A Social Network for Health Informatics Professionals and Students
Archetypes are the business objects of healthcare. They are central to the standardization of health care concepts and are formalised in the European standard CEN/ISO EN13606 , commonly known as EHRcom. Damon Berry of Dublin Institute of Technology (DIT) (the TeaPOT research group) organised a two day training course and symposium on archetypes. Here is a summary of the symposium.
Gerard Freriks, past chairman of CEN/TC251 WG1 and currently Services Manager at EuroRec and owner of Electronic Record Service BV (ERS) made a presentation on CEN/ISO 13606. He described the dual model approach of the standard; the Reference Model (En 13606 part 1) that contains the basic entities for representing information, and the Archetype Model (En 13606 Part 2) which is used to semantically describe clinical concepts using the entities of the Reference Model. His linked the Reference Model to the domain of traditional IT engineering and the Archetype Model to the domain of Health Information and Knowledge modelling.
The advantage of EHRCom is that it gives the power to the clinicians to define whatever concepts they need through the mechanism of archetypes. Gerard explained how clinicians can use archetypes to agree upon concept definitions and build a common Information Model. This model will achieve the common trust which is needed for semantic interoperability between systems. He gave as an example the concept of blood pressure and showed how an archetype can be constructed to represent a formal model of this concept. He stressed that the archetype should represent the maximal data set for the concept while a template is a subset of this maximal data set for use by particular specialists.
Gerard explained the differences between the messaging model HL7 and the two-level model EN13606, arguing that HL7 tells you nothing about the system from which it came, while in the two-level paradigm the common model is shared by all communicating systems. He also argued that the development time for agreed-upon HL7 messages can take years while archetypes can be agreed and developed in a matter of days, citing the epSOS projects (see part 2) as example of this. He also argued that HL7 messages are too static and do not leave any room for innovation in healthcare, again a weakness not present in archetype-based systems.
Gerard described a product developed by ERS called IC-EHR. This is an Enterprise Information Bus built around the EHRcom standard. He explained that systems that are based on archetypes are very flexible and can be added easily to this Enterprise Information Bus.
He spoke about the uptake of EN13606 in the UK (where it is being used within the NHSs Logical Record Architecture) Sweden, Germany, Japan, the Netherlands and other countries.
He concluded with some lessons learned from the pilot implementations
· It is imperative to have agreed standard modelling patterns to achieve Semantic Interoperability.
· One needs the complete machinery that 13606 provides
· Open standards are essential, proprietary specifications based on standards are not an alternative.
· To create semantic interoperability requires trust and the ability to share archetypes so archetypes need to be in the public domain – we need archetype libraries , both nationally and internationally to achieve the vision whereby all healthcare systems share archetypes and coding systems
· Archetypes need their own type of Standards Development Organization
· Archetypes need Quality Assurance at national and International level. EuroRec will play a part at the EU level (there is already a report available on Archetype Quality Assurance)
· There is a need for unique identifiers at different levels for these systems to work.
· Archetypes will be of great benefit to SNOMED-CT and the other terminologies. The codes can be used within Archetypes to remove ambiguity.
· All stakeholders must play a part; from legislators to insurers to IT vendors to national projects and health systems – the need for national projects was highlighted as a key enabler
· We need an “Infostructure” (archetype STOs, archetype QA, archetype libraries etc) and an “infrastructure” (unique Ids etc) to achieve semantic interoperability
· The "infostructure" and the "infrastructure" must be organized – it needs a director etc.
A general discussion raised some further points:
· The general consensus was that archetypes should be freely available and in the public domain. Archetypes are all about clinical best practice and so they belong to the community (though those who develop them should possibly be re-numerated in some way). The delegates agreed that knowledge should be freely available. The example of licensing costs for the use of some coding systems was cited as a barrier to the free availability of knowledge.
· EHR Extracts (based on EN13606) will open up the business of IT in healthcare
In part two of the meeting David Moner from the Biomedical Informatics Group, The Technical University of Valentia, described four pilot implementations of the EHRcom standard and he also described the LinkEHR tool (which was covered in day one of the meeting). This is an Graphical tool that allows the user to create archetypes based on the EHRcom standard.. Damon Berry gave his ten ingredients for Semantic Interoperability and made some proposals for the extension of EHRcom to include unique identifiers. Sheng Yu (DIT) described his work with termininology shadows. Xu Chen (DIT) spoke about a generalised identity model for e-health. Pamela Henry from DCU school of Nursing described the work being done in the EHRland and PARTNERS projects and Tony Kenny gave his observations of the EHR project in Ireland. I had to leave before the final presentations of the day to get through the Dublin traffic. Thanks to all concerned for an excellent symposium.