4694 Acute pancreatitis associated with hiv-infection: incidence and predictors of severity.

Abstract

The increased incidence of acute pancreatitis (AP) in HIV-infected patients is ascribed primarily to complications of medical therapy and opportunistic infections. Recent studies conflict in their assessment of the incidence of severe pancreatitis and the best method to assess the severity of AP in these patients. Our aim was to determine the incidence of severe pancreatitis in HIV- infected patients and to evaluate the accuracy of standard predictors of severity in the course of the disease. Methods: All patients known to be HIV-positive who had a documented episode of AP between 01/89 to 09/99 were identified through a database. AP was defined as an amylase of >200 U/L and 1) a symptom complex consistent with the disease or 2) radiographic evidence of the disease. Parameters required for predicting the severity of pancreatitis (Ranson, Modified Glasgow and APACHE II) plus demographic information, comorbid disease, AIDS-defining illnesses and CD4 counts were obtained. The etiology was determined from medication histories, appropriate investigations and lastly, reported opinion of the attending physician. Episodes of severe pancreatitis were defined by the 1992 Atlanta criteria. Results: Of 2300 patients with documented HIV-disease, 60 cases of AP were identified, of which 39 have beenreviewed. 22 (56%) patients had a history of an AIDS-defining illness (ADI) and a further 9 (23%) patients had CD4 counts of less than 200 cells/μL with no ADI. The etiology of AP was most commonly medication related (43%), followed by idiopathic (23%) and alcohol (19%). Of the 39 episodes of AP, 8 cases (5 deaths) were deemed severe by the Atlanta criteria. Ranson criteria were found to be 50% sensitive and 56% specific overall but 100% sensitive and 33% specific when only those patients with 9 of 11 parameters available were analyzed. The Modified Glasgow score was poor in predicting severity with a sensitivity of 37% and a specificity of 84%. This improved to 75% and 80%, respectively, with exclusion of patients with 5 or less of the 8 criteria required. The APACHE II score had the best overall accuracy with a 75% sensitivity and 83% specificity (88% of the parameters were recorded). There was no correlation between the CD4 count, the duration of HIV-disease or the presence of ADI and the severity of the episode of pancreatitis. Conclusions: In HIV-infected patients presenting with AP: 1) The incidence of severe AP is common, occurring in almost 20% of cases 2) The most accurate grading system to predict the severity of the attack is the APACHE II.