DHIN project manager Melissa Macolley notes that this long-term strategy of adoption over functionality was partly based on her belief that HIEs have only one chance to roll out a product correctly or risk losing stakeholder confidence and utilization. She says “quality over quantity” has been DHIN's motto from the start. Proof of its adoption efforts, DHIN has garnered 75 percent of Delaware hospitals and 80 percent of providers as members, as well as having brought two of the largest laboratories in the state, LabCorp (Burlington, N.C.) and Quest Diagnostics (Madison, N.J.) onboard. By the end of this fiscal year, Lee hopes that 80 to 90 percent of hospitals, several radiology groups, and major health plans will be DHIN financial contributors. DHIN also plans to roll out value-added services that align with MU requirements like public health immunization reporting, medication history, and CCD exchange.
Selecting a vendor that has a variety of core services was an important early step for the Kansas Health Information Network (KHIN), says Laura McCrary Ed.D, KHIN's executive director, since her state has providers in different stages of EHR adoption. KHIN sought a vendor that had a portal for practices that were still paper-based, an EHR-lite product for organizations that would not be implementing full EHRs, and functionality for organizations with EHRs. Over time as stakeholders' needs emerged, more products like secure clinical messaging were rolled out. “We need to be able to provide to our stakeholders the ability to connect through the Direct Project (for more on this, see page 14), so we asked our vendor to make sure we had a Direct product that we could roll out also,” adds McCrary.
TAKING A HYBRID APPROACH
Many in the health IT industry are seeing the technical architecture issue shifting from being less of a black and white issue to more of a gray one, as many HIEs are finding hybrid architectures suit their needs for core and value-added functionality. In a hybrid model, each organization manages its data separately, with some data physically stored and managed in a central location that allows for analytics and population health management. “[Central vs. federated] was a big deal to people a few years ago, but they've moved beyond that. A lot of people realized that everything is hybrid,” says Greg DeBor, who is one of the founders of the New England Healthcare Exchange Network (NEHIN). “Today, the more useful constructs are talking about core and value-added services you want to have in the middle and that determines your degree of centralization.”
Rossignol agrees that the industry has moved “much more towards a centralized hybrid model,” and toward focusing on data analytics and population management activities using de-identified patient data. “I believe the balance is going to accelerate toward a hybrid model that will at least offer you a good equilibrium of the need to ensure privacy and the need to have access to enough data to have analytics,” he says.
Of course, a minority of those developing HIEs still believe that a centralized model offers the best architecture for information exchange. “Inherently flawed in the federated model is the ability to accomplish some very core value-based missions,” says Devore Culver, executive director of HealthInfoNet, Maine's statewide HIE. One of those missions is disease reporting to public health, which can only happen if the data is centralized and standardized, Culver says. HealthInfoNet itself has grown from a five-year-old statewide claims database. Because of the value of that aggregated data set to public health and to the private sector, Maine leaders saw the importance of a clinical data exchange. After a request for information (RFI) with 36 vendors in 2005, Culver says it was evident that federated models couldn't provide the speed requirements needed for queries and reporting.
According to Jason Hess, general manager of clinical research at the Orem, Utah-based KLAS Research, the HIE market isn't mature enough to really assess which technical architecture best serves the needs of HIEs, as most are still struggling to stay viable and gain stakeholder buy-in. “I suspect that you're going to have HIEs running a federated model that, unless the vendor comes up with some unique hybrid angle on that, you're not going to be able to do some of the things you can do when you're aggregating data to some extent,” he says.
FOR BETTER OR WORSE THE WAY HEALTHINFONET WAS DEFINED FROM THE OUTSET, [STANDARDIZATION] WAS A COMMITMENT. I THINK THAT WILL BE FUNDAMENTAL AND CENTRAL TO THAT LARGER VALUE PROPOSITION. -DEVORE CULVER
Mark Allphin, research director at KLAS, adds that some vendors are banking on the centralized repository model to achieve their analytics goals. “A couple of vendors I've spoken with [say] they may support just the aggregated model specifically,” he says. “And for that reason they feel it will be more conducive to plug in an analytical piece and tools right on top of it and have all the data centrally organized.”
“We know it's not just about moving data around; we have to report out of [the HIE],” says Hess. “You have to have some kind of bringing together of that data to be able to do so, whether it's an aggregated model with comingling, or it's a hybrid.”
