Interpreting nurses' responses to clinical documentation alerts.

Area of Science:

• Health Informatics
• Nursing Practice
• Clinical Decision Support

Background:

• Electronic health records (EHRs) are crucial for patient data management.
• Accurate patient weight recording is vital for clinical decision-making.
• Automated alerts can flag potential data entry errors.

Purpose of the Study:

• To evaluate nurse responses to alerts detecting significant variations in recorded patient weights within EHRs.
• To assess the accuracy and impact of these alerts on clinical workflows.

Main Methods:

• Analyzing subsequent patient weights to classify alerts as true positive (TP) or false positive (FP).
• Observing nurse actions: overriding alerts, modifying entries, or abandoning data entry.
• Quantifying alert occurrence rates and nurse interaction patterns.

Main Results:

• Alerts were triggered in 2.74% of entries, with a 41.9% true positive (TP) and 58.1% false positive (FP) rate.
• Nurses overrode 30.3% of TP alerts and 97.3% of FP alerts.
• A low rate of entry modification was observed.

Conclusions:

• The alert system demonstrates an acceptable false positive rate.
• Nurses did not appear to alter entries to bypass the alert, indicating trust in the system.
• The implemented alert system enhances the accuracy of patient weight recording in EHRs.