Risk of falls for hospitalized patients: a predictive model based on routinely available data

Area of Science:

• Healthcare outcomes research
• Patient safety
• Epidemiology

Background:

• The incidence of patient falls is a key indicator of nursing care quality.
• Comparing fall rates across different healthcare settings requires adjustment for patient case mix.
• Accurate fall rate assessment is crucial for quality improvement initiatives.

Purpose of the Study:

• To quantify the incidence and circumstances of patient falls.
• To develop a predictive model for patient falls using routinely collected data.
• To enable case-mix adjusted comparisons of fall rates between hospital settings.

Main Methods:

• A prospective study of hospitalized patients over one year at a Swiss university hospital.
• Definition of a case as any systematically reported accidental fall.
• Analysis using Poisson regression with routinely computerized discharge data, including patient age, gender, morbidity, surgical procedure, and length of stay.

Main Results:

• The incidence rate of first falls was 2.2 per 1000 patient-days.
• Fall incidence rates increased with the number of previous falls.
• Age, gender, morbidity, surgical procedure, and length of stay were significant predictors of falls, with significant interactions between length of stay and age, and morbidity and age.

Conclusions:

• A validated model using routine data can predict patient fall risk, adjusting for case mix.
• This model facilitates fair performance evaluation of hospitals regarding patient falls.
• Recommendations are provided for standardized measurement, monitoring, and assessment of fall rates.