A critical review of antibiotic prophylaxis in severe acute pancreatitis.

The close association between infection and poor outcome in severe pancreatitis has led many investigators to hypothesize that antibiotic prophylaxis might reduce infection and thereby reduce mortality. However, despite this possible relationship, few studies of good quality have been performed in humans. Comprehensive searches using Medline and reviewing relevant published bibliographies of English-language human and experimental literature concerning acute pancreatitis or pancreatic tissue and antibiotic therapy or pharmacokinetics were conducted. Ample experimental evidence indicates that aminoglycosides penetrate pancreatic tissue poorly and that penetration of penicillins is variable, although the relevance of this is debatable, because most tissue that requires debridement in severe pancreatitis is necrotic peripancreatic retroperitoneal fat, not the pancreas itself. Although several animal studies suggest that antibiotic prophylaxis would be beneficial in severe pancreatitis, two recent randomized studies of intravenous antibiotics in humans provide conflicting data. There are insufficient data to recommend the use of selective digestive decontamination. Some justification exists for the use of intravenous antibiotic prophylaxis in severe pancreatitis, but the data are insufficient to mandate prophylaxis or to elevate it to the standard of care. If chosen, prophylaxis with the combination of a fluoroquinolone plus metronidazole, or monotherapy with a carbapenem antibiotic, would be most appropriate. Several other questions-including the minimum degree of severity that will benefit, the validity of endpoints other than mortality, and reduction of the need for surgical drainage-require additional trials.