SuO036UREA TO CREATININE RATIO AND ITS CHANGE IS A POWERFUL AND MODIFIABLE PREDICTOR OF AKI AND NON-AKI MORTALITY IN EMERGENCY HOSPITAL ADMISSIONS

Abstract

Introduction and Aims: Serious adverse events (SAEs) in a hospitalized population include acute kidney injury (AKI), unexpected intensive care unit admission, cardiac arrest, and death. Such events are common, but hard to predict. High circulating urea and low creatinine are each associated with specific pathophysiology and poor outcomes. Urea to creatinine ratio (ur:cr) may thus predict risk. Methods: A population based cohort study of all adult emergency patients with at least two pairs or urea and creatinine results admitted over 156 months to three UK National Health Service Hospital groups (early 2007 to mid-2013). We fitted a non-linear multi-dimensional model (support vector machine) to a subset of the data to enable prediction of in-hospital mortality. As widely used, ur:cr >=80 (mmol/L: mmol/L) was defined as 'elevated'. Results: Of 79,949 patients studied, ur:cr was elevated on admission in nearly half (36,339, 45.5%). Ur:cr, its change and AKI were all associated (p<0.01 in all cases) with differences in survivor median (interquartile range) LoS, ranging from 3.5 days (2-6) to 19 days (10-33.5). Mortality (6.4%, 5080) was similarly influenced, being 0.4% in patients 65y with ur:cr elevated on admission, which remained elevated and who developed AKI. This represents an 87-fold mortality increase. A support-vector machine model was able to predict in-hospital death (sensitivity- 60%, specificity- 76-82%). Conclusions: Both a high or a rising ur:cr are powerful markers of in-hospital mortality after emergency hospital admission. They interact with each other, and with patient demographic factors, in a complex fashion in determining such outcome. We have shown that ur:cr is a more powerful predictor of hospital mortality than AKI. Of the patients who die with AKI, 80% have an elevated ur:cr. As such ur:cr may be used as a tool to identify the subset of AKI patients with highest AKI mortality risk. Tracking ur:cr can be used to determine risk. The mortality risk of ur:cr is modifiable and falls with ur:cr normalization. This suggests that interventions to reduce ur:cr, such as resuscitation or therapy of sepsis, may reduce mortality. The use of ur:cr in guiding acute therapy in AKI and non-AKI patients should be evaluated prospectively in interventional studies. (Figure Presented).