Predicting cardiac arrest on the wards: a nested case-control study.

Area of Science:

• Critical Care Medicine
• Clinical Prediction Models
• Patient Monitoring

Background:

• Current rapid response team activation criteria lack statistical derivation from ward vital signs.
• Optimal vital sign predictors for cardiac arrest (CA) remain undetermined.
• The temporal relationship between vital sign changes and impending CA is not well understood.

Purpose of the Study:

• To identify the best vital sign predictors of CA on hospital wards.
• To determine the earliest time point at which vital signs can predict CA.
• To evaluate the effectiveness of the Modified Early Warning Score (MEWS) in predicting CA.

Main Methods:

• A nested case-control study involving 88 patients who experienced CA and 352 matched controls.
• Comparison of vital signs and MEWS on admission and during the 48 hours preceding CA.
• Analysis of prediction accuracy using receiver operating characteristic curves.

Main Results:

• Maximum MEWS was the strongest predictor of CA (AUC 0.77) in the 48 hours preceding the event.
• Maximum respiratory rate (AUC 0.72) and maximum heart rate (AUC 0.68) were also significant predictors.
• MEWS was significantly higher in CA cases by 48 hours prior to the event, with increasing differences over time.

Conclusions:

• MEWS differs significantly between CA and control patients by 48 hours before the event.
• Current MEWS criteria may not be optimal, as they include poor predictors (e.g., temperature) and omit significant ones (e.g., diastolic BP, pulse pressure index).
• Further refinement of MEWS or development of new prediction models incorporating key vital signs is warranted.