Background: The initial prognostic classification of a pathologic event has important implications for management, outcome, and resource utilization. A variety of prognostic classification methods have been developed, but none has thus far shown consistent predictive accuracy in individual patients. In an attempt to overcome some of the limitations encountered by earlier workers, we have developed a simplified system for evaluation and prognosis of acute pancreatitis.
Materials and methods: We studied retrospectively 306 patients with acute pancreatitis. Nonstandard statistical calculations were used to create risk indices for pancreatitis severity for all 13 variables which were included in a standard chart.
Results: The risk assessment rule's correct prediction is 81.32% (sensitivity) and 93.95% (specificity) for severe and mild pancreatitis, respectively. The total accuracy is 90.20%.
Conclusion: The results of this nonstandard small retrospective study seem promising. It should remain experimental until larger prospective studies confirm the usefulness of the method.
Copyright 2001 Academic Press.