event-icon
Description

Identifying patients at risk of deterioration in the hospital and intervening more quickly to prevent adverse events is a top patient safety priority. Early warning scores (EWS) identify at risk patients, but there is much opportunity for improvement particularly related to increasing lead time – the time from an alert trigger to adverse event (e.g., cardiac arrest, death). Our team develops healthcare process models of clinical concern (HPM-CC) and in this work has identified documentation signals that are proxies of nurses concern and can be used to predict patient risk earlier than current EWS systems that rely only on physiological data. We compared the performance of a validated EWS - the MEWS - to our novel model (MEWS-CC) comprised of MEWS criteria plus 3 proxy variables of nursing concern. MEWS-CC performed similarly to MEWS, with the added benefit of increased the time from EWS trigger to event by 5-26 hours.

Learning Objective: 1. To articulate evaluation metrics for early warning systems and their statistical and clinical relevance.
2. To understand proxies of nursing concern that can be measured in EHR data

Authors:

Sarah Collins Rossetti (Presenter)
Columbia University

Christopher Knaplund, Columbia University
David Albers, University of Colorado
Abdul Tariq, New York Presbyterian
Kui Tang, New York Presbyterian
David Vawdrey, New York Presbyterian
Natalie Yip, Columbia University Medical Center
Patricia Dykes, Brigham and Women's Hospital
Jeffrey Klann, Harvard Medical School
Min Jeoung Kang, Brigham and Women's Hospital
Jose Garcia, Brigham and Women's Hospital
Li-heng Fu, Columbia University
Kumiko Schnock, Brigham and Women's Hospital
Kenrick Cato, Columbia University

Presentation Materials:

Tags