Hemoconcentration and pancreatic necrosis: further defining the relationship

Abstract

Objectives: In the setting of acute pancreatitis, an admission hematocrit greater than or equal to 44% and/or a failure of hematocrit to drop at 24 hours have been reported as useful markers to predict subsequent necrosis. We aimed to validate the use of hemoconcentration as a marker to predict necrosis in adult patients presenting with acute pancreatitis.

Methods: Patients admitted to our medical center from 1990 to 2003 with a first presentation of acute pancreatitis were identified. Charts were abstracted for baseline demographic and clinical information, including admission and 24-hour hematocrit, and subsequent hospital course. Necrosis was determined based on computed tomography scan. We calculated the sensitivity, specificity, positive and negative predictive values (NPV) for different admissions, and 24-hour hematocrit levels in predicting the subsequent development of necrosis.

Results: Two hundred thirty patients were identified. Admission hematocrit (> or = 44%) was a poor predictor of subsequent necrosis with a sensitivity of 52.9%. The absence of hemoconcentration at admission or a drop in 24-hour hematocrit level was reliable in predicting that patients would not develop necrosis (NPV of 94.7% for hematocrit > or = 44%). Results were similar when we compared a range of admission and 24-hour hematocrit values.

Conclusions: In a community setting with low rates of necrosis, admission and 24-hour hematocrit levels were not helpful in predicting subsequent necrosis. The absence of admission hemoconcentration had strong NPV for necrosis. However, the actual clinical utility of this test to direct clinical decision making may be limited.