Development and Validation of an Electronic Adverse Event Model for Patient Safety Surveillance in Interventional...

Area of Science:

• Medical Informatics
• Patient Safety
• Radiology

Background:

• Comprehensive adverse event (AE) surveillance programs are uncommon in interventional radiology (IR).
• Developing effective AE surveillance is crucial for enhancing patient safety and quality of care in IR.
• Existing surveillance methods may not adequately capture all AEs in outpatient IR settings.

Purpose of the Study:

• To develop and validate a retrospective electronic surveillance model for identifying outpatient IR procedures with a high probability of AEs.
• To support patient safety and quality improvement efforts through proactive AE detection.
• To establish a reliable method for AE surveillance in IR.

Main Methods:

• Utilized electronic health records from the Veterans Health Administration (n=135,283 outpatient IR procedures).
• Developed electronic triggers based on post-procedure care to flag potential AEs.
• Employed mixed-effects logistic regression modeling, incorporating patient, procedure, and facility factors to predict AE occurrence.

Main Results:

• The developed model identified a >50% predicted probability threshold for likely AEs.
• The model demonstrated a positive predictive value of 68.9% and specificity of 95.3% in the development phase.
• During validation, the model identified approximately 70 IR cases per month with a likely AE, supporting its utility for ongoing surveillance.

Conclusions:

• An electronic, trigger-based approach to AE surveillance is feasible and effective in interventional radiology.
• This model can be implemented to enhance patient safety reporting and facilitate quality review processes.
• The developed surveillance system offers a scalable solution for monitoring AEs in IR procedures.