Purpose: (1) To test the hypothesis that the clinical features of patients with acute pancreatitis could be used to construct a new prognostic staging system, and (2) to compare the new system with the results of the existing Ranson system based on age and laboratory data.
Patients and methods: We obtained an inception cohort of 162 persons with 176 episodes of acute pancreatitis admitted to Yale-New Haven Hospital from January 1, 1987, to March 31, 1989.
Results: The main adverse outcome events, death and/or complications, occurred in 22% of episodes. Acute pancreatitis severity was classified in three stages, reflecting the presence of only primary features (no ileus or peritonitis), or the occurrence of secondary features that could be mild (mild ileus) or severe (severe ileus and/or peritonitis). In these 3 stages, the respective rates for outcome events were 13% (17 of 133), 43% (13 of 30), and 62% (8 of 13). The severity of comorbidity was classified according to Charlson's comorbidity index. For the 3 comorbidity stages, the outcome event rates were: minimal: 13% (12 of 92); intermediate: 22% (11 of 49); and severe: 43% (15 of 35). When the acute pancreatitis severity stages and comorbidity stages were conjoined to form the composite staging system, the corresponding outcome event rates were stage I: 8 of 107 (7%); stage II: 24 of 60 (40%); and stage III: 6 of 9 (67%). The composite staging system produced sharper gradients for the outcome events than Ranson's system, and was more readily applicable to all patients, particularly when the requisite special laboratory tests had not been ordered.
Conclusions: The clinical and comorbid features of patients with acute pancreatitis can be used to construct a clinically "sensible" composite staging system. The stages are easy to use, require no additional or special laboratory tests, and predict more accurately than Ranson's system.