Stream: social
Topic: Abolishing HL7v2
Brendan Keeler (Mar 21 2022 at 19:53):
https://twitter.com/healthbjk/status/1505994755225776128?s=20&t=TVd7reXB8yP1isevGgTHVw
so close, yet so far https://twitter.com/healthbjk/status/1505994755225776128/photo/1
- Brendan Keeler (@healthbjk) David Pyke (Mar 21 2022 at 19:54):
They're right! HL7 v3 is the future of health care interoperability. We should drop V2 now!
Mario Villace (Mar 22 2022 at 07:52):
That's not feasible because XML has been around for over 45 years...so it needs to be deprecated too!!
John Silva (Mar 22 2022 at 12:19):
Ah, the V3 model-based architecture where you could say anything but nothing conclusively ;-). It seems like the model was so complex to learn/understand that the 'feet in the street' (the s/w developers/integrators) couldn't get traction on it -- hence FHIR to address this.
Cooper Thompson (Mar 22 2022 at 13:21):
I feel like this is one of those "you forget how old you are" things. If HL7v2 is 20 years old, that means it was created in 2002?
Cooper Thompson (Mar 22 2022 at 13:23):
Which, interestingly, is about when JSON was created. So if JSON is around 20 years old, and that is the threshold for when something should be deprecated, then we should get to work deprecating JSON.
Cooper Thompson (Mar 22 2022 at 13:23):
We just need a new, modern, state-of-the-art interop format that has 30+ years of maturity in the industry.
Rik Smithies (Mar 22 2022 at 13:32):
then V2 fits the bill! It is 33 years old (1989).
René Spronk (Mar 22 2022 at 13:33):
Actually, make that 1979 (first direct precursor in healthcare, see http://ringholm.com/docs/the_early_history_of_health_level_7_HL7.htm). DICOM is even older. Stable, still going strong :-)
Rik Smithies (Mar 22 2022 at 13:55):
I took that date directly from your article! But agree the precursors to it were older :-)
John Moehrke (Mar 22 2022 at 14:06):
TCP/IP that we all are using for ALL of this is 40+ years old
John Moehrke (Mar 22 2022 at 14:07):
and TCP/IP has been replaced by three different protocols that were booked as better, only to fail because their benefit was not outweighed by the pain of change.
Yunwei Wang (Mar 22 2022 at 14:26):
Sounds like the same story as V2/V3.
Lloyd McKenzie (Mar 22 2022 at 14:55):
On the other hand, there's significant pain advantage in moving away from v3 :wink:
Joel Schneider (Mar 22 2022 at 14:55):
For me, the HL7v3 licensing was a killer. To read the spec, you had to agree to an NDA.
Cooper Thompson (Mar 22 2022 at 15:38):
We can't ignore switching costs. If HL7v2 or even HL7v3 are working in existing integrations, and maintenance costs are less that switching costs, then sticking with even HL7v3 can make sense. There are some areas of FHIR that are still very immature. We need to get past the Peak of Inflated Expectations before pushing rip-and-replace approaches. FHIR for greenfield interop is often as obvious option. But rip and replace is expensive. Sometimes very expensive.
John Moehrke (Mar 22 2022 at 16:03):
The nice part about FHIR is that it fits well as an API. where the grunt work is done by classic standards, but the API enables new participants to join. -- hence the message from IHE on using FHIR (MHD) -- https://profiles.ihe.net/ITI/HIE-Whitepaper/index.html
John Moehrke (Mar 22 2022 at 16:04):
yet FHIR is not only useful for API... for greenfields one can get fully working system using only FHIR. (also mentioned in that paper)
Kevin Mayfield (Mar 23 2022 at 09:45):
This sounds very optimistic.
Many of the FHIR equivalents of HL7v2 I've seen actually look more like HL7 v3.
The original v2 messages tend to look more like FHIR Resources..
The resource part of FHIR is being ignored to keep v3 alive?
Last updated: Apr 12 2022 at 19:14 UTC
