Article image

PDF version

of Gatefold

About the series

As health care moves rapidly toward a value-based delivery model, a greater emphasis will be placed on care coordination. We must ensure that patients not only get the right care at the right time in the right setting, but also that every part of the delivery system is connected and understands that a patient's need will be critical going forward. Information technology will be instrumental in making sure that these connections take place and in providing clinicians with valuable new decision support tools.

H&HN, with the support of AT&T, has created this yearlong series called Connecting the Continuum to explore how hospitals and health systems are addressing the care continuum in their strategic and operational plans. Each month, we will examine such topics as health information exchange, mobile health and transitions of care. Follow the Connecting the Continuum series in our magazine and in our e-newletter H&HN Daily.

Analyze this: You've assembled a strategically sound continuum of care. Information systems are connecting each site to one another as well as pulling data from all of them. Data are flowing — gushing — into a data warehouse, coalescing into registries that identify people at health risk. But for the most part, it's just potential for action. Where do you find the know-how and technological tools to harvest it all?

"We're grappling with the same issues around data as anybody else," says Adrian Rawlinson, M.D., director of medical informatics for Brown & Toland Physicians. "Are we creating data for data's sake, or are we actually creating good, actionable data?" The 1,500-physician independent practice association based in San Francisco files more than a million test results a month into an information network that includes an enterprise-level database. Only about 250 physicians are on the practice's electronic health record, but they're "the main chunk of our high-volume, preferred physicians," he says.

Brown & Toland tried to create its own data analysis and reporting tools but, ultimately, scrapped the expensive and "clunky" attempts, says Rawlinson, in favor of technology and services provided by Humedica, one of a slew of health care data-dicing firms stampeding into the population-management market opportunity. A recent report on the state of IT readiness for accountable care listed a baker's dozen firms, each with a sliver of market share [see chart].

These firms tend to be turnkey approaches to pulling data together from various sources, says Colin Buckley, strategic operations director at KLAS Enterprises, which co-authored the accountable care readiness report with Leavitt Partners. They have their own data repository for creating registries, sorting data and guiding provider priorities. Some vendors incorporate built-in intelligence based on what they've learned offering services in such other areas of health care as financial reporting, says Buckley.

One useful aspect is the "pursuit list," a breakout of certain types of patients who need set services aimed at preventing or managing illness, Rawlinson says. "For example, we could push a list to a primary care doc's office to say, 'Here's a list of patients 50 years and older who are overdue [for] a colonoscopy.' And then they would do whatever they want with that list." Brown & Toland can look at severity of illness scores, sort by ZIP codes or financial status of a patient, and stratify risk among people with diabetes, obesity, heart failure or heart disease. "All these things are now measurable," he says.

The next challenge, once health care providers organize their information trove for population-based management, is to bring relevant detail to the action level. Population health has a global feel to it, but basically it's a means to think broadly in order to act pointedly, Buckley says. "It's about narrowing down our efforts to where we have the greatest return."

Analytics used to be the province of IT professionals who had to build preset analytical queries proposed by clinicians, but now such tools are becoming more flexible, allowing enterprises to dream up analyses on the fly and run follow-up queries over a matter of hours instead of the days or weeks it once took, says David Krueger, M.D., medical director for 700-physician Bellin Health-ThedaCare Healthcare Partners in northern Wisconsin. It uses a tool called QlikView that is "turning the light on in the room," he says, performing and returning successions of queries "at a phenomenally high rate of speed."

Easy to use, these tools allow all clinicians to jump into clinical analysis, says Joseph Kimura, M.D., medical director of clinical reporting and analytics for Atrius Health, an alliance of six medical groups in eastern Massachusetts. The product it uses, from Verisk Health, is "a really nice, sort of exploratory tool, available 24/7, [that] doesn't really require human support," says Kimura. "Any doctor in the organization at 10 o'clock at night who has a brilliant idea can log in and see if his idea has merit."

Case Study

Managing patient health in return for fixed reimbursement leaves no room for error, or mystery. Bellin Health-ThedaCare had used homegrown patient registries and other clinical IT for years to perfect care processes and get to know its patients, but registry volume was limited to 15 or 20, and clinicians had been given certain pre-decided views of data. "You're stuck with what you developed," says medical director Krueger.

The QlikView analytical tool, which has been in operation nearly three years, now enables clinicians to "quickly decide to look at data in a different way. I can say, 'What I really want is: How many of my 39-year-olds are missing this measure?' " says Krueger. With a few hours of programming time, "the database is already set up to deliver it."

The application then can be asked question after question based on previous results, instead of having to develop a new query each time and get it back days later from a programmer. "I can go into the same database that we would have struggled with years ago and ask a question, run it, minutes go by and then I can say, 'No, that's not quite what I wanted.' " Tweak, run it again.

Case Study

Atrius Health sees benefits from the data analytics product it uses, but along with better ways to visualize and navigate through information came a new challenge: minor data differences with real impact on results. Insurers and vendors make assumptions on how they define and calculate measures, and those definitions can be just different enough to affect analysis of a situation being studied, says analytics director Kimura.

For example, when calculating percentages of generic products used by providers, Verisk Health would include non-pharmaceutical elements such as test strips in the overall measure, while Atrius excluded a lot of those elements. So, percentages would vary 5 to 10 percent based on that different starting point rather than actual variance among providers, Kimura says.

Billing variations also can be consequential, he says. Certain insurers may bill an aggregate lump sum for a service while others break the fee apart by professional and facility, "and that leads to variation by case per unit and total dollar amount that dumps into the analytics system."

Rather than have Verisk pull data directly from insurers, Atrius took the feeds in first, standardized them across all insurance products and then sent that information to Verisk. It added another month of lag time "but it eliminates at least some of the variation," Kimura says.