Factors associated with mortality in patients with infected pancreatic necrosis: the "surgery effect".

Severe acute pancreatitis complicated by infection is associated with high mortality. Invasive treatment is indicated in the presence of infected (suspected) pancreatic and/or peripancreatic necrosis (IPN) in the absence of response to intensive medical support. Step-up approach (SUA) has been demonstrated to lower complication rate compared to upfront open surgery. However, this approach has not been associated with lower mortality, and no factors have been studied that could help to identify the high risk patients. In this study, we aimed to analyse those factors associated with mortality following the invasive treatment of IPN, focusing on the role of surgical necrosectomy. A retrospective and observational study based on a multicentre prospective database was conducted. The database was coordinated by the Hospital General Universitario de Alicante, Spain and the Spanish Association of Pancreatology. Demographics, clinical data, and laboratory and imaging findings were collected. Atlanta 2012 criteria were considered to classify acute necrotizing pancreatitis and for the definition of IPN. Step-up approach was used in all centres with the intention of avoiding surgery whenever possible. Surgical necrosectomy was performed by open approach. From January 2013 to October 2014, a total of 1655 patients with the diagnosis of acute pancreatitis were included in our database. 1081 were recruited for the final analysis. Out of them, 205 (19%) were classified into acute necrotizing pancreatitis. 77 (8.3%) patients underwent invasive treatment of INP and were included in our study. Overall mortality was 29.9%. Upfront endoscopic or percutaneous drainage was performed in 60 (77.9%) patients and mortality was 26.6%. Out of 60, 22 (36.6%) patients subsequently received rescue surgery; mortality in rescue surgery group was 18.3%. Upfront surgery was carried out in 17 (22.1%) patients; mortality in this group was 41%. At univariate analysis, surgical necrosectomy, extrapancreatic infection, immunosuppression and de-novo haemodialysis were associated with mortality. At multivariate analysis, only surgical necrosectomy was significantly associated with mortality (p = 0.002 OR 3.89). Surgical approach for IPN is associated with high mortality rate. However, these data should be interpreted with caution, since we are not able to assess whether this occurs due to the need of surgery as the only resort when the other approaches are not feasible or fail.