Study of EHR-mediated workflows using ethnography and process mining methods.

Area of Science:

• Health Informatics
• Human-Computer Interaction
• Workflow Analysis

Background:

• Individual use of rapid ethnography and data mining has limitations in studying clinical workflows.
• Rapid ethnography struggles with small sample reliability, while data mining faces accuracy issues with automatic analysis.

Purpose of the Study:

• To explore the combined use of rapid ethnography and process mining (ethno-mining) for studying clinical documentation tasks.
• To compare the metrics of vital signs charting across three hospital sites using different electronic health records (EHRs).

Main Methods:

• Employed rapid ethnography and process mining, integrating them into a novel 'ethno-mining' approach.
• Studied vital signs charting task duration across three hospital sites with varying EHR systems.

Main Results:

• Individual methods showed significant discrepancies in task duration (e.g., rapid ethnography: 159.6s, 38.2s, 431.3s; process mining: 518.6s, 345.5s, 119.7s).
• Ethno-mining identified and explained outliers, leading to similar mean task durations across sites (approx. 70-80s) after outlier removal.
• Outlier removal in ethno-mining harmonized task duration metrics, suggesting improved accuracy.

Conclusions:

• Integrating rapid ethnography and data mining (ethno-mining) provides more meaningful insights into clinical workflows than siloed approaches.
• Ethno-mining enhances the reliability and accuracy of workflow metric analysis, particularly by addressing outliers.
• This combined methodology offers a robust solution for understanding and optimizing clinical documentation tasks.