Stream: implementers
Topic: HL7 Messages vs FHIR
Lauren Wolejsza (Feb 15 2017 at 16:40):
Hello! Can anyone give me some insight to how/if FHIR can accommodate HL7 messages? We are unfamiliar with HL7 messages and learning quickly about FHIR. Reason I ask is because we have existing telehealth databases and remote monitoring health devices that are generating data and we want to change the end point of the HL7 messages to utilize a FHIR-compliant MongoDB database. We are trying to get an idea for if FHIR can easily digest HL7 messages. Thanks in advance for your thoughts!
Igor Sirkovich (Feb 15 2017 at 17:02):
Hi Lauren, I believe by "HL7" you mean "HL7 v2" messages as HL7 is a set of standards, e.g. HL7 v3, CDA, FHIR, etc. If so, there is no magic bullet - you would either have to build an adapter to map HL7 v2 to FHIR or to use a commercial Interface Engine that supports such mapping. You can see an example of suggested mappings at http://build.fhir.org/observation-mappings.html, but every HL7 v2 spec is "unique" (i.e. not really standard) and you will need to know well each data element of the current HL7 v2 messages to figure out the best mapping to FHIR (including terminology mapping).
Lauren Wolejsza (Feb 15 2017 at 17:10):
@Igor Sirkovich Yes, I neglected to specify which HL7 standard as I do not know exactly which HL7 messages, v2 or v3, are being utilized today by the telehealth applications and device data flows. Also, thanks for clarifying the effort for mapping between standards! I see from the example link the complexity here from HL7 v2. Our client seems to think mapping the existing messages to FHIR will be an easy task and I want to make sure I can tactfully explain why this is not the case.
Lloyd McKenzie (Feb 15 2017 at 18:45):
The big challenge with mapping is it tends to be interface-specific. Once you're narrowed down to a single interface, if you've got an integration engine and some decent documentation, it generally isn't too hard
Lloyd McKenzie (Feb 15 2017 at 18:46):
(That said, mapping effort is always about degree of alignment, so estimating based on "hard" and then occasionally being pleasantly surprised isn't a bad strategy :))
Lauren Wolejsza (Feb 15 2017 at 19:58):
@Lloyd McKenzie Completely agree. We are mapping the UI fields to FHIR resource fields for the first of possibly 4+ applications for the first time now and fully expect to reuse this mapping for the next applications. Expect similar efforts to occur with the remote medical device data that currently utilizes messaging.
Last updated: Apr 12 2022 at 19:14 UTC
