Antibiotic therapy for prophylaxis against infection of pancreatic necrosis in acute pancreatitis

Acute pancreatitis is a common abdominal emergency with no specific treatment. Pancreatic necrosis may complicate severe attacks, detectable by computed tomography (CT). Necrosis can become infected, making surgical intervention necessary and increasing mortality to more than 40%. Experimental studies suggest that antibiotic therapy may prevent infection, but could promote resistance and fungal infection. To determine the effectiveness and safety of prophylactic antibiotics in acute pancreatitis complicated by pancreatic necrosis. The Cochrane Library (Issue 1, 2006), MEDLINE (January 1966-December 2005), EMBASE (January 1980-December 2005) and CINAHL (January 1982-December 2005) were searched. We also examined Conference proceedings. Randomised controlled trials (RCTs) comparing antibiotics versus placebo in acute pancreatitis with CT proven necrosis were sought using a detailed search strategy without linguistic limitation. RCTs. Initial searching was undertaken in November 2001. Latest update: December 2005. Two reviewers extracted data independently for rates of primary end-points: mortality and pancreatic infection rates. Secondary end-points included: non pancreatic infection and operative rates. Adverse events: antibiotic resistance and fungal infections. Subgroup analyses: antibiotic regimen. Five evaluable studies randomised 294 patients. Analysis suggested significantly less mortality with therapy (6%) versus controls (15.3%), odds ratio 0.37 (95% CI 0.17, 0.83). Infected pancreatic necrosis rates were not significantly different (therapy 20%, controls 27.8%), odds ratio 0.62 (95% CI 0.35, 1.09), and neither were operative treatment rates or non-pancreatic infection rates. Fungal infections were not significantly different at 4% with therapy versus 4.9% in controls, odds ratio 0.83 (95% CI 0.30, 2.27). There were no evaluable data on antibiotic resistance. Sub-group analysis was performed for antibiotic regimen: beta lactam (192 patients), and quinolone plus imidazole (102 patients). With beta lactam prophylaxis there was significantly less mortality (6.3%) versus controls (16.7%), odds ratio 0.34 (95% CI 0.13, 0.91), and infected pancreatic necrosis (15.6%) versus (29.2%) in controls, odds ratio 0.41 (95% CI 0.20, 0.85), but there were no significant differences in operative treatment rates or non-pancreatic infections. No significant differences were seen with quinolone plus imidazole. Antibiotic prophylaxis appeared to be associated with signi