Abstract 16310: Derivation of a Cardiac Arrest Prediction Model Using Ward Vital Signs
INTRODUCTION: Current rapid response team (RRT) activation criteria are not evidence-based, which may in part explain their disappointing results in clinical trials. We derived a cardiac arrest (CA) prediction model using ward vital signs and compared it to the Modified Early Warning Score (MEWS), a commonly cited RRT activation system.
METHODS: All patients hospitalized from November 2008 to January 2011 at an academic institution were included in this study. Vital signs recorded on the ward were extracted from the medical record from admission until discharge (controls), first intensive care unit (ICU) transfer, or first ward CA. Vital sign cut-points were determined using univariate logistic regression. Multivariate stepwise logistic regression was then used to derive the final model to predict CA, utilizing integer multiples of the model coefficients to allow comparison to the MEWS. Scores from the final model and MEWS were calculated for each vital sign set in the entire dataset. Each patient's maximum score during the study period was used to compare the area under the receiver operating characteristic curve (AUC) between the derived score and MEWS.
RESULTS: A total of 47427 patients were included in the study (88 CA patients, 2820 ICU transfers, and 44519 controls). The derived model is shown. The derived model was a better predictor of CA than MEWS (AUC 0.84 vs. 0.76; P=0.001). At a cut-off of >17, the derived model had a sensitivity of 53.4% and specificity of 89.9% versus the MEWS (cut-off >4) sensitivity of 47.7% and specificity of 89.9%. The final model also predicted ICU transfer better than the MEWS (AUC 0.71 vs. 0.67; P<0.001).
CONCLUSIONS: We derived a CA prediction tool using ward vital signs that is simpler than the MEWS and more accurate for detecting CA and ICU transfer. Implementation of this system may decrease RRT resource utilization while providing the best opportunity to improve patient outcomes compared to current systems.
- © 2011 by American Heart Association, Inc.