Using SNOMED and HL7 Together

Abstract

One of the lessons of the past few years is that we cannot separate the issues of information structure from those of terminology. The idea that syntax and semantics are independent has been shown to be an oversimplification. We cannot slot any terminology into any data structure and expect it to work. Perhaps, it is surprising that for so many years, so many people thought that we could. After all, in every spoken language, such as English, the grammar (syntax) and words (terminology) have evolved together.

For reasons, which seemed good at the time, the standards development organization responsible for healthcare terminologies (such as SNOMED CT), message syntaxes (such as the HL7 V2 and V3) and information models, developed independently and in parallel. The main justification was the perceived need to interoperate with a wide range of legacy schemes and to meet the needs of different national requirements and languages. SNOMED CT was designed to be syntax-neutral, so it could work with any syntax. Reference models such as the HL7 RIM and EN 13606 set out to be terminology-neutral.