Using a Text Mining Approach to Explore the Recording Quality of a Nursing Record System

Area of Science:

• Health Informatics
• Nursing Informatics
• Data Science in Healthcare

Background:

• Computerization of nursing records in Taiwan has generated vast unstructured text data.
• The quality assessment of these extensive electronic nursing records remains under-explored.

Purpose of the Study:

• To employ text mining for analyzing nursing record quality.
• To develop an auditing model and tools for electronic nursing records.
• To enhance the overall quality of electronic nursing records.

Main Methods:

• Retrospective analysis of 6,277 patient records from internal medicine departments (Jan-Jun 2014).
• Application of SAS Enterprise Guide and SAS Text Miner for text mining analysis.
• Expert review of records and comparison against a benchmark for evaluating SAS Text Miner's accuracy, sensitivity, and specificity.

Main Results:

• Analysis of 27,356 nurse-formulated events revealed an 8.08% similarity error with system-formulated events.
• SAS Text Miner achieved high performance with 96% (training) and 95% (testing) sensitivity, and 99% accuracy for both.
• Identified specific error rates including necessary names (29.72%), retained (17.53%), and error event names (10.15%), with 34.52% unclassified.

Conclusions:

• Text mining presents a viable and efficient strategy for auditing electronic nursing record quality.
• SAS Text Miner software demonstrates efficacy in rapidly detecting substandard nursing record content.
• Findings can inform in-service education to promote improved nursing documentation practices.