USING TECHNOLOGY AND PROVIDER ASSESSMENT TO REDESIGN CLINICAL RECOGNITION SYSTEMS

Purpose: This study aims to develop an integrated clinical recognition system for patients hospitalized outside of intensive care units to detect early signals of adverse events (defined as medical emergency team activation, transition to higher-level care, or death) 24 hours before they occur using weighted physiological measures coupled with electronically embedded nursing assessment. 

Method: We developed a framework to quantify the physiological deterioration that results in adverse events by integrating a published Early Warning Score and a new Nurse Screening Assessment (NSA) tool. NSA tool included an electronic form with assessment questions focusing on: food and nutrition, respiratory, neurological, musculo-skeletal, gastrointestinal, genitourinary, skin, and patient safety assessment. We collected data retrospectively over a three month period between January and March 2015 from a single facility. Our methods involved process review, staff surveys, statistical analysis using the Area Under the Receiver Operating Characteristic (AUROC) and Chi-square tests, and workload efficiency assessment.   

Result: Our clinical recognition system provided an AUROC of 0.81 with regard to adverse events within 24 hours. Each NSA assessment question statistically significantly improved the discriminating performance at a significance level of 0.05. The largest increase in AUROC was achieved by the skin assessment, followed by the genitourinary assessment. Chi-square tests showed significant differences between trigger system alternatives with and without NSA tool (p<0.01). Workload efficiency assessment provided a framework to balance the monitoring effort and early detection of adverse events.

Conclusion: Patient rescue is a complex system that requires an integrated approach to improve care and outcomes. A leading principle of our study is the consideration of early recognition of physiological deterioration as a multifaceted system with interdependent processes and personnel, rather than discrete environments. Our clinical recognition system has the potential to detect early signals of adverse events during hospitalization 24 hours before they occur by taking advantage of providers’ clinical judgement. Our multidisciplinary approach allowed creating an environment of non-punitive problem solving by engaging a wide range of stakeholders including nurse managers, frontline providers, respiratory representatives, patient safety and quality representatives, data scientists, biostatisticians and industrial engineers. Looking beyond the vital sign-based risk prediction and using technology solutions to introduce electronic nursing assessment at our institution allowed realizing opportunities with workflow, information presentation, and clinical documentation.