Clinical & business intelligence: the right stuff
This is Part II of our three-part June 2015 print cover story on healthcare analytics. Part I focuses on the first steps of launching an analytics program. Part II focuses on intermediate strategies, and Part III focuses on the advanced stages of analytics use.
Jeff Fuller has some advice for those healthcare providers in the middle stages of their analytics journey: Leadership comes from the leaders.
"It needs to be a strategic goal to become a data-driven organization," says Fuller, director in advanced analytics at Carolinas HealthCare System. "That needs to be stated from the very top level for your hospital or system."
But everyone else has to be on board too, of course – invested and ready to go where the data takes them: surgeons, nurses, staff, everyone.
Even that is not all, however. Making smart use of data means making sure those folks are using the best data.
"Many middle-of-the-road organizations, when it comes to data maturity, don't have their hands around all the data," says Fuller. "They're making decisions on the data that's readily available to them, which is not always the right data."
[Part I: A beginners guide to data analytics]
[Part III: Advanced analytics: All systems go]
Lots of organizations, in other words, think they're performing effective analytics by refocusing priorities based on evidence they've accrued via their electronic health records. But that may not always be true.
"There are sources of data you're used to going to because it's accessible, but you need executive sponsorship to find the right definitions for measures, the right data sources for measures, and hardwire it," says Fuller. "And that can't be done without leadership support."
There are many rubrics with which to measure an organization's clinical and business intelligence prowess. HIMSS Analytics has its DELTA Powered Analytics Maturity Model, which tracks providers on a five-step ascent (beginner, localized, aspiring, capable, leader). Another comes from analytics vendor Health Catalyst, which helps organizations find themselves on a scale from zero (fragmented point solutions) to eight (personalized medicine and prescriptive analytics).
(Or, for a metaphor that's more fun, the Salt Lake City-based firm offers the concept of a ski map, with trails representing various analytics projects marked green circle (easiest: opportunity improvement identification), blue square (more difficult: process improvement), black diamond (most difficult: outcomes improvement) and double black diamond (experts only: outcomes improvement with shared risk).)
It's the middle-ability "blue square" cohort of providers that may be most interesting to look at. They've got the tools in place and the organizational commitment to data-driven decision-making. They've already recognized some small successes in driving clinical quality and/or operational efficiency. Now they want to take it to the next level.
The advice here comes from some folks with advanced experience, but can be embraced by anyone who’s got the basic processes down.
At the Healthcare IT News/HIMSS Media Big Data & Healthcare Analytics Forum later this month (June 18-19 in New York), for instance, Fuller will present a talk aimed in large part, at that group. He bills it as “a more pragmatic approach for narrowing down the biggest bang for your buck in process improvement.”
Titled "Improving Value and Reducing Waste Through Targeted Analytics," it will show how sprawling Carolinas HealthCare – with its 900 locations and more than 60,000 employees – has managed to better focus its analytics initiatives, applying flexible business intelligence technology to a multitude of different variables to arrive at cost-saving insights much more quickly than in the past.
In his work at CHS' Dickson Advanced Analytics department, Fuller helps develop analytics models that can predict population health trends. Known as DA², the department is essentially a centralized analytics clearing house for CHS' entire clinical enterprise.
"We want to be apart of the planning cycle; helping us where to prioritize where to focus our attention," he says.
But as a centralized department, how does it make its insights felt throughout a health system comprising 41 hospitals with varying degrees of affiliation?
"We have a function within our team that's operations and communications," he says. "They're responsible for internal workflows, but also the way we face our customers, externally. They developed the portal for triaging ad hoc data requests." Newsletter communications and internal symposiums for sharing best analytics best practices are also key to spreading knowledge around CHS.
The buy-in has been gratifying, says Fuller: "What's been most effective is having physician- and administrator-led teams really orient themselves around what is the economics of the healthcare engine and the way we get paid, and to understand how their daily decision-making can be impacted through better information."
It's Fuller's job to "present physicians with the variables they need to understand before they make a decision, a checklist," he says. "Very simple: 'Before you use this device, here's some things you should know.' We don't want to inhibit the physicians from making the decision they feel best. We just want to present them with more information on the front end."
By and large, those docs are receptive: "More than ever, they're perceptive to economic conditions," he says, and understand that decisions based on data are usually the right ones.
"Mature analytics in a hospital needs to be consultative in nature," says Fuller. "You need to engage decision-makers. You're not going to them with answers. You're going to them with the tools and insights to help lead them to the right answer."
Tom Lawry, director, worldwide health at Microsoft, agrees.
