Brief Instrumented Mobility Testing Improves Fall Risk Stratification in Older Emergency Department Patients

Emergency department (ED) fall-risk screening often relies on measures that incompletely capture body movement signals relevant to future falls.

Test whether inertial measurement unit (IMU) features from a brief, modified, instrumented Timed Up and Go (miTUG) provide incremental prognostic value for 6-month falls after ED discharge beyond a clinical screening tool.

We conducted a prospective cohort study of community-dwelling adults ≥ 60 years discharged from an urban academic ED (September 2023-May 2024). Before discharge, participants completed a miTUG; four IMU features (sit-to-stand dominant frequency and duration; turn-to-sit spectral power and dominant frequency) were added to nine clinical predictors. The primary outcome was any fall within 180 days (6 months). Model performance was assessed using discrimination (AUC/C-index) and operating characteristics at ED-relevant thresholds. Secondary analyses examined models predicting time to first fall. Exploratory analyses examined patient sub-groups that may benefit from additional testing.

Among 360 participants, 94 (26.1%) fell within 180 days. The combined clinical+IMU model demonstrated modestly improved discrimination compared with the clinical-only model (AUC 0.72 vs. 0.67; Wilcoxon p = 0.19). At a prespecified 30% fall risk threshold, addition of IMU features improved sensitivity (0.57 vs. 0.45), specificity (0.80 vs. 0.76) and positive predictive value (0.50 vs. 0.39). In time-to-event analyses, the combined clinical+IMU model showed higher concordance (C-index 0.73 vs. 0.69) and better fit (likelihood-ratio p = 0.0006). Incremental gains were largest among adults ≥ 70 years, those with a recent prior fall, and those classified as lower risk by the clinical screen.

In older adults discharged from the ED, IMU features from a brief, mobility assessment added modest improvements in fall risk stratification beyond a clinical screen. These findings are hypothesis-generating and support the need for external validation and implementation studies before clinical adoption.