RESUMO
PURPOSE: This study was undertaken to establish a model to predict the post-operative mortality for emergency surgeries. METHODS: A regression model was constructed to predict in-hospital mortality using data from a cohort of 479 cases of emergency surgery performed in a Japanese referral hospital. The discrimination power of the current model termed the Calculation of post-Operative Risk in Emergency Surgery (CORES), and Portsmouth modification of the Physiological and Operative Severity Score for the enUmeration of Mortality and morbidity (P-POSSUM) were validated using the area under the receiver operating characteristic curve (AUC) in another cohort of 494 cases in the same hospital (validation subset). We further evaluated the accuracy of the CORES in a cohort of 1,471 cases in six hospitals (multicenter subset). RESULTS: CORES requires only five preoperative variables, while the P-POSSUM requires 20 variables. In the validation subset, the CORES model had a similar discrimination power as the P-POSSUM for detecting in-hospital mortality (AUC, 95 % CI for CORES: 0.86, 0.80-0.93; for P-POSSUM: 0.88, 0.82-0.93). The predicted mortality rates of the CORES model significantly correlated with the severity of the post-operative complications. The subsequent multicenter study also demonstrated that the CORES model exhibited a high AUC value (0.85: 0.81-0.89) and a significant correlation with the post-operative morbidity. CONCLUSIONS: This model for emergency surgery, the CORES, demonstrated a similar discriminatory power to the P-POSSUM in predicting post-operative mortality. However, the CORES model has a substantial advantage over the P-POSSUM in that it utilizes far fewer variables.
