Local Data Quality Assessments on EHR-Based Real-World Data for Rare Diseases

Area of Science:

• Health Informatics
• Clinical Research Data Management

Background:

• Rare diseases (RDs) require extensive data for research.
• The "Collaboration on Rare Diseases" (CORD-MI) project harmonizes electronic health record (EHR) data from German university hospitals.
• Integrating heterogeneous EHR data via Extract-Transform-Load (ETL) processes poses challenges to data quality (DQ).

Purpose of the Study:

• To investigate the impact of ETL processes on the data quality of rare disease information.
• To establish a methodology for assessing DQ before and after data transformation.

Main Methods:

• Evaluation of seven DQ indicators across three independent DQ dimensions.
• Comparison of DQ metrics and identification of DQ issues in RD data before and after ETL.
• Assessment of a methodology for real-world data quality evaluation.

Main Results:

• ETL processes significantly influence the quality of rare disease data.
• The developed methodology effectively detects DQ issues and validates DQ metrics.
• The study presents the first comparative analysis of RD data quality pre- and post-ETL.

Conclusions:

• ETL processes are critical and challenging steps in preparing rare disease data for research.
• The proposed methodology is effective for evaluating and improving the quality of diverse real-world data.
• This approach supports enhanced rare disease documentation and advances clinical research.