HealthInfoNet, Maine’s statewide HIE, started as a feasibility study in 2004, sponsored by the public/private partnership of the Maine Center for Disease Control and Prevention (Augusta), an organization that at that time was called the Maine Health Information Center (now the Manchester-based Onpoint Health Data), and the philanthropic Maine Access Foundation (Augusta). After engaging providers, consumers, and others, the actual planning process started in 2005 with coming up with core principles on what the exchange should address and a clinical and technical group that put together the first technical specifications for the exchange. Fast forward to this June as HealthInfoNet just finished its two-year demonstration phase with the state’s six major healthcare organizations, which represents about 50 percent of Maine’s population, sharing patient information. The organization is now plowing forward to get the rest of the state on board. HealthInfoNet’s Executive Director Dev Culver, previously the CIO of Eastern Maine Healthcare (Brewer), spoke with HCI Associate Editor Jennifer Prestigiacomo in a two-part interview.
Healthcare Informatics: Can you give me a little background on HealthInfoNet?
Dev Culver: In 2006 [HealthInfoNet] was incorporated as a not-for-profit and built as a public/private stakeholder organization. And what that means is, the board is a community board representing five stakeholder communities: consumers, providers, payers, business, and state government. There were also two standing committees created: a consumer advisory committee, which is populated by advocates like Maine Civil Liberties Union, Planned Parenthood, and other organizations that engage to support us in development, policy, operations, and communications; and a technical and professional practice advisory committee, which advises the board on where the exchange is heading.
We spent two years banging around and raising money. [He became executive director in June 2006.] In the very beginning of 2008, we executed a technical contract with 3M as our primary contractor to build a two-year demonstration phase, and that work began in February 2008. It involved six provider organizations, which represent facilities all over the state of Maine. Four of them are the four large delivery systems in the state, so pure go for volume early. One was a small, independent, rural hospital—we wanted to find out how hard this was going to be. And one was an ambulatory group practice with multiple locations around the state, so we wanted to understand the process around group practice workflow.
Those six healthcare organizations, which is 15 out of the 39 hospitals in the state, represent about 50 percent of all clinical care activity [Maine’s population is 1.3 million], so it’s a fairly healthy starting point for a demonstration phase. Part of the demonstration phase was connection to the public health system through Maine CDC to see if we could demonstrate the viability of automating lab reporting for mandated disease control.
It took about a year to build out the information exchange technically. We have a fairly robust clinical data set. It’s based on the continuity of care record, so it involves person identification, allergies, prescription medication, lab and radiology results—we’re not doing images at this point, just printed results—diagnosis, visit history, problem lists from the ambulatory space, and other transcribed notes. So every single registration event at any of those hospitals flows through us. We get our prescription data from the commercial side from [St. Paul, Minn.-based] Surescripts, and we are just now providing access to the Medicaid prescription data.
So, the first clinical use of the exchange started in July of 2009, and our demonstration phase officially concluded in June of this year. Now we’re taking those lessons learned and four months to retool. We’re changing out some of our vendors.
HCI: What vendors are you’re changing?
Culver: We’ve decided to change out having 3M serve as the prime contractor, and are in the process of taking out the 3M software that was in place. We’ve added two new vendors, Initiate [the Chicago-based Initiate Systems, recently acquired by IBM] to support our master provider and master patient index and we added [the Denver-based] Health Language Inc. to support our data dictionary and terminology coding needs. We’re very big on standardization and coding. So, all of the lab results are coded to SnoMed CT. Orion Health [Santa Monica, Calif.] stayed in place.
HCI: What did Orion provide that 3M didn’t?
Culver: In addition to serving the areas they did during the demonstration phase, and that was primarily the front-end with Rhapsody [Orion Health Rhapsody Integration Engine] as the interface engine and the backend with Concerto [Orion Health Concerto Physician Portal] as the presentation layer in the back [that leads] into the EMRs. They are now going to be responsible for the clinical data repository, which 3M had previously been responsible for. Clearly, 3M has a good product, but it’s getting a little old in some respects and pricing became a major issue.
HCI: Does Orion have any other U.S. HIEs deployed?
Culver: They’ve got a bunch in the queue that they are negotiating, but we’re kind their flagship in the U.S. They just won Singapore, which is no small feat. They’re in a bunch of provinces in Canada, New Zealand and a couple of states in Australia.
HCI: Is the HIE platform a federated or central repository?
Culver: It’s a central repository model. That was by design in 2005; you just can’t do public health reporting without that structure. That’s just a testament to all the work we’ve done developing trust. Because at the heart of the clinical data repository model is a trust model. The organizations participating are very much competitors, and in 2005 they made two very critical decisions that allow the clinical data repository model to even work. First, they decided to comingle their data. Secondly, they decided to stop competing with patient data. There was a lot of consensus building going on. Maine’s a very collaborative state; there’s a pragmatism in the state that’s pretty obvious. I think the organizations recognized that they’re becoming very visible targets because they’re expensive and large, and they needed to do some things to clearly impact the quality and cost of care. We became a very viable project.
HCI: What were some of the other lessons learned from the demonstration period?
Culver: Well, importantly, and technically it’s doable. [laughs] The potential there is for some real impact on utilization. The qualitative survey told us that given the right information at the right time, they [physicians] don’t order additional tests, and that’s an important lesson.
We also learned what the value propositions are. The speed to access is a huge value proposition for the provider community. The scope, seeing things they never saw before, is a huge value proposition. And part of the survey identified there was a high degree of trust in the data, but what that translated as a value proposition was expanded trust in what they were hearing from patients. In the demonstration phase, prescription medication information became the No. 1 valued [piece of] clinical data. It’s an easy win because it’s the most complicated area.
The other thing we learned is that getting adoption from physicians is harder than you think. It’s not because they don’t want to use the information, it’s just you’re asking them to change their workflow, and for many of them it looks like another step they have to do. So what we learned is focus on the workflow; find out who actually does the work, and get them connected to the exchange. For example, in the emergency room, it’s important to have the ED docs have access, but focus on the triage nurse because they’re the first point of contact, and you can also routinize (sic) the process of intake. Another lesson learned around prescription meds is to engage the pharmacists. In many institutions they’re carrying the medicine reconciliation process. I think incorporation of workflow is hard, but I think we did a pretty good job of reducing the barriers of access. You can get from inside your medical record to the exchange, not leaving your medical record, in three clicks. You do have to go through a security step and a break the glass step, associating yourself with the patient, but once you do those two things, you’re actually in the patient [record]. You don’t have to go search. Now, we did not focus on literally moving data into the [physician’s] EMR; that was not the intent of the demonstration phase because quite frankly the technical specs to do that weren’t terribly mature in 2005 and 2006. So the CCD-based exchange will come online next January.
The other thing we learned is that ‘if you can keep me from going out there because there’s no new information,’ that’s greatly valued. We actually have a solution that will be implemented shortly where if you click on the exchange inside of your EMR, a sign-in page pops up that will actually tell you if there’s new information on this patient you haven’t seen. When it fires off the link to connect to the exchange, that message incorporates some key things. It knows who the provider is, what organization they’re associated with, and who the patient is. So, what it’s able to do is look at that provider against that patient.