Introduction to FHIR Bulk Data

National Institutes of Health (NIH), Office of the Director (OD), Office of Data Science Strategy (ODSS)

June 13, 2023

FHIR® is the registered trademark of Health Level Seven International (HL7). Use of the FHIR trademark does not constitute an HL7 endorsement of this workshop.

What is FHIR Bulk Data?

HL7 specification, published in November 2021
https://hl7.org/fhir/uv/bulkdata/

From the spec:

Providers and organizations accountable for managing the health of populations often need to efficiently access large volumes of information on a group of individuals. For example, a health system may want to periodically retrieve updated clinical data from an EHR to a research database…

Why do we need FHIR Bulk Data?

From the spec:

Existing FHIR APIs work well for accessing small amounts of data, but large exports can require hundreds of thousands of requests.

FHIR Bulk Data provides a simpler, standard mechanism to extract large amounts of FHIR-format data from an EHR.

Why now?

The Office of the National Coordinator for Health Information Technology (ONC)’s Health IT Certification Program requires adoption of FHIR Bulk Data APIs:

“ONC Supports Adoption and Implementation of Bulk Data APIs”
§170.315(g)(10) Standardized API for patient and population services
FHIR Bulk Data group-export required for December 2022 certification deadline for USCDI data elements
- US Core FHIR Implementation Guide

Aside: EHI export

ONC’s program also requires population-level data export (“EHI export”):

§170.315(b)(10) Electronic Health Information export

Patient population electronic health information export. Create an export of all the electronic health information that can be stored at the time of certification by the product, of which the Health IT Module is a part.
§170.315(b)(10) fact sheet (pdf)

must be available to their customers by December 31, 2023
Can be implemented with FHIR Bulk Data, but does not necessarily have to be FHIR Bulk Data
- FHIR(ish) example: athenaPractice (née GE Centricity) https://docs.mydata.athenahealth.com/fhir-r4/ehiexport.html, which uses NDJSON, FHIR resources, and some custom resources defined in this Implementation Guide.

Vendor support

Widespread vendor support for FHIR Bulk Data export of groups of patients thanks to December 2022 ONC certification deadline
List of vendors with §170.315(g)(10) certification: https://chpl.healthit.gov
- Filter for Certification Criteria → 2015 Cures Update Certification Criteria → 170.315 (G)(10): Standardized Api for Patient and Population Services (Cures Update)
Examples: Epic, Cerner

FHIR Bulk Data specifications

Implementations use these specs:

https://hl7.org/fhir/uv/bulkdata/export.html defines the request flow for retrieving Bulk Data from a FHIR server
https://www.hl7.org/fhir/smart-app-launch/backend-services.html further describes the authorization/authentication flow

Types of Bulk Data requests

Patient, to obtain resources related to all Patients
Group, to obtain resources associated with a particular Group
System, to obtain all resources, whether or not they are associated with a patient

High level FHIR Bulk Data request steps

Get an access token from the FHIR server
Request: GET https://fhir.example.org/Group/[id]/$export
Receive the requested FHIR data (“NDJSON” format)
- Can be split into multiple files by the server to support arbitrarily large data sets

Characteristics of research that may benefit from FHIR Bulk Data

Need for direct access to (identified) EHR data for a specific group of patients
- Note that there is an open source library from Microsoft for de-identifying FHIR Bulk Data
The necessary data elements are part of USCDI, or otherwise included in EHR’s FHIR implementation
Real-time access is not needed (other FHIR APIs are better for this)
One-off data pull or regular updates
Multi-site research without existing shared CDM

Non-EHR uses of FHIR Bulk Data

From this FHIR DevDays 2022 presentation: