Hemoconcentration Alone Is an Unreliable Predictor of Mortality in Acute Pancreatitis

Abstract

Purpose: To examine whether hemoconcentration (HC) at admission predicts in-hospital mortality in acute pancreatitis (AP). To determine whether a new definition of intravascular volume depletion [Hgb + BUN] is a better predictor for in-hospital mortality. Methods: Patient data was generated from the Cardinal Health Research Database, a large population dataset. All cases from Jan 2004–Dec 2004 with principal diagnosis ICD9-CM 577.0 (AP) were included. Two cohorts of patients were identified: transferred vs. non-transferred pts. HC was defined as elevated hgb > 14.6 g/dL (hct 44%). The combination of elevated hgb > 14.6 g/dL and BUN > 25 mg/dL (Hgb + BUN) was used as a marker for intravascular volume depletion. Comparison was made between HC and our newly defined marker of intravascular volume depletion for prediction of in-hospital mortality in the two cohorts. Results: There were a total of 202 transferred and 17,852 non-transferred cases of AP. Overall mortality was 5.5% vs 1.2% in transferred vs. non-transferred pts respectively (P < 0.001 χ2). HC was a significant risk factor for mortality only among transferred pts. Combined (Hgb + BUN) was a significant predictor of mortality in both cohorts. Admission intravascular volume depletion as assessed by our new criteria was significantly more common among transferred vs non-transferred pts. (9.4% vs. 2.8%, P < 0.001 χ2). Conclusion: Hemoconcentration alone is an unreliable predictor of mortality in AP. Hgb + BUN was a more robust predictor of mortality for both transferred and non-transferred pts. These findings help explain discordant results from previous studies of hemoconcentration in AP.Table: Mortality Comparisons