Performance and Improvement of Track and Trigger Scores Before Rescue and Transfer in the General Care Setting

Objectives: This work aims to understand the performance of early warning scores calculated using intermittent vital signs to identify general deterioration in the medical/surgical setting and explore the feasibility of implementing near real-time scores using continuous monitoring as part of a systems approach to inpatient assessment. Methods: This retrospective study used comparative trends and performance analyses to compare the ability of 4 early warning scores to identify patients requiring rescue and/or transfer to a higher level of care. Simulation was used to explore improvement in deterioration recognition with estimated scores calculated every 5 minutes as compared with those calculated using available intermittent vital signs to understand the potential impact of score calculation where continuous monitoring is available. Results: The National Early Warning Score performed better than other scores in identifying patients needing rescue and/or transfer to higher levels of care, with a sensitivity versus specificity analysis area under the curve value of 0.82. The National Early Warning Score also produced clinically acceptable misclassification ratios of 1:1 at scores of 6-7 or above. Simulation using this score, estimated every 5 minutes, improved detection of rescue and transfer events by over 5 hours on average. Conclusions: Early warning scores can be used for the detection of general deterioration events and, if calculated frequently using parameters from continuous monitoring systems, can augment alarm-based continuous monitoring to create a system to further reduce unwitnessed arrests and deaths.