Abstract
BACKGROUND: Since its introduction, the Simplified Acute Physiology Score-3 (SAPS-3) has remained one of the most widely adopted scoring systems for mortality prediction in critically ill patients. This study aims to assess the performance of the SAPS-3 in a contemporary cohort of intensive care unit (ICU) patients in Mexico. METHODOLOGY: Data were obtained from a prospective database, covering admissions from September 2009 to May 2024. We evaluated the discriminative ability and calibration of SAPS-3 for mortality prediction using the area under the receiver operating characteristic curve (AUC) and the Hosmer-Lemeshow goodness-of-fit test. This study adhered to the guidelines for transparent reporting of multivariable prediction models for individual prognosis or diagnosis (TRIPOD). RESULTS: A total of 5,033 patients were included, with a mean age of 63 ± 18 years. In-hospital mortality was 1,211 (24.06%). The SAPS-3 score had a mean of 54 ± 19, corresponding to an estimated mortality of 1,560 (31.0%). For the entire cohort, the SAPS-3 demonstrated an AUC of 0.801 (95% confidence interval (CI), 0.787-0.815), and χ(2) = 31.2, P < 0.001. In the subset of patients cared for during the COVID-19 pandemic, the AUC was 0.791 (95% CI, 0.759-0.823), and χ(2) = 18.4, P < 0.02. CONCLUSIONS: Validation studies of SAPS-3 across diverse ICU populations consistently report good discriminative ability but limited calibration. This is the first study in Mexico with a large ICU sample, affirming that while SAPS-3 provides reliable discrimination, it tends to overestimate mortality risk.