Executive Summary

Alarm fatigue is a well-documented threat to patient safety and clinician performance in monitored hospital environments, driven by a high proportion of non-actionable physiologic monitor alarms that contribute to excessive noise, workflow disruption, and desensitization to clinically meaningful alerts.

A data-driven alarm optimization initiative was implemented in a Cardiothoracic Intensive Care Unit (CTICU) using Nihon Kohden’s AlarmSense™ to reduce non-actionable alarms through targeted adjustment of monitor parameters. Baseline alarm data were analyzed to identify high-frequency, low clinical actionability alarms, and parameter modifications were developed in collaboration with clinical staff and engineering stakeholders.

Following implementation, objective alarm data demonstrated marked reductions across multiple alarm categories, including irregular heart rate (97%), non-sustained ventricular tachycardia (95%), desaturation (49%), apnea (33%), and SpO₂ alarms (10%), with reductions concentrated in alarm types previously identified as high-volume and low-value.

Clinician perception of alarm burden, assessed via pre- and post-implementation surveys, showed consistent improvements in perceived alarm overload, workflow interruptions, and ability to concentrate on patient care. Additionally, 70% of clinicians reported noticing a reduction in nuisance alarms, 82% reported improved awareness of clinically important alerts, and 76% reported perceived improvements in patient care and safety.

Extended analysis demonstrated an 85% reduction in total alarm time per bed-day, reclaiming approximately 3.1 hours of alarm-free time per bed per day, along with substantial decreases in average alarm duration for vital sign (85%), SpO₂ (91%), and arrhythmia (70%) alarms, indicating reductions in both frequency and temporal burden.

These findings demonstrate that targeted, data-driven optimization of monitor settings can reduce non-actionable alarm burden while improving the clinical signal-to-noise environment, with no significant adverse events reported, and represent a scalable strategy for improving alarm management and clinician experience in monitored care settings.

• 3-way collaboration between MaineHealth Maine Medical Center, Roux Institute, and Nihon Kohden

• Real-time predictive modeling integrating bedside and EMR data

• 200+ patient variables collected per case for comprehensive analysis

• 11 key adverse outcomes predicted, including acute kidney injury and prolonged ventilation

• 1200+ annual cardiac surgery patients expected to benefit from the model at Maine Medical Center

• 50 patients across 6 ICUs  monitored with RemoteSense

• 24/7 remote monitoring enabled from centralized locations.

• Seamless integration with existing hospital systems.

• Interoperability, enabling scalable, future-proof remote monitoring.

• Simplified workflows by consolidating data