Modifying .reference elements in Patient record output

[ToInf1n1tyAndBeyond]

We are attempting to upload a large number of Synthea patient data into a FHIR server; however, we have discovered that all reference url's in the output are invalid and as a result are rejected by the FHIR profile. Solution is to either remove all .reference elements from the .json files or somehow have the program generate valid reference url's. Worst-case scenario would be to modify the core code to suppress the .reference elements. Can someone advise if...

1. There is a simpler way to suppress the .reference elements in the .json output, or
2. There is some way to have the code generate valid reference url's

[jawalonoski]

We regularly test this output with the HAPI server, so it would be useful if you can go into the specifics of your situation.

Synthea has a number of ways you can alter the FHIR output, specifically (abridged synthea.properties file):

# exporters that use XML or JSON can enable or disable 'pretty printing'
exporter.pretty_print = true
exporter.fhir.transaction_bundle = true
# using bulk_data=true will ignore exporter.pretty_print
exporter.fhir.bulk_data = false

• When exporter.fhir.bulk_data is true then most references appear as {ResourceType}/{ResourceID}.
  • In this case, each Patient and their associated Resources are contained in a Bundle
  • The exporter.fhir.transaction_bundle setting controls whether the Bundle is a collection or transaction.
    • If it is a transaction then the references of resources shared across Patient records (i.e., Organization, Location, Practitioner) are constructed as queries, e.g., Organization?identifier=https://github.com/synthetichealth/synthea|{ResourceID}. In this case, you need to load the hospitalInformation*.json and practitionerInformation*.json files before any of the patient records.
    • If your system cannot handle loading the shared resources in advance, and querying for the resource in the transaction, then use collection -- just note that shared resources will be duplicated.
• When exporter.fhir.bulk_data is false then references appear as urn:uuid:{ResourceID}
  • In this case, each Patient and their associated Resources are written into separate ndjson files.
  • Shared resources should be loaded in advance... for example, loading Organization before Location before Practitioner etc.

[ToInf1n1tyAndBeyond]

Attached is a .json with generated patient and encounter data. When trying to upload to my Smile FHIR instance all references generate errors...

• "HAPI-1091: Invalid match URL "Practitioner?identifier=http://hl7.org/fhir/sid/us-npi|9999869198\" - No resources match this search"
• "HAPI-1091: Invalid match URL "Organization?identifier=https://github.com/synthetichealth/synthea|17727201-cc60-3852-b377-9ae26b087fe0\" - No resources match this search"
• "HAPI-1091: Invalid match URL "Practitioner?identifier=http://hl7.org/fhir/sid/us-npi|9999998393\" - No resources match this search"
• "HAPI-1091: Invalid match URL "Organization?identifier=https://github.com/synthetichealth/synthea|c2b4bc00-156e-3478-a63f-4561dbab7e26\" - No resources match this search"

When .reference elements are removed the remaining data is uploaded successfully.

Sharon747_Fahey393_ca617c5f-828d-9a38-7e5c-41cb824c5ea8.txt

[jawalonoski]

Did you load the hospitalInformation*.json and practitionerInformation*.json files first?

Those two files contain the shared dependencies (Organization, Location, Practitioner) that I described above.

[ToInf1n1tyAndBeyond]

I didn't! Thank you.