The 22/11 Risk Prediction Model: A Validated Model for Predicting 30-Day Mortality in Patients With Cirrhosis and Spontaneous Bacterial Peritonitis

Abstract

OBJECTIVES: Clinicians do not have a validated tool for estimating the short-term mortality associated with spontaneous bacterial peritonitis (SBP). Accurate prognosis assessment is important for risk stratification and for individualizing therapy. We aimed therefore to develop and validate a model for the prediction of 30-day mortality in SBP patients receiving standard medical treatment (antibiotics and if indicated by guidelines, intravenous albumin therapy). METHODS: We retrospectively identified SBP patients treated at a tertiary care center between 2003 and 2011 (training set). Multivariate regression modeling and receiver operating characteristic (ROC) curves were utilized for statistical analysis. An external data set of 109 SBP patients was utilized for validation. RESULTS: Of the 184 patients in the training set, 66% were men with a median age of 55 years, a median MELD (Model for End-Stage Liver Disease) score of 20, and a 30-day mortality of 27%. Peripheral blood leukocyte count >= 11 x 10(9) cells/l (odds ratio (OR) 2.5; 95% confidence interval CI: 1.2-5.2) and MELD score >= 22 (OR 4.6; 95% CI: 2.3-9.6) were independent predictors of 30-day mortality. Patients with neither, one, or both variables had 30-day mortality rates of 8%, 32%, and 52%, respectively. The findings in the validation set mirrored the training set. CONCLUSIONS: In cirrhotic patients with SBP receiving standard therapy, MELD score >= 22 and peripheral blood leukocyte count >= 11 x 10(9) cells/l are validated independent predictors of mortality. The mortality in a patient without either poor prognostic variable is <= 10% and with both variables is <= 50%. Trials aiming to reduce mortality should target patients in the moderate-risk to high-risk groups.