Healthcare data is growing fast, but traditional ETL methods can’t keep up. Most systems still rely on hard-coded rules to move data, which are time-consuming to maintain and prone to errors. Domain-based routing offers a smarter approach. By using the built-in structure of medical codes like SNOMED and LOINC, it automatically maps healthcare data to the right place, without manual intervention. This shift to semantics-driven transformation simplifies data pipelines and gets you research-ready data faster.
ETL in healthcare has long been riddled with limitations. Traditional ETL processes rely on developers manually defining thousands of transformation rules such as:
if (code == “185349003”) then visit_occurrence_table
if (code == “38341003”) then condition_occurrence_table
While these rules technically get the job done, they often turn into maintenance headaches—hard to track, easy to break, and always needing updates. As healthcare data keeps growing in volume and complexity, it’s clear the old way isn’t enough. What’s needed now is a smarter, self-sustaining approach to ETL pipelines—one that can adapt and keep up without constant babysitting.
Introducing Semantic Routing in ETL Healthcare
Domain-based routing is a paradigm shift in ETL healthcare systems. It transforms static, syntax-based transformations into semantic-driven workflows that leverage the built-in classification of medical terminology standards to automatically determine where each data point belongs.
How Medical Codes Guide Intelligent ETL Healthcare Routing
Medical codes like SNOMED CT and LOINC aren’t just labels—they come packed with context. Each one helps sort data into meaningful buckets like “Visit,” “Measurement,” or “Condition.” That built-in logic isn’t just helpful—it’s key to making modern ETL strategies in healthcare smarter and more efficient.
SNOMED CT Example:
🔸 Code: 185349003
🔸 Description: “Encounter for check up”
🔸 Domain: Visit
🔸 Destination: visit_occurrence table
LOINC Example:
🔸 Code: 33747-0
🔸 Description: “Glucose [Mass/volume] in Blood”
🔸 Domain: Measurement
🔸 Destination: measurement table
By utilizing domain metadata, semantic routing eliminates the need for hand-coded transformation logic, revolutionizing ETL in healthcare data management.
The Power of Athena in ETL Healthcare
The OHDSI collaborative maintains Athena, a robust vocabulary service with over 6 million standardized medical concepts across:
🔸 SNOMED CT
🔸 LOINC
🔸 RxNorm
🔸 ICD-10
🔸 CPT
Each concept in Athena contains OMOP-specific metadata, including a domain_id, which allows ETL healthcare platforms to route data accurately without manual effort.
One FHIR Document, Multiple OMOP Records
Traditional ETL systems in healthcare often rely on rigid one-to-one mappings, which can be limiting and inflexible. Domain-based routing opens up a smarter way to work—with one-to-many transformations that better reflect real-world clinical data. Take a FHIR resource for a physical exam, for example—it might include:
🔸 Visit code (SNOMED: 185349003) → visit_occurrence
🔸 Hypertension diagnosis (SNOMED: 38341003) → condition_occurrence
🔸 Blood pressure measurement (LOINC: 8480-6) → measurement
This approach allows ETL healthcare pipelines to create multiple analytical records from a single source—improving both completeness and usability.
Behind the Scenes: How Domain-Based ETL Healthcare Works
The semantic routing engine processes data in four stages:
- Code Extraction: Parse FHIR resources for embedded medical codes.
- Concept Enrichment: Query Athena to retrieve metadata for each code.
- Domain Classification: Use the domain_id to determine OMOP destination(s).
- Record Generation: Create structured OMOP records with traceability.
This architecture enables faster, more accurate, and easier-to-maintain ETL healthcare infrastructure.
Key Benefits of ETL in Healthcare
Semantic routing delivers powerful advantages:
🔸 >2000 records/sec throughput
🔸 Parallel processing and vocabulary caching
🔸 Automated duplicate resolution
🔸 Consistent and accurate mappings
🔸 Full data lineage and quality controls
Unlike traditional systems, this method future-proofs your ETL healthcare pipelines against evolving vocabularies and standards.
Quality, Consistency, and Maintainability
When you build transformation logic around standard vocabularies, you’re not starting from scratch—you’re working with a shared language that already knows how healthcare data fits together.
🔸 Mappings remain consistent across time and systems
🔸 Unmapped codes are automatically flagged
🔸 Version-independent logic supports all OMOP CDM versions
🔸 Vendor-agnostic design works with any FHIR-compliant EHR
This elevates ETL healthcare workflows from being purely operational to becoming analytical and insight-ready.
The Future of ETL Healthcare Systems
Domain-based routing isn’t just a backend upgrade—it’s a smarter way forward. As healthcare teams push to turn routine data into real-world insights, this approach makes ETL more automated, context-aware, and ready to scale with the complexity of modern care.
Implementation Checklist
Before adopting domain-based routing, evaluate:
🔸 Coverage of vocabularies for your use cases
🔸 Volume of FHIR resource processing
🔸 Quality standards for research-readiness
🔸 Integration complexity with your current ETL healthcare architecture
Final Thoughts
Domain-based routing is changing how ETL works in healthcare. Instead of constantly rewriting fragile rules, teams can now use a smarter, standards-based approach that grows with their needs and doesn’t require constant upkeep.
This blog draws inspiration from the FHIR Analytics community and thought leaders like Carl Anderson, who are leading the way in making healthcare data truly interoperable and meaningful.
Domain-based routing is a semantic-driven approach to data transformation that uses medical terminology metadata (like SNOMED, LOINC, RxNorm) to determine where data should be routed in a target data model such as OMOP, eliminating the need for hard-coded mapping rules.
Traditional ETL relies on manually hard-coded transformation rules, which are error-prone, difficult to maintain, and inflexible in the face of evolving medical terminologies. This leads to high maintenance costs and potential data quality issues.
Standardized codes like SNOMED and LOINC include domain metadata (e.g., “Condition”, “Measurement”) that specify their clinical context. Domain-based routing engines use this metadata to automatically determine the correct OMOP CDM table for each code.
