Prediction of the severity of acute pancreatitis.

We conducted a prospective study to validate our previous finding that serum urea and plasma glucose levels on admission could predict the outcome of acute pancreatitis. Forty-two (24%) of 176 patients developed complications related to the attack of acute pancreatitis and were classified as having severe disease. By logistic regression analysis of 17 admission parameters, serum urea and plasma glucose levels were again the factors with independent significance in defining the outcome. By adopting the same cutoff levels as in our previous study (serum urea level greater than 7.4 mmol/L and plasma glucose level greater than 11.0 mmol/L), and the presence of either factor above the cutoff level as indicative of severe disease, the sensitivity of prediction was 79%, specificity 67%, and overall accuracy 70%. All the deaths were correctly predicted by this urea/glucose criteria. The overall accuracy was also found to be comparable with those of the APACHE II (cutoff level greater than 11) and Ranson's scoring systems. We conclude that the simple prognostic criteria for acute pancreatitis were validated; these criteria have the potential to stratify risk rapidly at the time of admission for patients who might benefit from an aggressive interventional protocol.