Background & aims: Routine laboratory tests that reflect intravascular volume status can play an important role in the early assessment of acute pancreatitis (AP). The objective of this study was to evaluate accuracy of serial blood urea nitrogen (BUN) versus serial hemoglobin (Hgb) measurement for prediction of in-hospital mortality in AP.
Methods: We performed an observational cohort study on data from 69 US hospitals from January 2003 to December 2006. Repeated measures analysis was used to examine the relationship between early trends in BUN and Hgb with respect to mortality. Multivariate logistic regression was used to evaluate the impact of admission BUN, change in BUN, admission Hgb, and change in Hgb on mortality. Time-specific receiver operating characteristic curves and multivariable logistic regression compared accuracy of BUN, Hgb, and additional routine laboratory tests.
Results: BUN levels were persistently higher among nonsurvivors than survivors during the first 48 hours of hospitalization (F-test; P < .0001). No such relationship existed for Hgb (F-test; P = .33). For every 5-mg/dl increase in BUN during the first 24 hours, the age- and gender-adjusted odds ratio for mortality increased by 2.2 (95% confidence limits, 1.8, 2.7). Of the 6 routine laboratory tests examined, BUN yielded the highest area under the concentration-time curve (AUC) for predicting mortality at admission (AUC = 0.79), 24 hours (AUC = 0.89), and 48 hours (AUC = 0.90). Combining admission BUN and change in BUN at 24 hours produced an AUC of 0.91 for mortality.
Conclusion: In a large, hospital-based cohort study, we identified serial BUN measurement as the most valuable single routine laboratory test for predicting mortality in AP.