Stream: bulk data
Topic: DocumentReference
Doug Pratt (Aug 25 2020 at 13:19):
The DocumentReference resource enables a FHIR server to return either links to documents or the documents themselves as embedded binary; the server makes the decision as to which. Are there any expectations with bulk FHIR? On one hand, the bulk paradigm would seem to favor sending the document contents in bulk as embedded binary. On the other hand, this could make the response data sets enormous. Any thoughts on what the current spec implies, and if we should plan an enhancement to enable the client to specify?
John Moehrke (Aug 25 2020 at 13:26):
there are IG for specific use-cases that declare constraints. Most that I know mandate use of Binary and not embedded base64 encoded. So, are you expecting bulk data api to override these?
Josh Mandel (Aug 25 2020 at 13:34):
This is one of the areas we took up as part of The Argonaut 2020 project on bulk data this year. We will be testing this among other improvements at the connectathon in September.
Josh Mandel (Aug 25 2020 at 13:34):
The specific changes we put in place with respect to attachment content (which includes document reference attachments) are visible at https://github.com/HL7/bulk-data/pull/64/files
Josh Mandel (Aug 25 2020 at 13:36):
To see them rendered in context you can browse the continuous integration build at http://build.fhir.org/ig/HL7/bulk-data/export/index.html
Doug Pratt (Aug 25 2020 at 13:37):
Thank you, I will review those links.
Josh Mandel (Aug 25 2020 at 13:37):
(search that page for the word "attachment" and the word "binary" to see the relevant guidance.)
Michael Turman (Feb 20 2021 at 17:44):
Hey – Michael from Cerner here, our team is working on capacity planning related to Bulk Data requests for DocumentReference and Binary resources. We wanted to solicit feedback/input from the client community on use cases that inform the latency, frequency, and volume of document content requests. Likely to be returned in the form of Binary resources.
For example, request patterns for Binary document content or use cases aligned with retrieving full document content in large volume.
Michael Turman (Feb 24 2021 at 15:00):
Bumping this thread.
Looking to understand when a 'Bulk Data Consumer' would want to return document content (binary) referenced within a DocRef resource. Do consumers expect to retrieve all document content on the heels of receiving DocRefs? Or to put it another way, what request patterns/use cases are being designed with respect to Bulk document data?
Last updated: Apr 12 2022 at 19:14 UTC
