Report from the FHIR Connectathon at the HL7: September 2017 Working Group Meeting

October 27th, 2017 / By Senthil Nachimuthu, MD, PhD

About a year ago, I wrote a blog about the convergence of two HL7 standards: Clinical Information Modeling Initiative (CIMI) and Fast Healthcare Interoperability Resources (FHIR). I attended the HL7 Working Group Meeting in San Diego last month where I followed the latest developments on these standards, and I wanted to post an update about the current activities among FHIR implementations. I hope this snapshot helps you understand the level of adoption of FHIR by various organizations, and helps you as a FHIR developer, implementer or end-user who benefits from the implementations.

Interoperability between disparate systems were demonstrated for a variety of use cases in the FHIR Connectathon. These include an Opioid clinical decision support system being connected to different EMRs in the CDS Hooks track, pulling data from EMRs for a breast cancer reasoning system in the Clinical Reasoning track, using Blue Button on FHIR/SMART on FHIR in the Consumer Centric Data Exchange track, as well as Devices on FHIR track focusing on point-of-care and personal health devices.

In the Genomics track, several organizations are working on supporting sequencing data on FHIR, including a Sequence resource and profiles for genetic orders, results and observations. New for this Connectathon, there was a ‘LastN Query’ track led by Cerner, Mayo Clinic and others, which returns the last N resources of a specific type, e.g. lab results, vital signs, etc., which turns out to be a frequently requested feature from hospitals.

Patient Reported Observations track presented their work on capturing sleep data reported by consumer sleep tracking devices. As a side note, I often wonder about the accuracy of these devices, since personal health devices are not periodically checked and calibrated as often as devices used in hospitals, and they often use proxy variables for measurements that rely on special sensors – e.g., accelerometers in mobile devices instead of EEG sensors used in polysomnography. The well-controlled studies that evaluate consumer devices against a gold standard are rather illuminating compared to YouTube/blog reviews that compare several personal health devices against each other.

The Patient track did a break-out session where about ten systems used Touchstone to run formal tests, and four were reported to pass the tests. It was encouraging to see that some organizations were working on provider directories and scheduling through FHIR as well.

Attachments track, understandably with participation from Lantana who did pioneering work on Clinical Document Architecture (CDA), was working on a document bundle that contains a composition and CDA on FHIR resources. I hope that this project provides a way for CDA implementations to migrate to FHIR.

AllScripts, VA and others are working on the Care Plan track, where they were creating PlanDefinition and ActivityDefinition to simulate the care plans of a handful of VA patients. There was also a Clinical Research track that works on using FHIR for clinical trials and modeling the interaction between patients, institutions and studies.

Not to be left behind by their clinical counterparts, the Financial Management track was working on electronic claims as well as medication coverage management using FHIR.

Last but not least, my reason for participation was the Terminology Services track. Several organizations were testing their terminology service implementations with testing tools such as Touchstone. In the future, this track plans to do formal tests of terminology services as well as interoperate with other tracks so that they can run their applications against different terminology servers that implement the FHIR terminology module. This will be a real world test bed for terminology services implementations, and I hope to provide updates as we do some testing this way.

I noticed that many tracks were testing their implementations using Crucible (MITRE) and Touchstone (Aegis) test suites. Though this is a necessary first step, the Connectathons will simulate the real world when different groups try to connect to each other and test operational interoperability. I’m confident that it is only a matter of time, possibly within the next two to three years. By then, I hope that the CIMI FHIR Profile as well as the U.S. FHIR Profile become DSTU/STU (draft standard/standard for trial use), if not normative, so that we can demonstrate seamless real world semantic interoperability using FHIR, CIMI and standard terminologies.

Senthil K. Nachimuthu, MD, PhD,  medical informaticist with 3M Health Information Systems’ Healthcare Data Dictionary (HDD) team.