Background: Simplification of APACHE II scoring system in the prediction of the outcome in critically ill patients with perforative peritonitis can be a useful and a cheaper model than the standard APACHE II system. We tested APACHE II and SAPS I scoring systems and variables of arterial pH, pO2, pCO2 and HCO3, cholesterol and albumin in the prediction of the outcome in these patients. Patients and Methods: The prospective study involved 145 patients of both sexes with perforative peritonitis. The main outcome of this study was peritonitis-related death. APACHE II and SAPS I scoring systems were calculated on the admission (during the first 24 hours). Cutoff points were specified and all values greater than the cut-off points were taken to predict death. Sensitivity and specificity are graphically shown for the different values of cut-off points. They are presented with the ROC curve. Variables of arterial pH, pO2, pCO2 and HCO3 were tested with Feed-Forward Artificial Neural Network which had 4 hidden layers with 8 neurons in the layer. We used Levenberg-Marquardt method for training, and 16 variables for the entrance in the network. We tested correlation between cholesterol and albumin levels with the patient outcome. Results: APACHE II ROC curve demonstrated that its discriminatory ability was better than the SAPS ROC curve. The area under the curve was 0.86 for APACHE II score in comparison to 0.83 for SAPS score. This illustrated that APACHE II is significantly better (P [Med Arch 2009; 63(5.000): 253-255]

peritonitis, APACHE II, Surgical Intensive Care Unit