There are few leaders in the health information exchange (HIE) space in healthcare IT whose work is more advanced than that of Devore (Dev) Culver, the executive director of HealthInfoNet, the statewide HIE in Maine. Culver, a well-known figure in the healthcare IT world, has been leading pioneering change at the Portland, Maine-based organization since 2006.
Most recently, HealthInfoNet has been moving into uncharted territory, leveraging a data analytics platform to create provider notifications based on clinical data, allowing physicians and others to know when patients in their care are admitted to hospitals or treated in emergency departments (EDs). Because of the maturity of HealthInfoNet’s work, Culver is in great demand as a speaker at conferences around the U.S. Indeed, he will be a panelist in the panel discussion session, “Health Information Exchange: New Models of Interoperability to Support Value-Based Healthcare,” on Wednesday, May 14, during the Boston Health IT Summit, sponsored by the Institute for Health Technology Transformation (iHT2). (Since December 2013, iHT2 has been in partnership with Healthcare Informatics through HCI’s parent company, the Vendome Group LLC).
In that context, HCI Editor-in-Chief Mark Hagland interviewed Culver recently regarding HealthInfoNet’s ever-expanding innovations. Below are excerpts from that interview.
Tell me about the latest developments at HealthInfoNet.
There are a number of pieces of good news for us, including our ongoing sustainability. As all the federal funding goes away, we’re making enough money to sustain ourselves through core subscriptions. We have annual subscriptions with different classes; there’s a whole pricing structure for hospitals, and a different one for physician practices; and within that physician pricing structure, specialists pay a little more than primary care physicians. We also have separate structures for long-term care and behavioral health. We recognize that some sectors, like long-term care, have very little margin; and yet they’ve become integral to the continuum of care around accountable care.
So at this point in time, every hospital in the state is under contract, and 34 of the state’s 37 are actively sharing data; and by the end of the year, we’ll have all of them connected. And we’re at 400 ambulatory practices connected, which encompasses about half of the primary care structure in the state. The specialty practices have been really slow to come to the table; at this point, I have one cardiology practice live. But employed practices that are hospital-based are already connected, and at this point, about 80 percent of physicians in Maine are employed by hospital-based organizations.
Why is it that most independent physicians haven’t yet connected?
I don’t think they fully understand the value to care management yet. The cardiology practice that is connected has set this up as a matter of policy.
So the physicians at that cardiology practice are recognizing the value to their patients?
Yes, absolutely. And the practice’s clinicians and staff used to have to call all over town, because their patients had been obtaining care at various locations. This is in the town of Bangor, which had one of the Beacon Communities. And we’ve got an FQHC [federally qualified health center], as well as the two hospitals in town, and specialty practices. Northeast Cardiology Associates is the name of the cardiology practice that’s connected; they’ve got 14-15 cardiologists, and in the past year, they became part of Eastern Maine Medical center. So we have a lot of specialists, but they’re hospital-employed.
And your connectivity across Maine has continued to grow successfully, hasn’t it?
Yes, we have over 1.3 million lives in the database; Maine is a state with 1.3 million people, but there are also out-of-staters who have data in the HIE. But we’ve got 90 percent of Maine residents with data in the HIE. And our opt-out rate remains at about 1.1 percent. So we’re seeing over 1.2 million messages a week being shared. Our user population is up about 2,000 users, and last month, we hit 22,000 patients accessed by those users. And that grew over 200 percent in the past year. Also, in January, we started doing real-time notification for inpatient admitting and discharge and ED admitting and discharge. So last month, we provided 8,000 notices of patient action, whether admission, discharge, or lab result coming in. And so care managers can associate themselves with patients. That’s really taken off significantly.
I definitely want to hear more about that. Broadly, with regard to the panel discussion that will take place in Boston on May 14, what are the most important steps healthcare leaders need to take around HIE for accountable care and value-based care?
They’ve got to clearly establish and demonstrate value; as with notification activity. But we’ve taken that to a whole new level this year. And we’ve got a couple of major projects right now around the Veterans Administration. In fact, 12-13 percent of the population of Maine is veterans; we have the fifth-highest ratio of vets to general population in the country. And we received a three-year-grant from HRSA [the federal Health Resources and Services Administration] to get the VA fully connected to us, and we’re involved in that right now. And we’ve received a SIMT grant—a States Innovation Model Testing grant. Those grants are all focused on significantly increasing the percentage of the population covered under some form of risk-based contract. In Maine, the goal was to help us get to 80 percent of the state’s population being covered by some kind of risk-based contract; we’re at 30 percent right now. But keep in mind that Medicare and Medicaid in Maine are already around 50 percent. It’s an older state and poor state.
