Using Electronic Health Records to Enhance Predictions of Fall Risk in Inpatient Settings.

Area of Science:

• Healthcare technology
• Patient safety
• Clinical informatics

Background:

• Hospitalized adults experience falls as a common adverse event.
• Current fall risk assessment tools lack sufficient accuracy.
• There is a need for improved methods to identify high-risk patients.

Purpose of the Study:

• To develop an automated, comprehensive fall risk score.
• To enhance the identification of hospitalized patients at high risk for falls.
• To evaluate the effectiveness of the developed risk score.

Main Methods:

• Developed the enhanced fall algorithm (EFA) using hierarchical logistic regression.
• Utilized data from 171,515 hospitalizations and 2,659 falls.
• Incorporated routine nursing assessments, labs, medications, demographics, and patient location from EHRs.

Main Results:

• The EFA increased model discrimination from 0.687 to 0.805.
• High-risk patient identification decreased from 28.0% to 16.2% without increasing fall rates.
• Fall detection increased from 3.1% to 5.1% with the EFA.

Conclusions:

• The EFA more accurately identifies high-risk patients than traditional methods like the Morse score.
• The algorithm redistributes patient risk categories effectively.
• The EFA integrates seamlessly with EHRs for automatic risk updates.