Semantic Interoperability: Share the Work, Share the Benefits

A September 2016 EU-sponsored report, ASSESS CT recommended SNOMED CT for EU-wide deployment to enable semantic interoperability of clinical information in Europe. 
 
This is great progress! The EU acknowledges that semantic interoperability is desirable, and that currently we don’t have it, and that terminologies, including SNOMED CT, are key to achieving it. 
 
The report also pointed out that SNOMED CT is part of a broader “terminology ecosystem” that includes WHO classifications, and local language interface terminologies.
 
But here’s the challenge: The terminology ecosystem really doesn’t exist yet. The report acknowledged  “terminology artifacts” would be needed, including standardized clinical information models that would be bound to SNOMED terms, maps from SNOMED to other terminologies in the ecosystem, and the development of quality assurance mechanisms to be built into a “terminology management cycle”.
 
We know from experience that terminology management is a low priority for software vendors. When I met with informatics teams last year, I learned that they were already living the terminology ecosystem, and that managing it was not easily repeatable, or scalable. This is partly because it has many moving parts: Many code systems with differing use cases and subtle incompatibilities, differing publishing cycles, competing standards, and the list goes on. Most importantly, was that the lack of software tools prevented enterprise wide governance of structured clinical content used in EMR applications. This, above all, was why terminology ecosystem management could not scale to the enterprise.
 
Recommending SNOMED CT as the “heart” of the ecosystem, the hub from which other terminologies extend, is a great step forward for semantic interoperability. Yet, for the terminology ecosystem to have the desired effect it must first exist, and must itself be semantically interoperable. The principles on which terminology artifacts are created and managed must be universally accepted and applied consistently. If clinical terminologies are going to be part of the solution, the people who make clinical terminology work - individual informatics teams - cannot work in silos. They must collaborate, across businesses, regions, borders and cultures. 
 
But will they collaborate? It is true that commercial software vendors compete with one another. Conversely, they also want to develop standards based products. The standards, terminologies, and other basic building blocks of semantic interoperability are not strategic. They offer no competitive advantage. They are required components of much larger systems. Standardization and consistently high quality are shared concerns for vendors.
 
And importantly, we are all united in the common purpose of improving the way healthcare works everywhere. Real semantic interoperability will revolutionise healthcare for the better. But it is a huge challenge. One informatics lead I met with called it a “holy grail”. To find it, we must change the way we work together. To do that we must change the way we think. If we share the work, we will share the benefits.
 
Next time: None of this will happen without technology support. There will be many opinions on this topic. I will share mine. If you made it this far, thanks for reading!