Multisite Pragmatic Cluster-Randomized Controlled Trial of the CONCERN Early Warning System

Area of Science:

• Clinical Informatics
• Patient Safety
• Healthcare Quality Improvement

Background:

• Delayed detection of patient deterioration often stems from limited data in early warning systems (EWS) and poor team situational awareness.
• This can lead to delayed clinical interventions and poorer patient outcomes.

Purpose of the Study:

• To evaluate the impact of the COmmunicating Narrative Concerns Entered by RNs (CONCERN) Early Warning System (EWS) on patient mortality and length of stay.
• To test the hypothesis that informing care teams with CONCERN EWS data reduces mortality and length of stay.

Main Methods:

• A one-year, multisite pragmatic clinical trial involving cluster-randomization of acute and intensive care units.
• The CONCERN EWS intervention utilized machine learning on real-time nursing surveillance documentation patterns to predict patient deterioration risk.
• Adult patients in acute and intensive care units were included, excluding those on comfort care or with specific advance directives.

Main Results:

• The intervention group (CONCERN EWS) showed a 35.6% decreased risk of in-hospital death and an 11.2% decreased length of stay compared to the usual-care group.
• A 7.5% decrease in sepsis risk was observed in the intervention group.
• An increased risk of unanticipated ICU transfer (24.9%) was noted in the intervention group.

Conclusions:

• Integrating a hospital-wide EWS based on nursing surveillance patterns into the EHR workflow significantly reduces in-hospital mortality, sepsis incidence, and length of stay.
• The CONCERN EWS demonstrates the potential to improve patient safety and healthcare efficiency through enhanced early detection of deterioration.