HIStalk Interviews Michael Oppenheim, MD, CMIO, North Shore-LIJ Health System

Michael Oppenheim, MD is CMIO at North Shore-LIJ Health System of New York, NY.

What is North Shore-LIJ doing with interoperability and HIE?

I’ll start at the end and then I’ll back up and explain the thinking that led us in this direction. 

We are making a huge investment — time, personnel, and focus — into developing an internal HIE, health information exchange. The reason we’ve done that — and I think a lot of other large integrated delivery networks have come around to this way of thinking — we were very eager participants when New York state initially put out the request for proposals to develop a number of RHIOs within the state. They’ve since consolidated — the HIEs, the RHIOs from across the state — into a single structure, SHINY, the State Health Information Network of New York. 

In the beginning, when you talked about doing internal HIE within an organization, everyone assumed that you were somehow trying to be exclusionist or not participate in the HIE. That’s far from the case. I think the state has come around on that, and many other IDNs have come to the realization that the kind of interoperability that we want to do goes much beyond what the mission and goals of the RHIOs are. 

The RHIOs are very, very much focused on getting as broad a look at patient information as they possibly can. That’s great because they help broker the politics and provide a common playing field for organizations that may be competitors in the marketplace, but are willing to jointly share data through the third party of the RHIO. You create a huge, consolidated record that people can go to and get a comprehensive look at the patient beyond just what they know from their four walls.

But there are a couple of things that HIE has come to mean to some of us that is beyond the scope of what the RHIOs or HIEs are focusing on today. One is around actionability of the data. The second is around not just aggregating and displaying data, but actually literally moving data from point to point without human intervention. 

The user experience with the standard type of HIE implementation is that the clinician first goes to the HIE or the RHIO to look up what history is there about this patient who I’ve never seen before. Then you go and you actually do your clinical documentation and order entry and everything else, in whatever transactional system, whatever EMR you happen to be using for your environment where you’re caring, whether it’s an office- or a hospital-based practice. 

There are two limitations to that in my mind. One is the intrinsic dissatisfaction with having to go to two places to look at data. The second being that the data out in the HIE is not necessarily actionable. I’m ordering a medication in my EMR and there’s a lab test that hasn’t been drawn in my office or hospital, but it’s known in the RHIO. Based on that lab test, I need a dosage adjustment or there’s a contraindication to the medication. My decision support engine doesn’t see that external data. 

Our focus has been on looking at how an HIE can bring data right to the clinician so that he or she can have one place where they do all their work, as well as have more of that data available for a decision support engine or for any rules or analytics or other things that you want to do on your data set, and have it all consolidated. 

We look at the HIE opportunity because internally, we can do a lot more. There’s a lot tighter integration and have a lot more actionability of the data by having an internal HIE that we control, that’s covered by our consents, and any number of other things that are facilitated by having an internal HIE.

We’re an Allscripts shop. We’re using Allscripts TouchWorks in the practice environment and  Sunrise in the hospital environment. We made this decision before the dbMotion acquisition and before some of the newer interoperability tools that they produced. Let’s put that off to the side for a moment.

The workflow that we wanted to enable was what we built so that when a patient comes to the emergency room, we pull a summary from the ambulatory environment. We place it into the Sunrise record, so it’s available and visible to the docs in the hospital. They don’t have to go out and look somewhere else. They don’t have to look at the HIE’s viewer, don’t have to look in the ambulatory record. It’s right there in the hospital environment.

At the same time, we use the data in some actionable ways. We’ve certainly done more sophisticated things than this, but even on the most basic level, we can fire off a notification. We can put a task notification in the task list of the primary care doc to say, do you know your patient is in the emergency room? If and when that patient gets admitted, we fire off a second notification saying, by the way, not only were they in the emergency room, they have been admitted to the hospital.

We begin to start to getting into what’s really business process management around the transition of care and moving the data for the user. Not requiring them to push it via Direct or something else — by sending off alerts and notifications to the primary doc so they can communicate with the hospitalist. That’s just one of the more basic examples.

To us, the HIE is much more of a process orchestration engine, not just simply a repository of data that someone can look at. It’s actionable. It’s delivered to the clinician when they need it, where they need it. That’s been the driving philosophy behind having an internal HIE rather than simply rely on RHIOs or outside entities.

The example I gave involves an ambulatory practice and a hospital. Certainly in some environments where you have a consolidated platform, maybe that’s not the most important use case. But even in hospitals that are using systems that share a record with their ambulatory facility, there’s always going to be other facilities in a large, integrated delivery network that’s not going to be on the common platform. We have nursing homes. We have a home care company. We have numerous other types of business entities that are relying on this flow of data so that their providers can work most efficiently in what we call the home system.

Whatever you’re used to work in, that’s where we want the payload delivered. That’s where we want alerts and notifications and things to arrive. That will be orchestrated through our HIE.

Will HIEs be challenged to provide business value to offset the cost?

If you look at where our future revenue opportunities are going to be, we’re moving away from our fee-for-service world and very much moving to the risk-based contract in a capitated world. We have numerous risk contracts with commercial partners. We’ve just launched our own insurance company, North Shore-LIJ CareConnect.

To us, orchestrating business process, eliminating redundancy by making sure that everybody’s got full access to the full corpus of clinical data, having a decision support engine that sits and looks at data and reacts to data across the entire health system … I couldn’t hand you a document today that says, “Here’s the amount of dollars I expect to improve my pay-for-performance and here’s how much I expect to cut my readmissions and here’s how much I expect to XYZ.”

