Terminologies: Critical, but often deferred in FHIR application development

Feb. 8, 2019 / By Tosh Kartchner, RN

The Health Level 7 (HL7) Fast Healthcare Interoperability Resources (FHIR) standard is getting widespread support and gathering momentum in adoption. In compliance with this standard, we’ve developed FHIR-compliant Application Programming Interfaces (APIs) for our terminology server, the 3M Healthcare Data Dictionary (HDD), and our team actively participates in FHIR initiatives and discussions. We’ve attended HL7 meetings and FHIR connectathons, and you can view our thoughts in our previous blogs.

We’re now participating in a new initiative, the HL7 FHIR Applications Roundtable. At this event, presenters are given 15 minutes to demonstrate the FHIR-compliant tools and services they provide to the market. These products range from Personal Health Records (PHR) and Electronic Health Records (EHRs) to Health Information Exchanges (HIEs), and there are also applications that implement FHIR services to support these platforms.

Given that terminologies are a foundational piece of implementing many of the FHIR specifications, I was surprised to see that only a handful of these demos mentioned standard terminologies. With the technical difficulties that inherently occur in live demos, that 15-minute time slot became very short to touch on every detail, even such an essential requirement as standard terminology implementation and support. Fortunately, I was able to track down presenters to dig in deeper and I found various situations at play. Some products did not have a need for terminologies, while others said they were in the process of evaluating a terminology server. The most interesting were those who had built a viable product without a terminology server chosen and in place. I was curious as to why terminology implementation was not included in their early planning—didn’t they know the importance of terminologies and standardization of data, or did they simply not care?

After conversations, I concluded that the answer was, “Yes, they do understand that it is necessary, and if not done the right way, they can end up with worse outcomes than not doing it at all.” However, due to various business or technical reasons, they decided to “act first and worry later.” There were also some who felt their terminology needs were small and that a robust terminology server was not necessary. I wondered if this strategy was shortsighted, but without a deep dive into the specific products, it was hard to determine the answer to that question. One thing I’ve learned is that “standard” does not equal “simple”—medical terminologies that reflect the art of medicine are complex, frequently confusing, and take dedicated effort and experience to master.

It was clear to me that, while we’re all dedicated to making medical data useful for clinicians and patients, we’re each working on one piece of the puzzle. As the HDD team remains heavily involved in informatics (specializing in terminology maintenance and mapping), how can we help the industry start the implementation of data standardization earlier in the process of developing tools so integration can be as seamless as possible? We’re already sharing our knowledge and experience by attending HL7 and other standards and informatics conferences; perhaps we need to engage more in “outreach” efforts or explore how to move terminology implementation towards a more “plug and play” approach—in line with the FHIR vision.

Tosh Kartchner is a registered nurse working in medical informatics at 3M Health Information Systems.