A comparison of the performance of Simplified Acute Physiology Score 3 with old standard severity scores and customized scores in a mixed medical-coronary care unit

Abstract

Background: Recently, the Simplified Acute Physiology Score (SAPS) 3 was developed. The objective of this study was to compare the performance of SAPS 3 with those of SAPS II, the Acute Physiology and Chronic Health Evaluation (APACHE) II and customized scores in predicting hospital mortality in critically ill mixed medical-coronary patients.

Methods: A prospective cohort study was conducted over a 2.5-year period in a 10-bed mixed medical-coronary care unit of a tertiary referral university teaching hospital in Thailand. The probabilities of hospital mortality of SAPS 3, SAPS II and APACHE II were calculated using standard equations.

Results: A total of 2022 patients were enrolled. The performance of SAPS 3 was similar to those of the old standard severity scores. All three scores had excellent discrimination, with areas under the receiver operating characteristic curve of 0.916 for SAPS 3, 0.914 for SAPS II and 0.936 for APACHE II. All scores overestimated hospital mortality, with standardized mortality ratios of 0.81, 0.78 and 0.80 for SAPS 3, SAPS II and APACHE II, respectively. The calibration of all scores was poor as demonstrated by the Hosmer-Lemeshow goodness-of-fit H and C statistics, which were <0.05. The calibration of all scores was improved by first-level customization.

Conclusion: The SAPS 3 score provided excellent discrimination but poor calibration, as did the SAPS II and APACHE II scores in our critically ill mixed medical patients. The customization of the severity score improved the calibration performance, and thus this customization may serve as a tool for adjusting the equation of the score to predict mortality and quality assessment in intensive care units (ICUs).