Validation of SAPS 3 Admission Score and comparison with SAPS II

The objective of this study was to validate the Simplified Acute Physiology Score SAPS 3 Admission Score (SAPS 3) and to compare its fit with that of SAPS II in an independent sample of patients admitted to a single-centre intensive care unit (ICU). The data for all adult patients consecutively admitted to an eight-bed ICU of a 700-bed university hospital between 1 January 2006 and 2 September 2007 were collected. SAPS II and SAPS 3 were computed, as well as the predicted hospital mortality. The calibration of SAPS II and SAPS 3, according to the general equation (GE), and equations for Southern Europe and Mediterranean countries (SE&MC), and Central and Western Europe (C&WE), were assessed by the goodness-of-fit Hosmer-Lemeshow H and c statistics. Standardized mortality ratios (SMR) with 95% confidence interval (95% CI) were computed for SAPS II and SAPS 3 equations. Six hundred and eighty-four patients were studied (males 63%). The median age was 73 (quartiles 65-80) years. The fit of SAPS 3 using the C&WE equation (H 13.49, P=0.095; c 12.73, P=0.121) as well as that of SAPS II was acceptable (H 6.02, P=0.644; c12.08, P=0.147), while SAPS 3 GE (H 23.36, P=0.002; c 22.37, P=0.004) and S&MC (H 25.73, P=0.001; c 26.19, P=0.001) did not fit well. SAPS 3 GE, SAPS 3 SE&M Countries and the SAPS II significantly over estimated the mortality. Only 95% CI of SMR for SAPS 3 C&WE included 1 (SMR 0.97; 95% CI 0.89-1.05). Each ICU should identify the SAPS 3 equation most suitable for its case mix. The SAPS II model tended to overestimate the mortality.