Post-surgical fall risk prediction: a machine learning approach for spine and lower extremity procedures.

Area of Science:

• Medical Care Quality
• Patient Safety

Background:

• Falls and pressure injuries are key indicators of clinical care quality in Taiwan.
• Patient falls can lead to a range of injuries, from minor bruises to severe trauma like fractures and head injuries.

Purpose of the Study:

• To identify factors contributing to inpatient falls among patients undergoing spinal and lower limb surgeries.
• To establish early warning measures for fall risk assessment.

Main Methods:

• Analysis of 2,948 medical records from patients at Veterans General Hospital, Taichung, Taiwan.
• Inclusion of patient demographics, vital signs, health conditions, diagnoses, medications, and admission type.
• Utilized logistic regression and random forest methods, incorporating fall history for model training and variable selection.

Main Results:

• Logistic regression incorporating fall history proved effective for risk assessment.
• Increased fall risk was associated with wheelchair/stretcher admission for spine/lower limb surgery and advanced age.
• Psychotropic and antihypertensive medications increased fall risk, while sleeping pills decreased it. Higher body temperature and poor vision also correlated with higher fall odds. Arthritis showed a decreased fall risk.

Conclusions:

• Refined fall risk assessment processes can improve patient care quality.
• Effective fall risk identification can help reduce caregiver workload.
• The study provides valuable insights for developing targeted preventive strategies for surgical patients.