But conceptually, we are all bought in that our entire financial success of the health system depends on the successful conversion to be able to do capitated and risk-based contracting. We don’t think we can do that without an HIE to coordinate the transition to healthcare managers and care navigators who identify patient activity, figure out who’s been where, get notifications when things happen that shouldn’t have happened, or get notifications when things that should have happened didn’t.

The HIE, for example, has in it the full ambulatory providers schedule. We can find if a patient has an appointment that’s been missed. We can fire off a notification out of the HIE. The HIE is so much more than information exchange.

The HIE platform also has registry function that allows us to load programs into it. If we have a heart failure program, we can either manually or automatically load in that these are all patients with heart failure that are part of this program. Or patients coming in with a certain payer. We can go into that payer registry and then we can make sure we do the right notifications to the right coordinators of those programs as either activity that should happens but doesn’t happen, or activity that shouldn’t happen but does happen, like unexpected specialist visits or ED visits or things like that. 

As an article of faith, we fully believe that in order to truly be able to coordinate care as an integrated delivery network and provide population health and be able to be financially and clinically successful in capitated arrangements or among our own insured population, the HIE has to be a critical enabler of that. I don’t have a specific financial ROI sheet that I can wave and say, “This dollar is going to be offset by that dollar,” but absolutely the direction of how the health system expects to care for patients in a longitudinal way and a holistic way requires this kind of technology.

Do you think the demands of population health management have turned the expectations for HIEs upside down? I’m referring to the RHIO-type organizations.

I’ll answer that in two ways. We’ve always had this intrinsic discomfort, as I started off by saying, that I’m going to look in point A and then point B and then point C, which is why we use the HIE as central consolidation point  to create a single, consolidated, comprehensive record which we can then push forward to the provider just in time as an encounter is about to happen. We anticipated that that kind of clinician reaction had to be overcome. That’s exactly why we did the things we did — get it in their face and not make them go hunt for it.

But how and when will the RHIOs retool? I think they have to. It’s really not as much their onus as it is the onus of the providers who are going to be held to different types of accountability standards to take on the responsibility to go search and find all of that data. That really is putting a tremendous burden on your providers. The value proposition goes up, but it’s on the back of the provider more than it’s on the back of the RHIO to do anything different.

The one thing, though, that I will say … I’ll editorialize a little bit … is that the RHIOs are being fundamentally pulled in the wrong direction on a lot of this stuff. Because at least in New York state, the privacy concerns around the RHIOs are, if anything, driving more and more and more restrictive rules around access to the data, sharing of the data, then sending us data. Within the context of a single organization that we control, we manage the consenting process end to end. There’s a lot more we can do.

When you get out into the state level or eventually the national level, a lot of the good intentions and the good clinical opportunities are potentially going to be stymied by the restrictive practices and policies that are being built around the RHIOs because of the patient privacy concerns. I don’t mean to minimize the privacy concerns. They’re certainly real and legitimate. But what they ultimately translate into from a regulatory statutory perspective, at least in New York state, runs a little bit counter to what we’re trying to accomplish by saying, hey, wherever this patient goes, we’ve got to be able to assure that everyone’s on the same plan of care. Everyone knows what’s already been done. It’s going to be very tough in the governmental RHIOs because of the privacy concerns and what they’re driving from a policy and practice perspective at the RHIO level.

You mentioned your Allscripts implementation earlier. I’m curious about how that’s going, especially now that they’re retooling into a population health management company.

It’s going well. We made this decision before they came forward with their dbMotion acquisition and some of the new tools that they’re bringing forth, which we’re very excited about. We just met with a number of them a few weeks ago. We have a whole bunch of folks coming on site.

We’ve been talking about population health management, trying to understand the respective roles for our internal HIE for what they’re trying to do to bring their products together. The newer front end that they have been talking about which fuses dbMotion with their front end products to make the community data and the local data appear seamlessly to a clinician look like a very, very attractive set of work flows. We are in detailed discussions with them about how we merge some of the things that we’ve done or are doing internally with some of the things that they’re doing, because we did set off on this track a little bit ahead of them.

What are your biggest challenges and opportunities as a CMIO over the next several years?

The HIE is probably one of the biggest. People think of it as a technology — and there is a lot of very, very valuable technology – but the HIE alone, just simply “data comes in, data goes out,” doesn’t accomplish the mission unless you build lots of clinical workflows over it and around it. You’re supporting any number of clinical programs or any number of potential patient flows or workflows. I have a big team focused just on that, which is working on how we take the power of the HIE and apply it to all the various different programs that are growing up around the system. That’s probably one of the biggest.

The other major area for us as a health system is the development of a data warehouse, which we don’t have today. We have a lot of individual analytic tools and products attached to our various EMRs, plus other types of warehousing — cost accounting, things like that.

We still have work to do with our EMR rollout. We still haven’t put physician documentation out beyond the inpatient space, beyond the admission and discharge documents. We still have to build out progress notes, consult notes, and a couple of other things. We still have about 30 percent of our medical group to whom we still have to roll out our ambulatory EMR. Those are all still in progress.

But my overall goal is to look at, as we make this transition to a different model of care, how do we orient everything we’re doing in the EMRs, align it with everything we need to do in the data warehousing space to provide the analytics that are needed to support these programs, and align all that with all of the clinical workflows that we’re building in the HIE to support the population health types of initiatives that we’re doing with the HIE? Making sure that all these three things work together properly, that they don’t overlap each other in what they’re doing, that we don’t leave gaps where I thought the HIE would do that or the other warehouse would do that. To make sure that all of these things align together to support all of the population health programs that we’re engaged in.