If you’re interested in how clinical information is organised, categorised and structured, a great place to start is Larry Weed’s 1971 lecture on problem-oriented medical records.
Larry may have been the first clinician to understand the challenges and to propose solutions. Almost half a century later, we are still grappling with them. Should a clinic letter from Breast Surgery be filed under “Surgery” or “Breast Surgery”? In other words, how should surgical specialties be sub-categorised?
When an anaesthetist encounters difficulty performing an epidural for a mother in labour, how should the details be recorded – what is the difference between an “attempt” and a “pass”? How should the concept of “guidance” be recorded: is loss-of-resistance part of the guidance element?
Questions of this nature are hard to answer – there will be a variety of views expressed by clinicians. When software engineers design, build, test and deploy software for the medical marketplace, they may solicit input from clinicians to help tease out some of the thornier problems. Developers may not understand fully the relationships between different concepts in medicine. This process is ‘clinical curation’, and unfortunately it’s often a case of rinse…repeat.
You might argue that what’s needed is a Clinical Terminology service such as SNOMED-CT. The trouble is, terminology will only get you so far. Sure, you now have access to about a hundred different concepts that relate to “blood pressure” but when you’re building your application, how do you pick the right concept for the context in which the blood pressure is being measured?
This is where clinical models are essential. They capture not only the concept, whether it be coded in SNOMED-CT or some other terminology, but in addition, they show how the concept fits within the clinical context. The leading clinical modelling framework is OpenEHR. In future posts, I hope to demonstrate how OpenEHR models provide a clear path to application development, without the headaches and overhead of thrashing out relationships and semantic meaning in medicine.
The other framework that you may have heard of is HL7-FHIR. There is lively debate in the clinical modelling community of the relative merits of OpenEHR vs HL7-FHIR. I’ve posted a few links below that will give you a flavour of the dialogue between the two communities.
If you’d like to get involved with the clinical modelling community, a good place to start is the OpenEHR discourse discussion forums. There is a good how to get started tutorial and also the introductory video from the 2017 OpenEHR day.
Links for more information:
- FHIR + OpenEHR – Alastair Allen, 2019
- FHIR compared to OpenEHR – Tom Beale, 2017
- Interopen paper on FHIR & OpenEHR – Ian McNicoll, Amir Mehrkar and Tony Shannon, 2019