Related Experiment Video
Updated: Aug 7, 2026
Design and Analysis for Fall Detection System Simplification
Published on: April 6, 2020
FALLPREDICT: Predicting Fall-Related Health Care Encounters in Persons with Vascular Lower Limb Amputation Using
Daniel C Norvell1, Alison W Henderson2, Elliott Lowy2
1VA Puget Sound Health Care System, 1660 S. Columbian Way, Seattle, WA 98101, USA; VA Center for Limb Loss and Mobility (CLiMB), 1660 S. Columbian Way, Seattle, WA 98101, USA; Department of Rehabilitation Medicine, University of Washington, 325 Ninth Avenue, Box 359612, Seattle, WA 98104, USA.
Objective:
To develop a novel preliminary prediction model (FALLPREDICT) that predicts the risk of falls in patients who have undergone a lower limb amputation (LLA) in the 12-months following their first definitive lower limb prosthesis (LLP) prescription. The model is designed for patients who have undergone initial transtibial (TT) or transfemoral (TF) amputation due to sequelae of diabetes and/or peripheral artery disease (PAD). The overall goal of this research is to identify individuals at relatively increased risk of falls to help inform targeted fall prevention strategies.
Design:
Retrospective cohort study of patients who underwent an incident dysvascular LLA at the TT or TF level who received a definitive LLP (considered time zero for the model). Patients were tracked 12 months forward to identify incident falls.
Setting:
The VA Corporate Data Warehouse and the National Prosthetics Patient Database.
Participants:
1,690 Veterans who underwent an initial TT or TF amputation due to diabetes and/or PAD and received a qualifying LLP between March 1, 2018, and November 30, 2020.
Interventions:
Not applicable MAIN OUTCOME MEASURE: A health care encounter that documented a fall code. Incident falls were designated as any record of an ICD-10 external cause code for falls (range W00-W19, e.g., W01.0XXA) identified after the prosthesis prescription date (time zero).
Results:
Variable selection, using a backwards stepwise logistic regression, led to a final model of nine predictors (amputation level, age, BMI, a fall prior to amputation, prior falls, depression, peripheral neuropathy, kidney dialysis and COPD) that demonstrated good discrimination (c-statistic 0.71) and satisfactory calibration across predicted risk groups, supporting its ability to provide meaningful risk stratification and reliable absolute risk estimates.
Conclusions:
The FALLPREDICT model can be used for identifying patients at high risk of a future fall to help guide targeted fall intervention strategies.
Related Concept Videos
01:25Errors occurring during blood pressure monitoring
Several factors...
01:23Peripheral Artery Disease V: Postoperative Nursing Management