Predictive analysis of the response to diuretics in patients with hepatic cirrhosis and ascites

Our aim was to develop and validate a prognostic model to predict the evolution of cirrhotic patients admitted to the hospital because of ascites and treated with diuretics. Two hundred and two patients were evaluated. After collection of clinical and laboratory data, a prognosis of the response to diuretics was given by one of us. After univariate analysis, 37 parameters were used to construct a database. A new prognosis was obtained for each patient comparing his/her data with the database, with the help of a computer and using Bayes' theorem. Gender: female, poor general status, disturbed consciousness, increased serum glucose, increased BUN, low prothrombin time, increased bilirubin, leucocytosis, thrombopenia and low urine sodium were associated with a poor response to diuretics. Sensitivity and specificity of clinical and computer prognosis were similar. Computer prognosis allowed the classification of patients in groups of risk which showed a different evolution and improved the accuracy of the clinical prognosis. We conclude that the objective analysis of clinical and laboratory data is useful to improve the accuracy of the prediction of the evolution in cirrhotic patients with ascites treated with diuretics.