I have been studying the USAM (unified service action model) - the clinical part of the HL7 v3 RIM. While (naturally!) I don't necessarily agree with everything I have read, it is a pleasure to read, due to its clarity. People here might be interested in a few early conclusions:
- HL7 is more process-oriented than GEHR. This is not surprising, given its hospital/pathology origins, and the general strength of health IT in the US in hospitals rather than primary care. In Europe, Australia/NZ it is the reverse, and consequently GEHR was originally conceived for primary care and hospital use, but was more knowledge oriented (GPs care less about micro-process than do hospitals).
- This philosophical and historical difference does not mean that GEHR and HL7 are incompatible; on the contrary, there are so many areas of compatibility, I was surprised myself. Another way of stating the different philosophy is that GEHR makes the EHR explicit, whereas HL7 models the whole process of health care, and (I assume) would see the "EHR" as a kind of constructed view of an HL7 system.
- A challenge for HL7 is to make definitional and result information more clearly available (i.e. filter out some of the more detailed process stuff) for clinicians to use; a challenge for GEHR is to include more process (currently underway - care plans and pathways are being modelled).
- USAM uses "mood" to differentiate between the meanings of different kinds of services - definition, intent, order, event (= satisfied order), goal (a predicate mood). The mood variable is central to HL7 information being able to properly carry the important semantics of natural language. In some ways mood simplifies the model (reduces number of classes etc) but I think a bit more work will be required to make the whole model comprehensible, since mood interacts with other variables like "status", and meanings have to be assigned to every mood X status combination. But.... it is clear that the purpose of mood is important. In GEHR's knowledge based approach, there are a surprising number of equivalences (following in the form of hl7 = gehr):
- definition mood = PROPOSITION_CONTENT
- intent mood = PRESCRIPTIVE_CONTENT, status = defined
- order mood = PRESCRIPTIVE_CONTENT, status = ordered
- event mood = PHENOMENON_CONTENT
- Service status describes a state model (new, held, cancelled, active...... completed), which corresponds to GEHR's PRESCRIPTIVE_CONTENT.status, and also a status variable in a care plan.
- Many of the service who/what/when/where attributes have a direct correspondence to the GEHR context attributes (ANY_CONTEXT, PHENOMENON_CONTEXT etc)
- A crucial aspect of GEHR is versioned transactions, enabling prior states of the record to be reconstructed. USAM mentions "snapshots" of the state of business objects at each stage in their cycle, for foresnic auditing - the same purpose as GEHR. This is encouraging, as it means that there is likely to be a direct equivalence for this once HL7 details more how it will be implemented.
- Almost all data types have equivalences in GEHR and HL7, and I suspect mapping these at least will not be too traumatic.
I am hoping my visit to Cleveland in 2 weeks will help the effort for true interoperability along, by finding a path for harmonisation.
- thomas beale
