Introduction: Effective triage of patients with acute pancreatitis is dependent on the ability to accurately predict a severe course. Predictors (e.g., APACHE II score of >8) have been tested against wide-ranging definitions of severity (prevalence, 15%-40%). To ensure uniformity in defining a severe course of acute pancreatitis, the Atlanta symposium of 1992 adopted all-encompassing criteria (local complications, systemic complications, need for surgery, or death).
Aims: To assess the prevalence of each Atlanta criteria for severe acute pancreatitis and to determine the sensitivity, specificity, and positive and negative predictive values of the APACHE II score as a predictor of these criteria for severe acute pancreatitis.
Methodology: We reviewed records of patients admitted to the University of Cincinnati Medical Center (Cincinnati, OH, U.S.A.) between 1994 and 1998 with acute pancreatitis. Exclusion criteria included referral from an outside hospital, immunocompromised state, and chronic pancreatitis.
Results: Seventy-four consecutive patients met our inclusion criteria. Ten patients (13.5%) had a severe course. Seven patients developed only local complications. Three patients had systemic complications. Pancreatic surgical intervention was required in four patients. No deaths occurred. An APACHE II score of >8 exhibited 50% sensitivity and 69% specificity (positive predictive value, 20%; negative predictive value, 89%). All patients with systemic complications and two of seven patients with only local complications had an APACHE II score of >8.
Conclusions: The prevalence of severity among our nonreferred patients with acute pancreatitis was less than previously reported. The APACHE II scoring system exhibited reasonable sensitivity in predicting systemic complications and/or the need for surgery, with a low positive predictive value. This most certainly is a function of the low pretest probability of severe pancreatitis. Future studies attempting to identify predictive systems that triage patients in a more cost-effective manner should restrict their analysis to Atlanta criteria other than local complications.