"What we have to do is get leaders to start thinking differently, based on the fact that the boundaries that gated their thinking are going away," he says.
"Most American providers, most of what they're doing is dated by old data and old processes," Lawry explains. "Frankly, the technology is the easy part. The real challenge is getting people to think and act and work differently – to think, if there were no boundaries to data, how would I do something different?"
Slowly, that change in thinking is taking hold as more and more providers gain analytics maturity.
"I think there were a number of people early on that believed that the installation of a new EMR was the solution to analytics," he says. "Larger organizations get the fact that they need to make better use of all data, what's coming out of the EMRs." Nowadays, many more are "moving from old data, old processes, to real-time, predictive analytics, driving toward self-service or research-on-demand."
Have a plan
Joseph Colorafi, MD, is vice president and chief medical information officer for Dignity Health, the fifth-largest health system in the U.S., spanning 17 states. Its analytics journey, he says, has been exciting – but also a bit daunting: "We're trying to strap on a very powerful jet engine to the wing while rising to 10,000 feet," he says.
One thing that's helped steady the ride, however, is a certainty about strategy. Colorafi echoes Fuller's advice: Have a plan.
"Wait to aggregate the data and then decide the business use cases? No," he says. "It's the other way around."
At Dignity, which recently contracted with SAS for a cloud-based analytics platform, "we're getting very deep into using certain tools that can easily visualize information, without a lot of data curation, that our leadership can look at … that would have the greatest business use case."
Readmissions are a chief concern. Colorafi points to a program based in Sacramento called CHAMP (Congestive Heart Active Management Program) that has led to one of the lowest readmissions rates of all Dignity Health's regions.
But there's room to be even better. "Their nurses spend up to two hours trying to come up with some kind of one-time readmission score to best care coordinate those patients once they leave the hospital," he says. "We want to be able to create a 'FICO score' for that, that's standardized and automated."
As always, staff buy-in is key.
"Fairly quickly, we can bring up our proof of concept at a couple sites, have a knowledge delivery system, get ownership by the clinician and nursing leadership," says Colorafi. "But how do we scale that to the whole organization?"
One helpful strategy, he says, is to "make the data self-service. We've talked about knowledge delivery, near-real-time dashboards for sepsis, readmissions, etc. When you can give them a dashboard where the nursing leadership can go online and every nursing bubble there represents nurse communication and what that did to patients, that's a game-changer."
Charles Macias, MD, chief clinical systems integration officer at Texas Children's Hospital, is pursuing population health improvement on many fronts at once. The health system has deployed an analytics framework for pop health across its locations, driving process improvement methodologies with predictive analytics, decision support and more.
Macias has something of a unique title.
"My role as chief clinical integration officer was created to be transformational," he says. "Trying to understand how we merge the core of science, what we know about evidence – our own evidence, published evidence – with what we could obtain through analytics and predictive analytics about our populations, specific to our institution and our enterprise, and link all of that to quality improvement transformation.
That holistic approach has already led to some tangible benefits.
"One of the greatest successes we've seen is with asthma, one of the most common chronic conditions of childhood," says Macias. "By developing very comprehensive shared baselines, evidence-based guidelines across the continuum of care that were multidisciplinary, we captured the best science and best evidence systematically.
"We created care process teams that help inform our data architects and the models we created for analytics so we understood, through near real-time dashboards, the outcomes of care for these populations."
How are these for some "pretty amazing outcomes"?
- Cutting length of stay for children with asthma "pretty much in half"
- Reducing unnecessary testing for chest X-ray utilization, almost to a third of what it was almost a year ago
- Improvements in patient satisfaction scores
- Improved timeliness of care with steroids and beta agonists.
All told, that relentless, multi-pronged plan of attack has "made the patients and their families happier," says Macias. "It's made our payers happier, but we've also improved profitability by making care more effective."
And that's just for asthma. "We've seen similar results with diabetes and suspected appendicitis," he says.
But taking "small steps" and focusing on one project at a time was critical to achieving success in each of them, says Macias.
"Understand that you can just do one disease process," he adds. "Start with that. Create your care process teams; give them the data they need; give them the best science, and allow them to improve organizational outcomes."
Macias says he would "highly suggest taking one opportunity to identify your biggest champions: Who are your innovators? Help support that group. Give them the right tools. Give them an infrastructure. Give them the data governance. Allow them to meet with the chief information officer, with whatever groups do analytics, with whatever group does high-level physician support.
"And then to share those successes," he says. "What those stories are. Not just with the C-suite or the board, but to colleagues: We are, as a team, as a collaborative effort, improving outcomes."