Sponsored by GE Healthcare
Developing an enterprise analytics model for value-based care
Hospitals and health systems are moving beyond the collection and management of electronic data and toward analyzing the data to improve patient outcomes and create value. This quest to harness data for value-based care is an exciting phase in health care delivery.
However, implementation of enterprise analytics tools sometimes can prove frustrating due to a lack of clear objectives or a mismatch between expectations and results. A systematic framework for implementing data analytics is helpful to achieve realistic goals. Experimenting with big data sets, algorithms and even artificial intelligence can provide insights into care quality and identify inefficiencies.
The value proposition of an enterprise analytics model
Eight hospital executives from around the country gathered at the American Hospital Association’s Annual Membership Meeting in Washington, D.C., recently for an hour-long dialogue on the topic of creating value through data analytics.
The executives shared their experiences in moving toward a data analytics framework. The value proposition of an analytics model came up in numerous areas of primary, acute and sub-acute care. These included mental health, imaging, high-risk patients such as the homeless, chronic care management, food security, hospice, care transitions, obesity, workflows and caregiver support.
Much of the value proposed from data analytics was aspirational, and not yet implemented. Still, the executives showed an enthusiasm for the potential of data analytics, and their first steps toward implementation provide insights into where the field is headed and the potential to create value for patients.
Analytics for interventions
Boston Medical Center, a safety-net hospital with a Medicaid insurance plan serving about 420,000 people, is using data analytics in its Medicaid accountable care organization (ACO). The ACO is in partnership with the state of Massachusetts, which has developed a risk-adjustment methodology, which, in addition to clinical factors, recognizes social determinants of health. These are homelessness, housing and security, severe mental illness and substance abuse. With approximately 3 percent of patients driving about 40 percent of spend, the medical center is searching for a data analytics platform that finds patients at risk of becoming those few high-cost patients so providers can intervene earlier.
"Our pilot data indicate that typical high-spend Medicaid patients are 57 years old, so we believe intervening when patients are in their 30s and 40s could help bend the cost curve," said Kate Walsh, president of Boston Medical Center. "If we could find a way to appropriately outreach to and better manage the health of those patients before they become the 3 percent, I think we'd do a much better job," she said.
As one step, Boston Medical Center has added a "thrive screen," asking patients eight questions about housing, food security, paying for medicine, transportation, utilities, child care, training and education, Walsh said. "Now that we've asked, we have to do something with the information."
Grady Health System, a public-private safety-net system in Atlanta, uses a predictive analytics tool called mobile integrated health to conduct outreach to patients in the community. Seventy teams of medical and behavioral health professionals are on the road providing care to patients, and each day they get a list of patients to visit, provided through an analytics tool. Additionally, the health system has a waiver allowing the teams to respond to 911 calls when first responders determine the patient doesn't need emergency medical care.
"We run one of the largest hospital-based EMS agencies in the country," said John Haupert, president and CEO of Grady Health System. "We are the largest provider of mental health services in Georgia."
Like Boston Medical Center, Grady Health System is combining analytics with social determinants of health research, said Haupert.
A predictive analytics tool used by Grady Health System providers tracks 20 social determinants of health on patients to identify which patients are more likely to be readmitted to the hospital within 30 days or to show up in the emergency department (ED). By knowing which patients to target for more intensive intervention, "we can better care manage those patients and a bit more effectively keep them healthy in our ambulatory environment and prevent readmissions and over-utilization of the emergency department and high-cost care," said Haupert.
Reducing infections with analytics
Publicly reported data can serve as a jumping off point to further data analytics, according to Rick Stevens, president of Christian Hospital in St. Louis.
Christian Hospital deploys paramedics to the homes of patients who are high utilizers of hospital services, as identified through analytics of ED and admissions data, Stevens said.
Additionally, Christian Hospital had a high catheter-associated urinary tract infection (CAUTI) rate. Not only did the hospital implement protocols to reduce infections, but "we also measured steps along the way," Stevens explained. These included the reasons behind catheter usage and whether each catheter was necessary in patient care. As a result, the hospital got its CAUTI rate down to zero, he added.
"When you say deliver on the value proposition, there's a lot there from that standpoint," Stevens said. "We use [analytics] for ... how we deliver care and set up programs in the community as well."
