Multicenter Evaluation of an Interoperable System for Automated Guideline Adherence Monitoring in ICUs

Area of Science:

• Critical care medicine
• Health informatics
• Clinical quality improvement

Background:

• Adherence to critical care guidelines is crucial for patient outcomes but challenging to monitor.
• Current manual methods are time-consuming and prone to error.
• Variability in real-world adherence necessitates robust evaluation systems.

Purpose of the Study:

• To develop and validate an automated system for critical care guideline adherence.
• To compare the system's performance against manual expert review.
• To identify factors influencing guideline adherence in diverse hospital settings.

Main Methods:

• Retrospective, multicenter observational study across five university hospitals.
• Development of an automated system using digitally encoded guidelines and patient data.
• Comparison of automated monitoring with expert manual review for applicability and adherence.

Main Results:

• The automated system achieved 97.0% accuracy, significantly higher than human reviewers (86.6%).
• System processing speed was over 2000 patient days/second versus 2 patient days/minute for manual review.
• Substantial variation in adherence rates was observed across sites and time.

Conclusions:

• The automated system is accurate, efficient, and scalable for critical care guideline adherence monitoring.
• System performance depends on consistent and structured clinical documentation.
• This technology enables automated quality management in intensive care settings.