Reverse ETL for FHIR: Getting Clinical Data Products Back into the Chart
Reverse ETL for FHIR: Getting Clinical Data Products Back into the Chart - Hakkoda
Summary
This article discusses Reverse ETL for FHIR, a process designed to move curated clinical data products (such as readmission risk scores) from data warehouses back into the Electronic Medical Record (EMR) for clinical action. It details how to ensure data reaches the point of clinical decision-making rather than remaining in BI tools.
Details
The article addresses the challenge of Reverse ETL, which involves taking data products computed in a data warehouse's 'gold layer' and delivering them into the clinical setting's Electronic Medical Record (EMR). It uses FHIR as a test case, comparing the two extremes of the 'coupling spectrum': posting a transaction Bundle via REST API versus landing NDJSON on an external stage. The core difficulty lies not in the transport, but in the resource itself. To transform a computed score into a conformant FHIR Observation resource, critical steps include 'terminology mapping' (normalizing local codes to standards like ICD-10, LOINC, or SNOMED) and 'reference resolution' (generating resources in dependency order). Successfully completing these steps allows the data to reach the point of clinical decision-making.
Original content copyright by respective publishers