, 650 providers overall) at CareMount involved in their third year of participating in Medicare’s Next Generation ACO (accountable care organization) Program; they also have Medicare Advantage contracts with four different commercial health plans.
As to how long he and his colleagues at CareMount have been actively leveraging analytics to support their value-based contracting, Conroy says, “About five years ago, we decided we wanted to become more of a value-based player. We were an MSSP [a Medicare Shared Savings Program ACO] back in the 2012 to 2013 timeframe,” and then shortly after that, “we went into CPC+”—the Medicare program’s Comprehensive Primary Care Plus payment model. “That was in 2016. We used that program as a building block, and invested in our care coordination and analytics efforts.” Importantly, he says, the medical group’s board and leadership made the conscious decision to invest strongly in those programs; and then, in 2017, he reports, “We began to lay the framework to participate in NextGen.” And, to be successful in NextGen, the most rigorous CMS program, he says, “You need analytics to risk-stratify your patient base, analyze your patient base, and develop care management plans and care coordination tools.”
Very importantly, Conroy says, in order to be successful at value-based care contracting “You need to make sure that your patients, your physicians, and your analytics infrastructure are all aligned. Who do we need to see and who do we need to treat? Who has multiple comorbid conditions, so that we need to proactively develop care plans for them?” What’s more, when physicians are given the data and information that a good analytics program can provide them, it motivates them. “With analytics, it’s easier for them to see who’s seen their patients both inside and outside CareMount. And we’re taking lab and radiology data under the same umbrella. All of that’s making the difference: we can now brag about our annual wellness rates and we can brag about our quality scores. It’s improving patient care, outcomes, and the health of the community.”
Survey findings: a journey in progress
One thing that everyone in U.S. healthcare can agree on: this journey into fully leveraging data analytics to facilitate success under value-based contracts remains a journey in progress, with the leaders of patient care organizations nationwide gradually moving through its stages.
That reality is affirmed by our own data research at Healthcare Innovation. Our State of the Industry Survey, some of whose results were published in the January/February issues of this publication, inclusive of responses from some 200 senior healthcare executives, found a very mixed picture with regard to the state of analytics usage.
We found a range of responses with regard to our question around analytics: 24 percent agreed with the statement, “We are advanced in our level of analytics usage, fully leveraging broad and deep data analytics to power our population health management and care management work at every level”; 49 percent agreed that “We are early on in our journey of leveraging data analytics to power our population health management and care management work”; and 27 percent agreed that “We have not used data analytics yet in any significant way to power population health management and care management work.”
Those results are particularly interesting in light of the survey’s results around participation in value-based contracting. Among survey respondents, as of January of this year, 32 percent of respondents were participants in the MSSP program; 12 percent, participants in the NextGen program under Medicare; 21 percent, participating in a Medicaid ACO; and 35 percent, participating in a private insurer ACO. Among those participating in one or more of those types of value-based contracts, 12 percent were in downside risk under MSSP or NextGen; 11 percent in a Medicaid program; and 64 percent in programs with private health plans.
Health systems learn on their collective feet
Even as pioneering organizations like CareMount Medical move boldly forward, leveraging data analytics to support their advanced work in population health management and care management under value-based contracts, the current situation involving the COVID-19 pandemic is helping to ignite tremendous data analytics-based work to help the leaders of patient care organizations manage the clinical, operational, and financial zigs and zags of hospital and health system management under pandemic conditions.
Mike Ross, vice president of population health & enterprise analytics at OhioHealth, a 12-hospital integrated health system based in Columbus Ohio, spoke of those challenges and opportunities during a June 10 webinar presented by Healthcare Innovation and sponsored by Informatica. At OhioHealth, Ross told webinar attendees, “We had selected our data governance tools and were in the process of installing those tools; our target, quite frankly, was June, the end of our fiscal year, to go live” with the robust new data analytics program. “When COVID-19 hit,” Ross said, “People needed to know about beds, ventilators, tests, drugs—and we realized very quickly that we needed to get our hands around this, and dramatically accelerated the release of our data governance capability. We needed to get our hands wrapped around that very quickly, in terms of capacity and capability. Somebody once said to me, ‘never let a good crisis go to waste.’”
“We had also been on a journey around predictive analytics and AI,” Ross told his audience. “And while that was on our roadmap, we quickly realized we needed to do that right away. So we took our data scientists from across the organization, formed them into a single group, named a director, and had him start them running. What is our predictive model for COVID? How are we going to predict elements like beds, supply chain, and workforce?” A key element, Ross said, was to help to train end-users so that they could become “a community of developers.”
Looking forward into the next two years, Ross said on June 10, “We still actually have two crises going on; we are still seeing a high census of COVID-positive patients. And we’re trying to develop a delivery model of the future, including reassuring patients of safety.” Importantly in all this, he added, “We’ve been able to distinguish between immediate needs and longer-term needs; because if we constantly act on ad hoc requests, we won’t be able to fully serve our organization.”
Looking past the pandemic
But—whenever it is—once the pandemic-related crisis has passed, how will the leveraging of data analytics evolve forward into the future? “For one thing, we’ll look at what we learned during this experience, and how we can turn what we’ve learned and implemented, to help make healthcare better in the future? Among other things, telehealth and remote care delivery will continue to advance,” for the convenience of patients, says Scott Weingarten, M.D., who is both chief clinical and innovation officer at the Charlotte-based Premier Inc., and also consultant to the CEO of the Cedars-Sinai Health System in Los Angeles, and a professor of medicine at Cedars-Sinai.
Importantly, Weingarten emphasizes that “The analytics and the technologies have advanced, thankfully, over the past ten years; but the way that you impact a population of patients is one patient at a time.” Asked to respond to the case of a theoretical patient, Mrs. Smith, who has congestive heart failure (CHF), diabetes, and hypertension, and how the leveraging of data analytics might help her in her situation in the future, he says, “In terms of Mrs. Smith, we might look at her blood pressure and her hemoglobin a1c, and her medications, but we’ll also be able to review a lot of information around her social determinants of health. Can she afford her medications? Can she afford healthy foods? We’re going to look for information about medication adherence, and we’ll look at what’s being documented in the notes. And if she has CHF, we need to read and interpret her last ECHO report, including her ejection fraction. That information is often embedded in the text. So we’ll need NLP”—natural language processing.
All those interviewed for this article agree that, for the moment, managing aspects of hospital, medical group, and health system operations in the context of challenges coming out of the COVID-19 pandemic will continue for the time being. And that clearly will be important for patient care organization leaders to manage.
COVID-19 and the supply chain—policy and operational impacts
Meanwhile, one of the strange ironies of the COVID-19 pandemic’s impact has been a massive turn of industry leader attention towards the enormous gaps and challenges in the nationwide healthcare supply chain that the pandemic has exposed. As Blair Childs, senior vice president of public affairs at Premier Inc., puts it, “What has occurred in the past three to four months is that the government has become acutely aware of how antiquated and really backward our health information technology infrastructure is overall.”
Together, all the parties involved in that collaborative have been working on “everything from drug shortages to a lot of attention to PPE”—personal protective equipment, Childs says, noting that they have been interacting directly with FEMA (the Federal Emergency Management Association) and the White House, “non-stop” since the COVID-19 pandemic hit the U.S. A key learning, he says, was for everyone to see clearly how very may steps are involved in acquiring PPE, among other important examples. “Most of the regulatory changes that have been made” around the nationwide supply chain in the past few months have come out of the work of the collaborative, he adds, and “many have come from Premier specifically, because we would surface something and identify a problem, which would lead to an adjustment, whether it was an emergency use authorization, or sterilization, or flying over swabs from Italy because there weren’t sufficient swabs.”
