Rethinking the Multi-Institution Clinical Data Repository | David Raths | Healthcare Blogs Skip to content Skip to navigation

Rethinking the Multi-Institution Clinical Data Repository

February 22, 2018
| Reprints
Exploring the changes under way at the clinical data repository for South Carolina hospitals
Last month I wrote a short news item about how Health Sciences South Carolina (HSSC) won Best in Show at the FHIR Connectathon in New Orleans for its clinical data repository for South Carolina hospitals. Last week I had the chance to interview HSSC executives in more detail about this effort. The evolution in approach at HSSC may signal a sea change in how large clinical data repositories work. 
Established in 2004, HSSC was the first statewide health data and research collaborative in the United States. Today, HSSC enables multi-institutional health research through its clinical data warehouse and associated governance and research tools. It has data on 4.5 million patients. Recently, HSSC has been working with a company called Simpatico to deploy a FHIR-based clinical data repository called the Smile CDR that spans several facilities.
Les Lenert, M.D., the SmartState Endowed Chair in Biomedical Informatics at the Medical University of South Carolina, said, “We got interested in the FHIR clinical data repository architecture as a way to convert from having a large centralized data source to a federated model that was closer to our clients. That would meet our needs for a new business model we saw. Previously we had an approach of ‘if you build it, they will come’ with a large central database where people could perform operations. What we thought was that being separated from our customers like that was not a good idea.”
Under the new architecture, the idea is to set up a bunch of site-specific clinical data repositories. Each hospital in the network has its own CDR with a bunch of clinical data for that institution. Each facility is able to have a dedicated repository of clinical data populated in real time based on feeds from EHR systems. Hospital data is normalized into FHIR Patient, Encounter, Condition, Observation, and other resources. This data infrastructure is then combined with an enterprise master patient index (EMPI) to provide standardized research data reporting, and to enable SMART on FHIR-based apps with a longitudinal view of data across institutions.
Health systems have started developing their own clinical data repositories and have the ability to do much more management of their own information, so the need for an organization to provide warehouse services to them is not as great anymore, Lenert said. 
“We are finding new ways to envision ourselves and what our data services and products are,” Lenert said. The Smile CDR helps us do this by not only creating a local repository, but having this idea of enrichment of that repository through subscriptions, and yet the same subscription mechanisms allow us to introduce things like a master patient identifier across systems so we don't have trouble merging data when we need to for other operations.” 
The institutions control their own data and are able to add proprietary data and information when necessary for business operations, added Kenneth Deans, HSSC’s CEO.
“We wanted a platform that would not only allow us to transform, normalize and standardize the information,” Lenert said, “but also one where we could enrich it from outside sources through subscription. That might include our research EMPI but it might also include death data from the state death index or environmental sensors or data available through geocoding. It is really about this notion of transformation of our members’ information sources, and making them more accessible, but also enriched at the same time.”
Previously HSSC had real-time HL7 feeds from most of its sites. The initial vision for this was a parallel EHR that was a few minutes behind. “What was hard was to get to work flow integration,” Lenert said. “With the Smile CDR through its FHIR interface, that is what we are excited about. We are able to build things that interface with the workflow inside EHRs or other population health systems of our members.” 
Deans spoke about creating “communities of practice” around types of data. “We believe the sky might be the limit,” he said. “That is part of the value of moving to the FHIR methodology because we can ingest public as well as institutional data in a much more meaningful way than we have done in the past. The phrase we use is ‘technology-enabled communities of practice.’ If you bring together public health data and institutional medical record data, it allows you to get to very specific communities of practice, and the options of those practices are limitless. They can focus on those that are most meaningful to them — problems such as antibiotic resistance, obesity, or the opioid crisis.” 
The new architecture should help improve research efforts, Lenert said. For instance, a research pilot in the Carolinas Collaborative spans the EHRs in emergency departments at MUSC, Wake Forest and the University of North Carolina. “They each have to modify their EHRs separately and you can’t write once and run many places. With our new network,” he explained, “when we do something across members of HSSC, it will be write once, run everywhere because they can all hit the same Smile CDR for their FHIR applications. They can collaborate on building a single FHIR app or deploying one FHIR app that has already been built and work on the business processes, evaluation and outcomes together.”