In 2006 there were only seven or eight clinical specialty registries in the United States. As part of HITECH Act, specialists got credit for submitting patients to a specialty registry. In 2016, 10 years later, there were 165 specialty registries.
Years ago, registry data was done manually by chart abstraction. Gradually that evolved to using electronic chart abstraction forms, but it is not true electronic data capture. Russ Leftwich, M.D., senior clinical advisor for interoperability at InterSystems and co-chair of its the Learning Health Systems Work Group at HL7, says the only feasible way to enable future registries is by using data captured in EHRs and other systems in a reproducible way. “That suggests using FHIR,” he said.
Health systems might have 165 registries coming to them wanting their data, and they can’t give it to them in 165 different ways, he said. “There needs to be a standardized way to get those same data elements — and for the most part it is the same data elements — that different registries want. The specialty part comes down to a small percentage of the total data that systems hold.”
From a technical standpoint, he said, it needs to be provided according to a specific data model or data definition. You can’t have different ways of asking for the same data. “FHIR is the ideal technology to provide that data in a standard format, using the same FHIR profile or application programming interface to provide that data.”
The Healthcare Service Platform Consortium and HL7’s Clinical Information Modeling Initiative are working together to create information models that are then expressed as FHIR profiles so registries could ask for that data in the same format.
Informaticists are working to create a repository of FHIR profiles that can be interoperable. “Interoperability doesn’t depend on FHIR, per se, but on using the same FHIR profile for the same data,” Leftwich explained. “If people hadn’t seen a zebra before, and one called it a white horse with black stripes and the other called a black horse with white stripes, any human would understand they are talking about the same thing, but computers can’t do that. Computers have to have a very specific definition of that data concept to exchange it with another computer.”
Another thing that makes FHIR the solution to the registry of the future, Leftwich added, is that subscribing to data is easier with FHIR. There is a mechanism in the FHIR specification that allows you to subscribe to data the same way we subscribe to someone’s Tweets or Facebook posts. “So if registries have entered an individual as a patient in the registry, they can subscribe to the data from that individual that will be sent to them as FHIR resources as it becomes available.”
EHR vendors are enthusiastic about this approach. “They were one of the first to recognize what the F in FHIR stood for,” Leftwich said. “It stands for fast. When they realized they could create interfaces to their data so much faster with FHIR, they were all over it from the beginning.”
“When the vendors realized that a new person with a computer science degree could in a couple of days start to understand FHIR and build interfaces in a few days that previously took six weeks, it was obvious to them what the value of FHIR was going to be,” he said.
Likewise, registries need a new way to get data or it will become increasingly burdensome to invest the time to feed the data into them.
“Data in healthcare is growing exponentially, and yet we are using the technology and standards invented 30 years ago to move that data around,” Leftwich said. “It is amazing we have gotten as far as we have.”