Dereddi Raja Reddy, MD, FACP FCCP
Assistant Professor, Program Director MS4 McGovern Medical School
MD Anderson Cancer Care Center
Houston, Texas
Disclosure information not submitted.
Joshua Botdorf, DO
Assistant Professor
University of Texas MD Anderson Cancer Center, Texas, United States
Disclosure information not submitted.
John Crommett, MD
Associate Professor
The University of Texas MD Anderson Cancer Center, United States
Disclosure information not submitted.
John Cuenca, MD (he/him/his)
Clinical Research Assistant
The University of Texas MD Anderson Cancer Center
Houston, Texas
Disclosure information not submitted.
Imrana Malik, MD, DABSM
Associate Professor
The University of Texas MD Anderson Cancer Center, United States
Disclosure information not submitted.
Robert Wegner, MD, FASA
Assistant Professor
The University of Texas MD Anderson Cancer Center, United States
Disclosure information not submitted.
Petra Grami, CCRN, DNP, MSN, RN
Director Specialty Units
The University of Texas MD Anderson Cancer Center, United States
Disclosure information not submitted.
Joseph Nates, MBA, MD
Professor, Deputy Chair, Director ICUs
University of Texas MD Anderson Center
Bellaire, Texas, United States
Disclosure information not submitted.
Title: Performance of an Automated Deterioration Index as an Early Warning System for ICU Outreach
Introduction/Hypothesis: Automated predictive scores have gained popularity as part of early warning systems. Several systems have been implemented with success including the Early Warning Score (EWS) and later Modified Early Warning Score (MEWS). Results from MEWS implementation have shown an increase in the number of deterioration calls, a decrease in cardiac arrests, and a 50% reduction in code calls. As a part of our ICU Outreach Program, we investigated the implementation of a real-time automated deterioration index score (DIS) to assist with the identification and early triage of ICU admissions.
Methods: A prospective cohort of 860 patients admitted to the ICU from between 9/1/2020 and 2/1/2021. Real-time monitoring of all hospitalized patients with the deployment of our Rapid Response Teams (RRT) from the ICU Command Center was implemented. We collected the DIS before, during, and after ICU admission. We evaluated the mortality at the time of hospital discharge of all patients admitted to the ICU based on two DIS ICU admission thresholds, 50 and 60.
Results: Over a 6-month period, 771 ICU patients admitted were discharged from the hospital and were included in the outcome analysis. Patients with a DIS threshold < 50 (437 patients, 56.6%), the mortality was 16.9%, reflecting average ICU mortality around the world. Patients with a DIS >50 (334, 43.3%) had a mortality of 46.4%. Patients with DIS < 60 (563, 73%) had a mortality of 21.49%, while patients with DIS >60 (208, 26.9%) had a mortality of 51.9%. For DIS thresholds of 50 and 60, the sensitivity was 67.69% versus 47.16%, and the specificity 66.97% versus 81.55% respectively. The Positive Predictive Values for DIS 50 and 60 were 46.41% versus 51.92%; the Negative Predictive Values for DIS 50 and 60 were 83.07% versus 78.51% respectively.
Conclusion: The automated DIS can be a useful additional early warning score for ICU outreach teams’ deployment and triage. DIS thresholds of 50 and 60 had low sensitivity, specificity, and predictive values. The determination of an optimal DIS threshold to trigger RRTs deployment requires further analysis.