Validation of a Fall Injury Prediction Model for use in Chronic Care Hospitals: A Multicenter Retrospective Observational Study

Abstract

Background: The self-developed Saga Fall Injury Risk Model (SFIRM) consists of six items: sex, age, presence of a referral letter, ambulance transportation, history of falls, and Bedriddenness Rank: and demonstrated good discriminative ability in a preliminary study conducted in acute and chronic care hospitals. This study validates the SFIRM for use in chronic care hospitals, where many older patients at high risk of fall-related injuries are admitted. Methods: This multicenter retrospective study targeted patients aged 20 years who were admitted to three chronic care hospitals from April 2018 to March 2021. Data on information and factors related to falls and fall-related injuries were collected from medical records. The primary outcome was the presence or absence of a fall-related injury, and the Area Under the Curve (AUC) of SFIRM was calculated. Results: The data of 1,184 patients (median age = 86 years; men = 45.6%) was available for analysis. Among them, 103 had fall-related injuries that occurred within the hospital. The AUC of the SFIRM was 0.856 (95% CI: 0.820- 0.891, p <0.001). Conclusion: The SFIRM demonstrated good accuracy with data from chronic care hospitals, indicating that it is a simple and helpful model applicable to diverse populations.