A series of randomised controlled trials (RCTs) has led to confusion as to the precise role of antibacterial prophylaxis against infection of pancreatic necrosis (PN) in acute pancreatitis. Published RCTs comparing antibacterial therapy versus placebo in patients with CT proven pancreatic necrosis were sought. Seven were found (Jan 1993-Sep 2007): five evaluated beta-lactams (n=302) and two quinolone/imidazole (n=102). None was adequately powered, end point reporting varied and overall study quality was variable. Three studies were double-blinded, but the latest only provided extractable data on PN infection rates. Meta-analysis was performed using ‘RevMan 4.2’ software (Update, Oxon UK).Mortality was significantly lower after therapy (6.8%) versus control (15%), odds ratio (OR) 0.44 (95% CI 0.21-0.91), p 0.03, in 6 fully evaluable studies (n=322). However infected PN rates did not differ significantly in all 7 studies. Non-pancreatic sepsis was significantly lower after therapy (20%) versus controls (30%), OR 0.54 (95% CI 0.30, 0.98), p 0.04, although only 4 studies (n=236) provided these data. Four evaluable beta-lactam studies showed a trend to lower mortality (7.4%) versus controls (16%), but this was non-significant and no significant survival differences were found in two quinolone/imidazole studies. Beta-lactam therapy showed a trend towards less infected PN (16.8%) versus controls (24%), OR 0.61 (95% CI 0.34, 1.08), p 0.09, but there was a non-significant trend to higher infected PN rates with quinolone/imidazole therapy. Non-pancreatic infections were significantly lower after beta-lactams (15.4%) versus controls (28%) in 3 evaluable studies, OR 0.43 (95% CI 0.20, 0.92) p=0.03. Only one quinolone/imidazole study reported non-pancreatic infections which were not significantly different.This results suggest that antibacterial prophylaxis does reduce overall mortality, but not by reducing infection of PN. The mechanism is unclear, but contrary to the rationale for these RCTs, prophylaxis may be protective against fatal non-pancreatic infections, such as respiratory infection. More data are awaited from the most recent study, but we conclude that further, adequately powered, double-blinded studies, perhaps targeting beta-lactam agents, are required to elucidate the mechanism for reduced mortality.REFERENCES:Pederzoli et al SGO 1993;176(5):480-483.Sainio et al Lancet 1995;346:663-667.Nordback et al J GI Surg 2001;5(2):113-118.Schwarz et al Dtsch Med Wschr 1997;122:356-361.Isenmann et al Gastroenterology 2004;126:997-1004. Rokke et al Sc J Gastro 2007;42:771-776.Dellinger et al Ann Surg 245:674-683.