This article originally appeared at http://www.healthcare-informatics.com/article/moving-out-its-emergent-phase-healthcare-data-analytics-becomes-more-real on March 29, 2016.
As healthcare organizations move further into data analytics work, what are they learning in the process?
The healthcare analytics phenomenon has been a leading edge topic for several years, with exciting discussions among health IT leaders around the capabilities of Big Data and predictive analytics tools at virtually every industry conference and event. What’s more, the practice of data analytics has now grown out of its infancy stages as more providers move forward, and move more deeply into, the challenging work of setting up the data warehouses and business intelligence capabilities needed to support analytics.
As further proof of the growth of the healthcare analytics market, last month Watson Health, IBM’s artificial intelligence computer system, acquired the Ann Arbor, Mich.-based Truven Health Analytics, a provider of cloud-based healthcare data, analytics and insights, with its more than 8,500 clients, for $2.6 billion. Upon completion of the acquisition, IBM’s health cloud will house one of the world’s largest repositories of health-related data.
It’s clear that data analytics is here to stay; yet the discussions around it are changing, health IT leaders say. Keith Figlioli, senior vice president of healthcare informatics at the Charlotte-based Premier, Inc., recently spoke with Healthcare Informatics on the state of healthcare analytics and he referenced Gartner’s Hype Cycle schematic around the maturity and adoption of new technologies as a good parallel. Gartner is an information technology research and advisory company, and according to Gartner’s Hype Cycle schematic, there are five key phases of a technology’s life cycle beginning with technology trigger and then moving into the peak of inflated expectations. That’s followed by a downswing, called the “trough of disillusionment,” and then eventually a gradual upswing, referred to as the slope of enlightenment and finally the plateau of productivity.
“We were in this hype cycle for so long, and I think we’re starting to get into the trough of disillusionment,” Figlioli says. “So I think many people are testing a lot of concepts, seeing if this stuff would work, and now they’re starting to see, one, how hard this work is, and two, how immature the space is.”
He continues, “We were in a four to five-year hype cycle in this area, and now we’re in this trough, and right after that, it picks back up and you get productivity gains. I think in the next two years or so, we’ll come into that upswing again. But we’re just in the early stages of getting into this trough, where people are really digesting what they spent and the tools that they have and asking the question, Is it really going to get them there?”
While the description may be stark, Figlioli uses it to illustrate how the discussions around data analytics are changing as the industry moves into second and third itineration buying cycles.
“I think now the discussions are, Wow, this is hard; we’re in it for the long haul here,” Figlioli continues. Even with our business, people who didn’t talk to us on the first round are coming back and want to know more about deep enterprise data management capabilities. And what I see is that we’re in second and third round buying cycles, which tells me that people are getting smarter, and they are realizing the level of complexity here. We see more attention to data management, data acquisition, data transformation and enterprise data management,” he says.
Four years ago, Premier launched its PremierConnect Enterprise business intelligence and enterprise-wide analytics platform, which includes a cloud-based data warehouse, which is vendor- and payer-agnostic.
“The storyline here, to me, is more about data management. When it comes to data analytics tools, it might be a great dashboard or a great predictive algorithm or actuarial analysis on risk stratification, but if the data coming into those systems is not good, those tools are almost useless, and most health systems don’t have good data,” he says.
Figlioli also is involved with his organization’s Data Alliance Collaborative, an organization that includes representatives of 12 of the largest health systems focused on co-developing data analytics methods and sharing best practices. So what have these pioneering health systems been learning in the process?
“It’s a marathon, not a sprint,” Figlioli says, “I think the two biggest lessons are, one, at what stage are health systems from a cultural readiness standpoint to do this type of work? So, are they socially and culturally ready to use data to run their business?”
And many health IT leaders say that cultural readiness must come from the top down. “Clinical leadership has to be 100 percent aligned that this is the way we’re going to run the enterprise and deliver care,” Figlioli says.
The second learning centers on effectively prioritizing the analytics work across the enterprise.
“When you do these types of projects what becomes very clear is that you have multiple stakeholders with multiple interests across these larger enterprises so, how do you think about the priority order to do the work that gets you the biggest bang for the buck?” Figlioli says.
