Background/aims: This study aims to determine predictability of death in acute pancreatitis at a secondary-care hospital in Germany.
Methods: This study is part of an ongoing study on the epidemiology of acute pancreatitis and covers 368 patients with a first attack of acute pancreatitis in Lüneburg county from 1988 to 1999. Early and late mortality were defined as < or = 1 weeks and > 1 week after admission. The following parameters were used to establish on admission likelihood of death: admission within 24 h or later with an acute attack, abdominal tenderness, signs of peritonitis, amylase and lipase in serum, leukocytes, hematocrit, potassium, sodium, calcium, creatinine after rehydration, blood glucose, bilirubin, serum glutamate-oxalacetate transaminase (SGOT), serum lactate dehydrogenase (SLDH), arterial pO2, APACHE II score, Ranson and Imrie scores.
Results: Of the 368 patients 17 (5%) died, 7 early because of multiple organ failure and 10 late because of septic complications. Mortality rates in interstitial and necrotising pancreatitis were 3 and 17%, respectively. Only an elevated serum creatinine (> 2.0 mg/dl) and a blood glucose > 250 mg significantly correlated with mortality. Ranson and Imrie scores were also significantly correlated with mortality; however, they were not obtained on admission, but only after 48 h. In univariate analysis, APACHE II score > or = 6 on admission and lipase > 1,000 U/l on admission provided a high sensitivity and negative predictive value for early and late mortality patients.
Conclusion: Approximately half of the deaths in acute pancreatitis occur because of multiple organ failure or septic complications. New approaches have to be found to counteract these severe complications. A fatal outcome may be predicted by simple laboratory parameters such as a high serum creatinine and blood glucose. An APACHE II score > or = 6 and a lipase level on admission > or = 1,000 U/l indicate severe pancreatitis.