All of this work, Childs says, has led to intensified awareness of the nationwide issues around the healthcare supply chain across the U.S., with federal leaders realizing “how backward the U.S. healthcare IT system is nationwide.” As a result, he notes, in May, the Health, Education, Labor and Pensions (HELP) Committee published a report around all these issues, and the members of the committee, “are going to try to move legislation on this regarding supply chain and the strategic national stockpile. How do we prepare for the next pandemic, or next terrorist attack or other disaster?” he asks.
Included in that work will be efforts to reengineer how the Centers for Disease Control and Prevention (CDC) collects epidemic surveillance data, which remains mostly paper-based. Moving forward, he reports, federal regulators and legislators are looking at data collection processes around disease testing and symptom identification.
Individual efforts in California
Christian (Chris) Pass, CFO at the three-hospital John Muir Health in Walnut Creek, Calif., a suburb of San Francisco, has been deeply involved in COVID-related efforts inside his own health system. “The Bay Area, with its five or six counties, really got ahead of things, and enforced shelter-in-place, and it really paid off” for the healthcare organizations in that region, early in the COVID-19 pandemic, Pass notes.
Like leaders at many patient care organizations nationwide, those at John Muir Health quickly created a COVID-19 command center, and, from the start, Pass reports, “We would have daily phone calls, and started to call out data, which led me to call the analytics team and ask them to track phone calls, visits, requests, to see if we could identify trends around potential surging. And they asked me if we’d like to look at data models. So they brought forward some models in terms of PPE, space needs, and bed needs.”
The process at John Muir Health has evolved forward in a totally iterative way, Pass notes, with the organization’s leaders making the decision in real time to delay the implementation of an EHR upgrade once the pandemic had hit the region, in order to avert any potential for implementation-produced downtimes. What’s more, he notes, “We set up an enterprise system for prioritization across the enterprise, prioritizing the sickest of the sickest, and dealing with the unknowns around COVID. And then we looked at volumes of patients.”
With regard to analytics, Pass says, what’s been very heartening is that “We’ve become much more analytically driven as an organization. The predictive analytics we did around what was going to happen and how to prepare for surges—we made many decisions based on data. That was a great example of using analytics to run the business better.”
Analyzing financials going forward
Leveraging analytics to help strategize forward will be exceptionally important for the leaders of hospitals, medical groups, and health systems, say Tim Zoph and John Klare of the Naperville, Ill.-based Impact Advisors consulting firm. And in that regard, Klare says, “We’re trying to help hospitals understand how the process will unfold in terms of both patient care and financials. There are models that can help you predict the waves and peaks.” They and their colleagues at Impact Advisors are actively working right now with their hospital and health system client organizations, using what they’re calling a recovery model for health systems.
“If you start to consider when you can predict patient volumes coming back, you can look at it in terms of the OR and ED,” Klare says. “But frankly, the more interesting perspective is based on the disease or the DRG, and in that, you’re seeing different behaviors across the country. Most organizations are rebounding” from the financial losses created by the shutdown of elective procedures this spring, “but many more aren’t bouncing back as fast as they’d expected. And what’s it going to look like if there’s a second wave, and its timing? That’s the perspective we’re using.”
Importantly, Zoph says, one of the lessons that’s being learned now during the COVID-19 pandemic is that “The old crisis-management playbooks really are not adequate to respond to a crisis of this magnitude. Crisis and surge management in the future will involve predictive modeling, looking at early warning signs in terms of other disease spread or whatever the crisis might be; the goal will be to put into place tools that will allow you to respond proactively.”
The big-picture challenge? “Healthcare has been really good at retrospective analytics,” Zoph notes. “We’ve got data ‘til Sunday around how we get paid and are regulated—even retrospective quality data. What we haven’t figured out is how to use data to look forward. Data is still very siloed, but we need to combine data for forward-looking decision-making. We still have a ways to go as an industry to really gather the data that allows us to be more far-sighted in how we use the data.”