Strategies to Predict and Prevent In-Hospital Cardiac Arrest

The purpose of this study is to evaluate the accuracy of medical personnel in their ability to predict the likelihood of non-intensive care (ICU), ward patients to clinically deteriorate (defined as a cardiac arrest, unplanned ICU transfer, or unexpected death)via the use of a clinical judgement-based tool designed for this study, Patient Acuity Rating (PAR), to predict short-term clinical deterioration. We will compare the ability of this tool to predict clinical deterioration compared to accepted physiology-based tools and tools combining judgment and physiology as well as other markers of deterioration such as physician order changes. We will compare the sensitivity, specificity and area under the curve of these combined models to the predictive models including only physiology or clinical judgment. We will assess the correlation between specific physician orders and patient deterioration to determine whether specific clinical activities, such as emergently obtained radiology exams, predict impending deterioration. We hypothesize that PAR will be a useful tool for predicting clinical deterioration across the institution and that it will have a higher average accuracy for predicting clinical deterioration in non-ICU inpatients within 24 hours than the physiology-based tools alone. We further hypothesize that a combined metric which includes both the PAR and the individual physiologic components that comprise physiologic tools will not significantly improve prediction over the PAR alone. We further propose to use PAR to prospectively risk stratify patients for preemptive evaluation by the Rapid Response Team. We hypothesize that intervening on high risk patients by preemptively activating the hospital's Rapid Response Team (to assess and treat patients as needed) will decrease cardiac arrest rates and mortality.