At the five-hospital ThedaCare integrated health system based in Appleton, Wis., senior executives, and quality, clinician and IT leaders have been busy visioning the future of healthcare. For several years now, the organization has been deeply involved in applying Lean management principles to care delivery across the ThedaCare continuum. Indeed, the health system’s former CEO, John Toussaint, M.D. became such a proponent of the application of Lean principles to healthcare that he founded the ThedaCare Center for Healthcare Value, and has been a key organizer in the Lean Healthcare Transformation Summit, an annual event that was held this year on June 4-6 at the Hilton Lake Buena Vista in Lake Buena Vista (Orlando), Fla.

While attending the Lean Healthcare Transformation Summit earlier this month, HCI Editor-in-Chief Mark Hagland interviewed Julie Bartels, executive vice president of national health information at the ThedaCare Center for Healthcare Value, to talk about the new Clinical Business Intelligence Network that the Center for Healthcare Value is participating in. Hagland also spoke with Brian Veara, Clinical Business Intelligence Program manager at ThedaCare, regarding his current work in the clinical business intelligence area at the health system. Below are excerpts from that interview.

What do you do at the ThedaCare health system?

I’m director of decision resources. So my responsibility is working with all data and information from a clinical and operational perspective—anything clinical and operational. From a Lean perspective, I manage the data and information needed for that. We have 27 clinics now and five hospitals [four owned and one managed], and we have both senior services and home health and employee health. And then we also have about 30 independent specialists I work with.

And we do a lot of quality outcomes and other reporting. Our reporting encompasses meaningful use-related reporting, reporting for PRQS [the Physician Quality Reporting System], immunization registration, and other areas. And we host the reporting capability for Bellin Health in Green Bay, using our same Epic license [from the Verona, Wis.-based Epic Systems Corporation]. In fact, Bellin saves $5-7 million a year by being on our system for quality reporting.

And your position is embedded in the IT department at ThedaCare?

When I first started, I reported to the chief medical officer, who’s now our CEO, and that was intentional. They wanted the primary direction of our work to be clinically focused and improvement-focused. When our CMO became CEO, the new CMO didn’t want that responsibility, so I now report to Keith Livingston, our CIO.

And your organization recently became a Pioneer ACO under the Medicare Shared Savings Program for accountable care organizations, correct?

Yes, we’re a Pioneer ACO. And there are many data and outcomes reporting challenges involved in participating in the program. They drill down to such issues as the fact that the Pioneer ACO Program’s definition of a diabetic is different from the definition of a diabetic under programs from other payers, for example. So there are a lot of very detailed things to work through.

Some of the challenges have been procedural?

They’re related to physician engagement and the reasons we’re doing it. There’ve been some technical issues, but those haven’t been the big struggle. There have been clinical definition challenges, but it’s more been around why physicians should be engaged. The thing is that the physicians realize what’s going on in general with Medicare, but they don’t understand the risk element; and putting some of their clinical performance at risk has been challenging for them. And there have been issues around defining patient attribution. Some of those have been “purified,” I would say. CMS [the federal Centers for Medicare & Medicaid Services] has actually done a very good job communicating and trying to do their best to make this work. And there’s a lot of materials available, and explanations.

But when you drill down to detailed information and data levels, things get complicated. For example, some of the attribution logic is really complex. For example, one of the measures is LDL management. And who “owns” the management of that, from the standpoint of rewarding or penalizing a provider? If a patient is seeing a primary care physician and a cardiologist, who is responsible for managing your LDL level? They’re using a really complicated algorithm to determine that.

So in a way it’s both, right?

From a patient perspective, yes. I can also see that if one is not managing me well, their job is to inform the other of it, and the process around that is not really well-defined, either. There are very thick walls of silos in healthcare, and the workflow doesn’t support it. One of our physicians who’s in our network says he sees one patient at a time and is used to thinking in that way; population health is an abstract for him. And as you become population health-oriented, those communication breakdowns have become more apparent. We’ve been live on Epic for 13 years now.

The irony—and Keith Livingston, our CIO, can articulate better than I can—is that it was doctors’ peers being tired of having to fax things to them, that pushed some of the lagging doctors towards EHR [electronic health record] adoption; they said, we want you on Epic. Their peers were threatening to stop sending their patients to them. So that moved their colleagues forward.

What are the big-data challenges and the data analytics challenges?

One major challenge is that not everything we do from a clinical perspective gets into an EHR; not everything that is entered shows up on a claim; not everything that shows up on a claim gets paid. So you have to understand the perspective and context. At first, everything was focused on trying to analyze claims data. Now, the service data—when did something actually happen—becomes the most important. The question becomes, how many patients did I actually see yesterday? We average about 15 percent of our cases month to month not getting coded from previous months; typically, there’s a three day lag. I’m talking about in an ambulatory setting. The entire financial structure is driven by posting. And some of that data gets kind of squishy because it’s not coded yet. We know the activity happened, it’s in the EMR, but it still has to get coded. So the real-time data and real-time analytics become a challenge if you haven’t yet optimized your processes around documentation.

Isn’t it also tough that you are having to go through the transition to the new ICD-10 coding system while you’re doing all this other process and data work, and change?

Longitudinally, yes. It’s an analytical challenge; it’s really not a data challenge. You set the rules. CMS actually created the crosswalk to them, but in the case of one to many, you’ve got to pick one. Now, the good news is, the people who have really studied this have said, among the old codes, they only used a small percentage of the codes. They’ll probably end up using among the most frequent codes they’re already use; and ICD-10 is far more specific, of course.

I’m actually going to be more worried about the payers than our providers. Medicare a few years ago put in preexisting conditions—2010—documentation. December 1 was the date it was supposed to go live; almost every payer wasn’t ready; the government was.

What should CIOs and CMIOs at other organizations be doing right now to prepare for all the kinds of work that you and your colleagues have already undertaken?

I think you can’t lose focus on why we’re doing this. This is about the patient, and about delivering better care to the patient. At the end of the day it’s about, are we doing a better job taking care of patients? And we also can’t lose focus on the people who provide the care. It’s not about the technology, but about the caregivers and the workflow.  And when you look at an EMR, it’s designed to deliver care and get a claim paid; it’s not actually designed for care management.