SHM Converge Scientific Abstract ePoster Gallery-Validation of Saga Fall Risk Model: A Multicenter Retrospective Study

Pdf Summary

This multicenter retrospective observational study validated the Saga Fall Risk Model 2 (SFRM2), an eight-item inpatient fall prediction tool, across eight hospitals in Japan. The study included all inpatients aged 20 years and older from April 1, 2018 to March 31, 2021, covering chronic care, acute care, and tertiary settings.<br /><br />The main outcome was in-hospital falls. Researchers calculated individual SFRM2 scores and assessed model performance using the area under the curve (AUC), confidence intervals, and shrinkage coefficients. They also developed a revised version, SFRM2.1, by re-estimating coefficients with logistic regression using the same eight items, then validated it on a separate data subset.<br /><br />Overall, SFRM2 showed good predictive ability, with an AUC of 0.687 and a shrinkage coefficient of 0.996, suggesting stable performance. However, external validation across the hospitals indicated the AUC may be lower in some facilities, implying performance can vary by setting. Using the Youden index, the cutoff for SFRM2 was -2.85, corresponding to a fall probability of 5.5%, with sensitivity of 71.0%, specificity of 57.2%, positive predictive value of 3.9%, and negative predictive value of 98.8%.<br /><br />SFRM2.1 improved sensitivity to 81.1% while maintaining similar specificity (56.0%), with a cutoff of -4.02 and a fall probability of 1.8%. Its negative predictive value was 99.2%.<br /><br />The authors concluded that SFRM2 is a simple and useful tool for inpatient fall prediction and that accuracy may be improved by adjusting coefficients while keeping the model structure unchanged. This could help hospitals allocate resources efficiently and prevent falls more effectively.

Asset Subtitle

Risa Hirata

Meta Tag

Author List Chihiro Saito, Eiji Nakatani, Hidetoshi Aihara, Hitomi Shimada, Kaori Amari, Kazuya Kurogi, Kiyoshi Shikino, Maiko Ono, Mariko Yoshimura, Masahiko Ezoe, Masaki Tago, Midori Tokushima, Motoshi Fujiwara, Naoko E. Katsuki, Risa Hirata, Shizuka Yaita, Sh

Category Research

Concept Saga Fall Risk Model 2

Concept Inpatient Fall

Concept Discrimination

Concept AUC

Concept Recalibration

Distinguished Non-Finalist

Presenter Organization Saga University Hospital

Presenting Author Risa Hirata

Track Patient Safety

Keywords

inpatient falls

fall risk prediction

Saga Fall Risk Model 2

SFRM2

multicenter study

retrospective observational study

hospital validation

AUC

logistic regression

patient safety

Saga Fall Risk Model 2

Inpatient Fall

Discrimination

AUC

Recalibration