Leveraging Analytics Locally, and Enterprise-Wide
Al Villarin, M.D., CMIO at Staten Island University Hospital, agrees that a cultural shift is vital for analytics work. “We have an entire country of clinical people who learned to practice medicine not using analytics, so it’s going to take a generation of training with the use of analytics to change clinical practice,” he says.
Villarin, who also is a practicing emergency physician, serves as the director of Staten Island University Hospital’s Division of Quality Analytics, which is part of Northwell Health (formerly North Shore-Long Island Jewish Health System), a 21-hospital integrated health system serving eight million people in the metropolitan New York area. When the hospital rolled out its electronic medical records (EMR) system, Villarin led a hospital-wide workflow analytics group (WAG) to address how well processes were working, identify where variations occurred, measure performance and improve patient flow.
“You have to have your entire team—clinical team, administration, finance—on the same understanding of how analytics can affect them and how they can affect the success of analytics. With WAG, we had that buy-in and that’s why we were successful in approaching the transparency of the data,” he says.
Villarin continues, “Data is great, but for the most part, it’s locked behind computers, and you have to bring it out front and create an information portal that allows information to affect and change practice. And that is the hardest part to do, changing clinical practice with people who have worked in one way all their lives.”
More broadly, Northwell Health is taking steps to leverage data analytics enterprise-wide as part of its overall strategy to provide managed patient care. The health system’s shift from traditional fee-for-service to one based on shared risk and pay for performance has created a need to collect, collate and harness data across the Northwell health system. The health system has set up a data warehouse where the insurance claims and electronic health record (EHR) data collected from the hospital sites is aggregated and analyzed.
The goal, says Villarin, is to create a “network-wide clinical decision support element that gives information right back to the clinicians in real-time.”
“We want real-time surveillance, and real-time data analytics, so we can impact, immediately if necessary, patient care to prevent further demise or support an efficiency in outcomes. So, if the patient is better, the physician can get an alert to send the patient home. So it can help reduce length of stay, improve patient satisfaction and reduce healthcare costs. The core of all this is learning about the patient and that’s where data analytics comes in,” he says.
And, Villarin acknowledges that issues around governance and data management and mapping the data to end users so it’s delivered to those end users in a meaningful way, are significant hurdles in the process.
“I’m an ER physician, so I want to know when my patient is having problems,” Villarin continues. I’m not just sitting there waiting for a problem to happen as I’m looking at other patients, so I don’t want to worry about it until the system, the analytics, tells me that the patient is having a problem. So the governance around that, the alerting, the timing, the management, that’s still going through the creation process right now,” he says.
Many health IT leaders say that as healthcare organizations begin to recognize the scope of the work involved with setting up the business intelligence capabilities and data governance and quality initiatives needed for data analytics, it’s imperative that the leadership have a strategic vision for the end goal.
Leveraging data analytics to improve outcomes is the holy grail, Villarin says, “That front end piece, taking the information and delivering it at the point of care, that is where we’re headed.”
“Opening up the pipes of data moving to the data warehouse, training clinical people in the role of analytics and then finding better tools to look at the data and then transfer the information and the knowledge that is found in that data back to the end users—those things are big steps in the next five years that we embracing for our network,” Villarin says.
Driving Analytics to the Bedside
Many leading healthcare organizations have moved beyond setup to actually applying analytics to multiple use cases. The 42-hospital Charlotte-based Carolinas HealthCare System has invested significant resources to leverage analytics capabilities to support evidence-based population health management, individualized patient care and predictive data modeling. The health system’s Dickson Advanced Analytics (DA2) department integrates and analyzes clinical data to manage the health of individual patients and of communities and to develop analytics models that can be used to predict population health trends.
Through the use of its readmission risk analytics, the health system’s hospitals are now able to predict, with nearly 80 percent accuracy, a patient’s risk for being readmitted within 30 days after being discharged by pulling electronic medical records data for 40 different patient variables.
“Carolinas HealthCare System has good examples of where they’ve used good data management capabilities, good data governance and they have set up a good data governance committee across the enterprise. They have forensically gone through and made sure that their data is right and they’re using real analytics in the workflow and in clinical settings to be able to drive behavioral change,” Figlioli says.