BACKGROUND AND AIMS: Despite intensive management, spontaneous bacterial peritonitis (SBP) is associated with poor prognosis especially in hospitalized patients. Therefore, the aim of this study was to determine prognostic factors for SBP-related in-hospital mortality, and to evaluate the predictive power of Child-Pugh (CP), model of end-stage liver disease (MELD), creatinine modified Child-Turcotte-Pugh (CrCTP), and integrated MELD (iMELD) for identifying the best score to predict mortality. METHODS: Predictors of SBP-related in-hospital mortality were assessed using regression analysis over 100 cirrhotic patients. Predictive abilities of CP, MELD, CrCTP, and iMELD were compared using the area under receiver operating characteristic curve (AUC). RESULTS: SBP-related in-hospital mortality was 22%. Age, serum creatinine, bilirubin, sodium, CrCTP, MELD, and iMELD were associated with mortality. Using AUC, CrCTP, and iMELD was significantly better than CP and MELD in predicting in-hospital mortality, where iMELD had the highest AUC (0.862). The cut-off with the best ability to predict in-hospital mortality was 43.5 for iMELD. CONCLUSION: Age, serum creatinine, bilirubin, and sodium were associated with SBP-related in-hospital mortality. The incorporation of these variables into CP and MELD significantly improves their predictive ability. iMELD followed by CrCTP provided useful prognostic information for critically ill patients with SBP.
BACKGROUND AND AIMS: Despite intensive management, spontaneous bacterial peritonitis (SBP) is associated with poor prognosis especially in hospitalized patients. Therefore, the aim of this study was to determine prognostic factors for SBP-related in-hospital mortality, and to evaluate the predictive power of Child-Pugh (CP), model of end-stage liver disease (MELD), creatinine modified Child-Turcotte-Pugh (CrCTP), and integrated MELD (iMELD) for identifying the best score to predict mortality. METHODS: Predictors of SBP-related in-hospital mortality were assessed using regression analysis over 100 cirrhotic patients. Predictive abilities of CP, MELD, CrCTP, and iMELD were compared using the area under receiver operating characteristic curve (AUC). RESULTS: SBP-related in-hospital mortality was 22%. Age, serum creatinine, bilirubin, sodium, CrCTP, MELD, and iMELD were associated with mortality. Using AUC, CrCTP, and iMELD was significantly better than CP and MELD in predicting in-hospital mortality, where iMELD had the highest AUC (0.862). The cut-off with the best ability to predict in-hospital mortality was 43.5 for iMELD. CONCLUSION: Age, serum creatinine, bilirubin, and sodium were associated with SBP-related in-hospital mortality. The incorporation of these variables into CP and MELD significantly improves their predictive ability. iMELD followed by CrCTP provided useful prognostic information for critically illpatients with SBP.