Computer-assisted analysis of clinical and laboratory data in acute pancreatitis: early formulation of a predictive...

The results of a retrospective study of 111 cases of acute pancreatitis are reported. Based on diagnostic investigations and clinical outcome, patients have been divided into three groups: oedematous, limited and extensive necrotizing pancreatitis. The step-wise statistical analysis of clinical and laboratory parameters at the time of admission related to the pathological findings has allowed the formulation of a severity score to be applied to every new case of acute pancreatitis. By discriminant function coefficients a concordance percentage of actual and predicted classification has been obtained in over 90% of the cases. The presence of shock after 12 hours of intensive treatment, the hemorrhagic appearance of the peritoneal fluid and tachypnea are the variables with highest discriminating power. In view of the high concordance percentage between actual and predicted classification results, this predictive score could be applied, to every new case of acute pancreatitis at admission.