1

Abstract

Introduction: Today, critical care in the United States costs more than $80 billion annually. With an aging population and the growing demand for critical care, it is predicted that the number of staffed intensive care unit (ICU) beds will become increasingly inadequate. Limited resources, intensivists, and ICU beds warrant investigating models for predicting who will benefit from admission to the ICU. This study presents a possible model for identifying patients who might be too well to benefit from admission to the ICU. Methods: We retrospectively identified all patients admitted to our 54 bed medical-surgical ICU between 11/2009 and 2/2013. We used the APACHE Outcomes database to identify patients who on day one of ICU admission received one or more of 33 subsequent active life supporting treatments. We compared two groups of patients: Low Risk Monitor (LRM) (Patients who did not receive active treatment on the first day and whose risk of ever receiving active treatment was <= 10 %) and Active Treatment (AT) (patients who received at least 1 of the 33 ICU treatments on any day of their ICU admission). Results: There were 2293 admissions (29.7%) in the LRM group, and 5430 admissions (70.3%) in the AT group. APACHE IV score was 34.3 (+/- 13.4) for the LRM group versus 58.7 (+/- 25) for the AT group (P < 0.0001). ICU length of stay (LOS) in days was 1.6 (+/- 1.7) for the LRM group versus 4.3 (+/- 5.3) for the AT group (P < 0.0001). ICU mortality was 0.7 % for the LRM group compared to 9.6 % for the AT group (OR = 15.0; 95% confidence interval [CI], 9.2 – 24.8, p <0.0001). Hospital mortality was 1.8 % for the LRM group compared to 15.2 % for the AT group (OR = 9.8; 95% CI, 7.1 – 13.4, p <0.0001). Conclusions: The outcome for low-risk monitor patients in our ICU suggests they may not require admission to intensive care. This may provide a measure of ICU resource use. Improved resource use and reduced costs might be achieved by strategies to provide care for these patients on floors or intermediate care units. This model will need to be validated in other ICUs and prospectively studied before it can be adopted for triaging admissions to ICUs.