Predictive accuracy of the APACHE IV scores on mortality and prolonged stay in the intensive care unit of Dr Sardjito Hospital.

INTRODUCTION Acute Physiology and Chronic Health Evaluation (APACHE) is the most widely used scoring system in the intensive care unit (ICU). The APACHE IV showed a good level of discrimination and calibration on predicting mortality and prolonged stay (PLOS) in some countries. This study is aimed to determine the predictive accuracy of the APACHE IV score on mortality and PLOS at the ICU of Dr Sardjito General Hospital (SGH). MATERIALS AND METHODS This study involved all adult patients at the ICU of SGH during 2018 that met the inclusion criteria. The discrimination of APACHE IV scores on mortality and PLOS was analyzed with Receiver Operating Characteristic Curve, and the optimal cut-off point was assessed with the Youden Index. The calibration of the APACHE IV score was assessed with the Hosmer-Lemeshow goodness-of-fit test, and a p-value of >0.05 is considered a good calibration. RESULTS From the data of 742 patients, only 476 were included. The overall mortality and PLOS rate was 25.4 % and 15.1 %, respectively. The mean of APACHE IV score was 66.27±27.7. The area under the receiving curve with a 95% confidence interval for mortality is 0.99(0.97-1.00) and for PLOS was 0.68(0.62-0.74). The optimal cut-off point of the APACHE IV score for mortality was 78.9, with a sensitivity of 0.96 and a specificity of 0.96. The optimal cut-off point of the APACHE IV score for PLOS is 62.5 (in the 6th percentiles), with a sensitivity of 0.72 and a specificity of 0.61. The calibration is good for mortality prediction (p=0.98) but is poor for PLOS prediction (p=0.01). CONCLUSION APACHE IV score has excellent accuracy for mortality prediction but is poor for PLOS prediction in patients in the ICU of SGH.