SNOMED CT is a modern, comprehensive medical terminology. For a detailed understanding of SNOMED CT, please see Chapter 4 in my clinical design document.

A key issue with SNOMED CT is an effective implementation. A good implementation, in my opinion, results in a system in which users do not actually know that they are using a terminology system at all. It should be intuitive and most importantly, usable from a clinical point-of-view.

I have been using SNOMED CT since 2009. I think my electronic health record system (PatientCare) was the first deployed system in the UK to use SNOMED CT. SNOMED CT is the lingua franca within the system, internally the application and its business logic layers make ubiquitous use of SNOMED CT to make inferences about the care of a patient and to update the user interface to enable or disable functionality depending on the data entered so far.

In my opinion, while one can bolt-on SNOMED CT as a terminology as an after-thought, it is preferable to design systems from the ground up to use and understand SNOMED CT.

However, this doesn’t mean every system needs to re-implement SNOMED CT from the ground up. Indeed, I’d suggest using my new SNOMED CT terminology microservice. This provides very fast browse and search function as well as subsumption so that one can ask whether, given a diagnosis of progressive bulbar palsy, does a patient have a type of motor neurone disease, and get “yes” as the answer.

It also provides an electronic prescribing function for real-time natural language parsing of prescriptions. There is more work to do, and I continue to work on new developments including round-tripping from VTMs to VMPs and AMPs with automatic dose calculations and support for automatic natural language post-coordination. Post-coordination is, in my opinion, in scope of the information model in which SNOMED CT terms are used, but automatic NLP-based post-coordination will make entry more intuitive for the user.