A Practical Approach to Assessing the Completeness of Electronic Health Records for Medical Research: A Data Quality...

Area of Science:

• Medical Informatics
• Health Data Science
• Clinical Research Data Management

Background:

• Data quality is crucial for medical research and AI, but real-world assessment faces challenges.
• Existing frameworks lack practical guidance for diverse data environments.
• A need exists for practical, interpretable data quality assessment approaches.

Purpose of the Study:

• To propose and validate a practical approach for assessing electronic health record (EHR) completeness for medical research.
• To integrate structural, rule-based, and descriptive analyses for robust completeness evaluation.
• To demonstrate meaningful interpretation of data quality metrics in practice.

Main Methods:

• A three-part completeness assessment was applied to a large-scale EHR dataset (2005-2023).
• Methods included structural completeness, rule-based missingness checks, and descriptive analyses of data diversity.
• Clinical data quality assessment tools were utilized on 1,798,153 patient records.

Main Results:

• Structural assessment revealed 7 of 39 tables were empty, indicating limited clinician free-text data capture.
• Rule-based assessment found significant missingness in vocabulary (30.6%) and observation data (23.8%).
• Descriptive analyses showed balanced gender distribution and predominantly Korean ethnicity.

Conclusions:

• This study demonstrates a practical, multi-perspective approach to measuring EHR completeness for medical research.
• The findings provide practical insights into data quality assessment and interpretation.
• The approach is valuable for researchers conducting data quality assessments and applying quality dimensions.