Effectiveness of Sensor-Based Fall Prevention Systems in Hospitals: A Systematic Review and Meta-Analysis

Why clinicians should know about it

• Picked for Health Informatics (paper of the day, 9 October 2026): Sensor-based fall prevention systems, ML impact

Abstract

OBJECTIVE: To synthesize evidence on the effectiveness of hospital sensor-based fall prevention systems and examine whether technology type, patient risk, and implementation strategy modify their impact. METHODS: We searched MEDLINE, Embase, CINAHL, CENTRAL, Web of Science, and trial registries through April 30, 2025. Interventional studies evaluating hospital sensor-based fall prevention systems were eligible, including randomized controlled trials and quasi-experimental designs. Two reviewers independently extracted data and assessed risk of bias using the Cochrane Risk of Bias tools. Random-effects meta-analyses applied the Paule-Mandel estimator with Hartung-Knapp-Sidik-Jonkman adjustment. Incidence rate ratios (IRR) were pooled for fall rates and fall-related injuries, and odds ratios (OR) for patients with at least one fall. Subgroup analyses examined technology generation and risk profile, with restricted analyses for mandated implementation. Certainty of evidence was rated with the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach. RESULTS: Sixteen studies were included, seven in meta-analysis. No significant impact was found for fall rates (IRR 0.77, 95% CI 0.38-1.58; I2 = 92%), fallers (OR 1.24, 95% CI 0.76-2.02; I2 = 73%), or fall-related injuries (IRR 0.98, 95% CI 0.81-1.18; I2 = 0%). Machine learning-based systems reduced fall rates (IRR 0.50, 95% CI 0.32-0.79), while rule-based systems did not. Mandatory implementation was associated with lower fall rates (IRR 0.51, 95% CI 0.37-0.70) but not fewer fallers. Certainty of evidence ranged from very low to low. CONCLUSION: Sensor-based systems showed no consistent benefit for preventing hospital falls or injuries, though machine learning and mandated use appeared more promising. Given the limited number of trials, heterogeneity, and low certainty, further research should focus on rigorous evaluation, workflow integration, and cost-effectiveness. REGISTER: This review was registered with PROSPERO (CRD420251123960; registered on 11 August 2025). CLINICAL TRIAL NUMBER: not applicable.

Abstract as published, via PubMed.