Reconstruction of a Large-Scale Real-World HL7 v2.5 Database Using HL7 FHIR®

Area of Science:

• Health Informatics
• Clinical Data Management
• Interoperability Standards

Background:

• The 6NC-EHRs Database (6NC-EHRsDB) aggregates clinical data from six national electronic health records (EHRs).
• HL7 v2.5 is a widely used standard for EHR data exchange.
• HL7 FHIR® (Fast Healthcare Interoperability Resources) is a newer standard aimed at improving data interoperability.

Purpose of the Study:

• To evaluate the reproducibility of a database constructed in HL7 v2.5 when converted to HL7 FHIR®.
• To identify discrepancies and areas for improvement in FHIR implementation for clinical data.

Main Methods:

• Mapped 80 items from 6NC-EHRsDB to FHIR resources (Condition, Medication Request, Observation LabResult).
• Compared FHIR-extracted data with 6NC-EHRsDB for record count, data item coverage, and value accuracy.
• Assessed agreement rates for specific data points like prescription start dates and standard values.

Main Results:

• 74 out of 80 items were successfully mapped to FHIR.
• High record match rates were observed (99.9% Condition, 99.2% Medication Request, 99.9% Observation LabResult).
• Significant issues with data item coverage (missing standard codes) and value accuracy were identified, with lowest agreement rates for Medication Request (69.1%) and Observation LabResult (58.9%).

Conclusions:

• Reproducibility of HL7 v2.5 databases using FHIR can be enhanced by addressing identified implementation deficiencies.
• Improving FHIR implementation, particularly with standard code assignment, is crucial for better data utilization and interoperability in medical information systems.