Quantifying clinical narrative redundancy in an electronic health record

Area of Science:

• Health Informatics
• Clinical Documentation Analysis
• Bioinformatics Applications in Medicine

Background:

• Electronic health records (EHRs) offer advantages over handwritten notes but can increase documentation time and data redundancy.
• Quantifying information redundancy in clinical notes is crucial for optimizing EHR systems.

Purpose of the Study:

• To quantify the extent of text redundancy within physician notes in an electronic health record system.
• To assess the feasibility of using sequence alignment algorithms for analyzing clinical documentation redundancy.

Main Methods:

• A retrospective design was used, analyzing electronic admission, progress, resident signout, and discharge summary notes from 100 patient admissions.
• A modified Levenshtein edit-distance algorithm was applied to align and compare documents, measuring duplicated text.
• The percentage of duplicated information was calculated as (total words - new words) / total words.

Main Results:

• Significant redundancy was observed, particularly in signout notes (average 78% duplicated) and progress notes (average 54% duplicated).
• Substantial information duplication occurred between different note types, such as from admission to progress notes.

Conclusions:

• The study demonstrates the feasibility of using bioinformatics sequence alignment algorithms to analyze redundancy in EHR narrative records.
• Findings provide a basis for further investigation into the clinical utility and potential risks associated with information redundancy in EHRs.