Repeatable process for extracting health data from HL7 CDA documents.

Area of Science:

• Health Informatics
• Biomedical Data Science

Background:

• Clinical Document Architecture (CDA) is a standard for electronic health records.
• The Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM) standardizes health data for research.
• Converting real-world CDA data to OMOP CDM is crucial for large-scale health research but presents challenges.

Purpose of the Study:

• To develop and validate a repeatable Extract-Transform-Load (ETL) pipeline for converting Estonian HL7 CDA documents into the OMOP CDM.
• To address the initial steps of data extraction and restructuring before the mapping phase.
• To create a high-quality, structured data format adaptable to evolving data exchange standards and diverse study needs.

Main Methods:

• Developed a repeatable ETL pipeline for data extraction, cleaning, and restructuring from CDA to OMOP CDM.
• Designed the pipeline for adaptability to data format changes and various CDA document subsets.
• Iteratively developed the pipeline to ensure swift error detection and correction.

Main Results:

• Successfully transformed diagnosis codes, body weight, eGFR measurements, and PAP test results from CDA to OMOP CDM.
• Demonstrated ease of extracting structured data, though challenges in harmonizing coding systems and extracting free-text lab results were noted.
• The iterative process enhanced pipeline efficiency and facilitated error correction.

Conclusions:

• An ETL pipeline was developed over a decade to effectively transform HL7 CDA into OMOP CDM in Estonia.
• The pipeline is repeatable, adaptable to various data subsets, and valuable for health data researchers.
• While tested in Estonia, the principles are broadly applicable to varying international health data standards and CDA's continued relevance for retrospective studies.