How Can Health Research Data Infrastructures Be Sustained?
Many people have voiced concerns about the sustainability of state-level health information exchanges bolstered by ARRA stimulus spending. HIEs had to use the federal funding to get established while simultaneously looking for a sustainable revenue model. A similar dilemma faces other forms of multi-stakeholder health IT architecture projects established with grant funding. Many of the projects involving patient-centered outcomes research and comparative effectiveness research will need to find long-term business models to continue.
Recognizing that these projects have to solve the sustainability question, the latest issue of the eGEMS Journal is devoted to sustainability considerations for health research and data infrastructures. (eGEMs, a publication of the EDM Forum, is an open access journal focused on using electronic health data to advance research and quality improvement.)
I had a chance recently to speak with Adam Wilcox, medical informatics director at Intermountain Healthcare, who served as guest editor for the sustainability issue. “One of the positive side-effects of the way the [ARRA] funding was done is that it is allowing us to think about what sustainability really means and how can we address it more directly,” he said. “Our goal was to for the first time get a collection of papers that say here are different ways people have found sustainability working or here are lessons learned.”
Wilcox returned to Intermountain a year ago after working as a professor at Columbia University, where he led an ARRA-funded project called the Washington Heights/Inwood Informatics Infrastructure for Comparative Effectiveness Research (WICER). It involved setting up a data infrastructure to measure the health of a particular New York community over time. After meetings with AHRQ and Academy Health officials to talk about sustainability, Wilcox sought to apply some principles they had discussed to the WICER project. “I started writing a paper about it. In that process, others thought we needed an issue of the eGEMS journal to discuss this more broadly because it keeps coming up that we don’t have good models for sustainability,” he said.
Projects described in the issue include i2B2 (Integrating Biology and the Bedside), the HMO Research Network (HMORN), the Translational Research Informatics and Data Management (TRIAD) Grid, and the DartNet Collaborative.
Wilcox said that a key lesson learned is that one of the biggest long-term operational costs besides technical infrastructure can be maintaining the collaboration itself. He found great disparities in the WICER project between enthusiastic and reluctant partners. “We found if the organizations were willing to collaborate, it was so much more efficient in terms of costs,” he said. One of the WICER partners had great institutional support, but another participant had to convince their organization of the value. Wilcox compared the progress of the two. “It took the enthusiastic partner a third of the time to get through the first three steps of the project. For the other, the issue of getting consensus and agreement took so long that it made the tasks take three times as long. So it was three times as expensive to work with the reluctant partner.”
Another lesson, Wilcox noted, is that that costs can be covered if participants perceive that value is being created. Once infrastructure is developed with grant funding, maintenance costs may shift to the researchers who are using the infrastructure directly. They may be able to contribute to the maintenance once the infrastructure is sufficient to benefit their research projects, he noted.
And while having multiple stakeholders increases the opportunity, it also adds complexity to sustaining an infrastructure, Wilcox said. An example is i2B2, whose initial 10-year grant funding is nearing its end. “They could partner with industry, but they felt it would be difficult to maintain governance if they did that. It would make it easier to sustain, for sure. But they thought governance and management of relationship of different participants was critical and they didn’t want to involve another stakeholder because it could misdirect where they wanted to go,” Wilcox said. Instead, i2B2 is looking to establish a foundation, he said. “So there is an example where adding a stakeholder would have increased the opportunity for sustainability, but also the complexity level.”
He added that i2B2 is also an example of the power of crowd-sourcing. They disseminated a tool and created a network of users and enhanced the system. The organization itself was able to sponsor two plug-ins, while the community contributed 30, Wilcox said. “The crowd-sourcing model led to a lot more expansion than they were able to do by paying for it. That is a really interesting lesson. If you can reach the tipping point, crowd-sourcing can decrease your enhancement costs.”
A final common lesson he identified is that stakeholders may support an infrastructure to keep flexibility in an emerging area. Wilcox cited as an example New York Presbyterian Hospital’s participation in the WICER project.
“I don’t believe their participation was because it was currently achieving value for them,” he said. “I think they saw it could achieve value later. They were hedging their bets on big data. A lot of people are talking about this emerging trend and they don't want to be caught flat-footed. They decided it would be more cost-effective to keep this project going than to try to play catch-up later.”
All of the papers in the eGEMS issue offer interesting insights into long-term business models and commercialization of health research infrastructure.
