Related Experiment Video
Updated: Sep 4, 2026
Design and Analysis for Fall Detection System Simplification
Published on: April 6, 2020
Moderate and severe falls: predictive model based on six years of reports in southern Brazil
Sandiely de Araujo Mees1, Greici Capellari Fabrizzio1, Michelle Mariah Malkiewiez1
1Universidade Federal de Santa Catarina, Departamento de Enfermagem, Florianópolis, SC, Brazil.
Objective:
To characterize the sociodemographic and clinical profile of patients and identify predictive factors associated with moderate and severe falls.
Method:
An analytical cross-sectional study conducted using 300 fall reports (2019-2024). Poisson regression with robust variance was used to estimate prevalence ratios (PRs), logistic regression was performed for predictive modeling, and the area under the receiver operating characteristic (ROC) curve (AUC) was used to assess the model's discriminatory performance.
Results:
Most patients were male (55.0%), with a mean age of 56.4 years and a high prevalence of chronic diseases (86.2%). Most falls resulted in mild harm (54.2%) or moderate harm (44.6%). A hospital stay longer than 14 days (PR = 7.21; 95%CI: 1.68-26.47), male sex (PR = 2.89; 95%CI: 1.15-7.40), falls occurring during the morning shift (PR = 2.54; 95%CI: 1.17-6.73), disorientation (PR = 1.86; 95%CI: 1.11-4.55), chronic diseases (PR = 1.36; 95%CI: 1.11-3.69), and diseases of the circulatory system (PR = 1.58; 95%CI: 1.03-4.31) were associated with a higher risk of moderate/severe falls. Previous hospitalizations were identified as a protective factor (PR = 0.37; 95%CI: 0.22-0.94).
Conclusion:
falls were associated with clinical and healthcare-related factors. The model demonstrated moderate discriminatory performance (AUC = 0.77), supporting its potential use in the prevention of moderate and severe falls.
