SENTRY — A Systems-Based Model for Quality and Safety Surveillance with Continuous Quality Improvement in Emergency Care

Two decades after the patient safety movement reshaped how medicine understands harm, emergency care remains a complex and risk-laden environment. Despite major advances in measurement, reporting, and accountability, most safety frameworks still look backward - analyzing errors after they occur rather than anticipating and preventing them. The next frontier in safety is not more reporting; it is real-time learning. It demands systems that can detect and respond to risk as it unfolds, while learning and adapting in response to every signal. To meet that challenge, the authors developed and, in January 2023, implemented a systems-based model for quality and safety surveillance that turns the emergency department (ED) into a continuous learning environment. This model, Safety Evaluation and Networked Tracking for Real-Time Yield (SENTRY), combines Safety I (error prevention) and Safety II (adaptive resilience) principles to build a continuous surveillance and feedback ecosystem across three phases: prehospital care, emergency care including hospital-based transfers, and postdischarge care transitions and follow-up. Safety signals are captured through multiple inputs - including incident reports, key performance indicators, rapid-response activations, and 72-hour return audits - and reviewed through a standardized rubric grounded in national frameworks such as Agency for Healthcare Research and Quality Patient Safety Indicators and Reason's Swiss cheese model. Structured data from each case are integrated into a centralized Research Electronic Data Capture (REDCap) repository, enabling longitudinal analysis, recognition of recurring patterns, and prioritization of targeted interventions. Since implementation, automated pharmacist consults for prostacyclin medication have been associated with no observed delays in recognition and medication reconciliation. Comparing preintervention with postintervention, the mean and median time from ED arrival to prostacyclin reconciliation decreased by half. Automated aspiration precautions and nothing-by-mouth standing orders for high-risk patients were associated with the prevention of aspiration-related ED safety events, which decreased to zero events in 2024-2025, from an average of three identifiable events in 2023-2024. Use of an artificial intelligence-based triage tool was associated with improved throughput, alignment of acuity with clinical risk, and improved triage inequities. Same-day specialty pathways, including an outpatient diuresis clinic for heart failure patients, have been credited with averting more than 27 inpatient admissions over a 7-month period. Beyond these discrete outcomes, the framework has reshaped how safety is lived in daily practice - transforming surveillance from a retrospective exercise into a culture of continuous learning. Embedded feedback loops, charge nurse reporting, and nurse-driven safety rounds sustain engagement and momentum. Quantitative and qualitative insights now drive workflow redesign, policy, and education, allowing the ED to function as both a mirror reflecting vulnerability and a compass guiding improvement. While challenges such as alert fatigue, infrastructure demands, and sustainability remain, this model shows that continuous surveillance and adaptive learning can coexist within the realities of emergency care.