Performance of SAPS3, compared with APACHE II and SOFA, to predict hospital mortality in a general ICU in Southern Europe

BACKGROUND AND OBJECTIVE: Simplified Acute Physiology Score (SAPS3) has not been validated in Southern European countries. The purpose of this study was to validate the ability of SAPS3 to predict hospital mortality in adult patients in an interdisciplinary intensive care unit in Southern Europe, compared with Acute Physiology and Chronic Health Evaluation (APACHE) II and Sequential Organ Failure Assessment (SOFA). METHODS: This is a cohort study of 864 patients with a prospective collection of SAPS3 and SOFA variables and retrospective analysis of APACHE II in a mixed intensive care unit at a teaching hospital in Spain throughout 2006. The performance of the systems was determined by examining their discrimination and calibration. RESULTS: The discrimination of SAPS3 was excellent, with an area under the receiver operating characteristic curve of 0.916, similar to APACHE II (area under the receiver operating characteristic curve = 0.893) or SOFA (area under the receiver operating characteristic curve = 0.846). The calibration was good for SAPS3 but insufficient for APACHE II. Hospital death rates were lower than those that were predicted by the models. CONCLUSION: Our data demonstrate a better calibration of SAPS3 than APACHE II. Calibration was sufficient only for SAPS3. Hospital mortality was lower than predicted by both models. The discrimination of SAPS3 is excellent, and, when it is customized for Southern Europe, SAPS3 accurately predicts mortality risk in our adult mixed-case ICU.