But the most exciting thing, and this is where I think the HIEs are particularly well-positioned if they do it right, is entering into the world of near-real-time analytics. So we’ve got a set of analytics based on clinical and encounter data, predicting things like when a patient will be admitted or visit an ED. This is live now; we started the validation testing in September of last year; and we’ve got three contracts being developed—one in hand, St. Joseph’s Hospital in Bangor. We’ve got others coming up in the near term. These are organizations engaged in managing populations. Two of the factors are focused on a six-month timeframe—who will be an inpatient, who will show up in the ED, and who will be an expensive patient to manage, in the next six months? And this model is pretty sophisticated, and was originally created to predict variations in the genome. The company is HBI solutions, and we’ve been working with them for the past two years, and now, basically, we’re offering a production service to providers in the state.
That’s very impressive.
It’s pretty cool. I don’t know of any other HIE trying to do this with clinical data. You’re looking at the encounter data, so coding, but also labs, medications, and diagnostic imaging studies. And they’ve done a variant analysis that basically lines up predictable events. So you can take a patient and see what’s driving their risk, in any one of those three categories. So that’s an example of really understanding the value proposition. From the outset, HealthInfoNet decided it was (back in 2005-2006), going to take the extra time and effort to map as much of the clinical content as possible to clinical terminology—so, SNOMED, LOINC, NDC [a pharmaceutical cataloguing structure], ICD-9, ICD-10. So all that hard work is now paying off, because it standardizes the results across the state; therefore, you can actually do the analytics, because you can standardize data from across the state. So this investment from six or seven years ago is really helping us now. And whereas most of the modeling systems done today have been created with claims data, which is three to five to seven months old, in this case, the next lab result or diagnostic test, changes the calculation. So it’s pretty cool. And this is something that individual provider organizations don’t have. And your data is basically minutes old, and you put a structure around this that can turn into actionable data.
What are the mechanics of this?
So basically right now, you as a care manager or physician, come into a portal environment, with a data warehouse sitting underneath it. And you’re provided with filtered views of the data in the database. So when we’re looking at the population risk patient view, we’re looking at who’s going to become an inpatient or ED patient or an expensive patient, and you can look at date range or diagnosis, or whatever; so you can basically get down to a list of your patients who are at certain risk levels, based on the community data on that patient, not just from your institution. And the 30-day readmit is the same basic mechanics, but run while the patient is in the bed.
Do you know of anyone else doing this?
No, not from a clinical standpoint. But I’ve got to believe that others are sitting there with the same concepts; what they lack sometimes is access to a large clinical database that’s standardized.
Maine has enjoyed an environment of collaboration among providers statewide, correct?
Actually, we’ve had problems in the past, not because people weren’t willing to share data, but because of technological issues. In Maine, people are no less competitive, but they’ve come to understand the value proposition of having a shared vision. The hardest thing is to have them understand that patients aren’t loyal. The numbers are actually amazing that way. And what we’ve tried to do, ever since I came into this role in 2006, from the board and management team, was that we would make a profit to keep the lights on, even though were a non-profit. And we’re in a good place. We also happen to be a data course for others, who put it into their analytics. There’s an organization in the Northeast formed through investment partnership between Eastern Maine, Maine Health, Dartmouth College, Dartmouth-Hitchcock, and Fletcher Allen Health. Those five organizations have invested in an analytics company called Northern New England Accountable Care Collaborative; we basically feed data to them for their ACO. But we’re also running our tools, too. It’s coopetition, really. And some days are better than others.
What would you advise our core readership of CIOs and CMIOs to think about, in all this?
From my perspective, that balance of competition versus coopetition. I like that. There is a value proposition in exposing yourself to that cooperative side of the equation, and I think that calculus needs to be thought about very carefully. And while I’m already starting to see some issues of channeling, trying to keep people within your structure—and we’re seeing some of that coming back again—but I think for your readership, the question of value or opportunity lost as a result of competition, is something that ought to be thought through and weighed. It’s not right for every market, but it’s a tradeoff between what you think you hold, and a competitive advantage that could actually be more via synergy. That’s the hard calculus that people need to make. It works in Maine, but that’s only because we’ve made it work in Maine.
And our marketplace, like everybody else’s, is changing very rapidly now. We’re seeing consolidation, an increase in physician employment—over 80 percent of our physicians are employed now. So what does that mean to an exchange when you’ve only got a few large systems, and no independents? That’s a challenge to face. We’re in a really wonderful place right now; the work that’s been done in the last seven or eight years is positioning us extremely well for the moment; but it could be a totally different scenario in two or three years, and we’ll have to adapt to changes to continue to be successful.