(A-214) Harmonizing Electronic Health Records from Multi-Healthcare Systems in Saudi Arabia: A Comprehensive Guide with SQL Script

Background: The harmonization method has been applied to Saudi Real-World Network (SRWEN), and structured query language (SQL) has been used to give users access to homogenized, organized, and cleansed electronic health records (EHRs). This method simplifies the complexity of various EHR variables, enhancing data usability and consistency. Moreover, it enables databases with different structures and formats to be mapped into a single standardized schema based on domains such as demographics, drugs, diagnosis, procedures, lab data (measurement), and visit data (emergency, inpatient, and outpatient).

Objectives: The objective of harmonizing SRWEN is to bring together EHR data from various healthcare systems, and ensure that the massive volumes of diverse EHR data are on the same scale, and have the same domains, formats, structures, and dimensions.

Methods: To map and consolidate SRWEN data extracted from multiple healthcare systems into a unified schema, we use PostgreSQL—relational database management system. Therefore, the process of data harmonization begins with creating a target harmonized schema for each domain to hold the unified EHR data, and a clustered index is added to specific columns to improve the query performance. Then, we create cross-database connections. Enabling the dblink, which is a core Postgres extension, allows direct connections so data can be retrieved from other Postgres databases. The next step consists of data extraction and transformation, namely, fetching data from several source databases and ensuring that schema mapping and wrapping are performed to align variables with the standardized structure and thus ensure consistency of formats. To identify the origin column, data integration, validation, and quality checks are carried out to bring the retrieval data into a target harmonized schema, and a tagging source is added to the identifier column for each schema to provide completeness, accuracy, and eliminate duplication.

Results: The unified structure of the SRWEN data from four different databases consolidates over 3.7 million unique patient records into standardized tables in 36 seconds.

Conclusions: By harmonizing SRWEN, researchers can effectively utilize existing healthcare data and obtain a comprehensive view of data structures in a single table for each domain. Additionally, this process enhances accuracy, consistency, and reliability due to the use of real-world data (RWD), supporting informed decision-making through research, studies, and further analytical processing.