Analytics and successful community partnerships
A high rate of obesity among youth in Bristol, Conn., spurred a novel community partnership. Bristol Hospital, a 150-bed facility, partnered with community groups to address the issue. Identifying at-risk children through the Women, Infants and Children (WIC) Food and Nutrition Service, the children and their families participated in cooking, nutrition, gardening and movement classes. "The improvement in [body mass index] for these kids was outstanding," said Kurt Barwis, president and CEO of Bristol Hospital.
Similarly, Lee Health, a $1.8 billion health system in Fort Meyers, Fla., focused on keeping people healthy at home after they leave the hospital. Lee Health partnered with community organizations to deliver healthy food to people's homes during care transitions.
"We've had great, great outcomes as a result of this initiative," said Scott Kashman, chief acute care officer at Lee Health, citing an almost 40 percent drop in readmissions.
Analytics for improved quality of life
The executives expressed great hope that analytics can help solve some of their most intractable problems.
"Our big focus is how do we get better, robust analytics primarily for the predictive analysis that can help us identify not just multiple readmissions, but also those patients who have chronic disease — just to make sure they are getting follow up with regard to medications and treatments," said Mark Merrill, president and CEO of Valley Health System, a six-hospital system based in Winchester, Va.
Valley Health System established a chronic disease resource center for patients with multiple co-morbidities for coordinated care management, Merrill said.
Stevens of Christian Hospital agreed that analytics could help identify patients for interventions and improve quality of life. "We have to have the analytics to help guide us toward our high-risk patients and those high utilizers of service," Stevens said. "Not because they are high utilizers but because there is a better way."
Leaders whose systems have been managing data for many years said that there is room for improvement in terms of analytics use.
"Mental health is still somewhat wide open in what's available," said Doug Struyk, president and CEO of Christian Health Care Center, a post-acute mental health provider in Wyckoff, N.J.
Synthesizing data between the acute and post-acute environment is also ripe for innovation, Struyk added. "We've been trying to figure out how to manage the data we collect with data from acute care partners, on different EHRs, measuring data in very different ways themselves," he said.
Data analytics can reveal missed opportunities in care. Chadron Community Hospital in Chadron, Neb., used its data analytics platform to learn that only 2 percent of patients had completed an advance directive. This offered an opportunity to work with patients earlier on to find out their end-of-life wishes and to start the conversation earlier about hospice.
"We've got a great hospice program, a great home health program, but we seem to not be getting these folks in during very vulnerable times," said Anna Turman, CEO of Chadron Community Hospital.
Armed with this data, the hospital started conducting education outreach to patients and families about advanced care directives. This included evening showings of the documentary "Being Mortal," based on the Atul Gawande bestseller of the same title, which Turman personally bought a hundred copies of to hand out to patients.
At the movie showings, the patients fill out an advance directive, with staff members on hand to help them if they have any questions, Turman said.
"We bring people in, we give the books out," Turman said. "They fill out the advance directive and get their five wishes done ... they have to bring a friend, and then that friend brings a friend .... So, it starts getting bigger and bigger."
Identifying patients who could benefit from hospice or palliative care sooner, based on frequent admissions, diagnosis codes and emergency room visits, is a shared goal of Valley Health System, said Merrill. "This is one area where we are trying to figure out how we can develop the (data) tools," he said.
Sustaining value with analytics
Rob Nelson, global marketing director of artificial intelligence and analytics at GE Healthcare, said analytics models must be sustainable to create value in health organizations.
"One of the things I'm particularly interested in is, as you look at analytics, how do you build that model to make it sustainable to create value for your system," Nelson said.
Artificial intelligence (AI) applications can help to improve efficiencies in clinician workflow, helping to sustain value, Nelson added.
"AI can take care of some routine tasks," Nelson said. "No human intervention is needed any longer once you do it well. The one thing that we have a lot of are devices and there's a lot of data on those devices. So, we're starting to connect those devices and utilize AI to drive better processes, for improved asset utilization, throughput and capacity."
The ability to integrate data to create a more complete picture of patient health is of great interest. Additionally, having data analysis available in real-time (as opposed to using data from a few months prior) is highly sought because it is more accurate and actionable.
"Absent the integration part you are going to have the same problem with some of these machines," said Walsh of Boston Medical Center. "Everyone is talking about it in terms of predictive analytics. How about in-the-moment analytics?"
The hope is that sophisticated, real-time data analytics could help care teams get back to their core purpose of treating patients, the executives agreed. That would prove the value of a data analytics framework.
"When you talk about population health, you think about how much money you're all spending on [electronic health records systems], how much money we're spending on data and overlaying this," said Barwis of Bristol Hospital. "At the end of the day, it's about the community and what you can do to help the community."