Microsimulation Model Using Christiana Care Early Warning System (CEWS) to Evaluate Physiological Deterioration.

Area of Science:

• Biomedical informatics
• Health informatics
• Clinical interventions

Background:

• Evaluating clinical interventions like rapid response team (RRT) activations based on scoring systems is understudied.
• Physiological warning signs before hospital deterioration are widely researched, but intervention effectiveness needs more study.

Purpose of the Study:

• To evaluate Christiana Care Early Warning System (CEWS)-triggered RRT policies using patient demographics and policy structures.
• To identify optimal score thresholds for RRT activation to minimize critical patient conditions.

Main Methods:

• Developed a microsimulation model with regression analysis using retrospective electronic health records (2015-2016, 6000 patients).
• Compared RRT policies on subpopulations (age, gender, comorbidities) to find score thresholds minimizing time above critical CEWS values.
• Analyzed policy sensitivity to threshold changes based on average scores versus current value and change in CEWS.

Main Results:

• Policies using average scores were more sensitive to threshold changes than those using current value/change.
• A CEWS score threshold of 10 resulted in the lowest percentage of time spent in critical condition for most subpopulations.
• Identified specific score thresholds that optimize RRT activation policies for diverse patient groups.

Conclusions:

• The proposed microsimulation model offers a novel framework for evaluating RRT policies and simulating patient deterioration.
• Data-driven models are crucial for personalized care and optimizing early warning systems (EWS) for clinical interventions.
• This research provides a significant opportunity for health informatics to design and evaluate EWS-based interventions.