Only patients with predicted severe acute pancreatitis are randomised. Patients with mild acute pancreatitis are considered not eligible for randomisation because of their low risk to develop infectious complications. If such patients would be included, a very large number of patients would be needed to demonstrate a significant effect between the treated and the non-treated patients. To properly introduce the concept and potential of probiotic prophylaxis, the use of prophylactic antibiotics was discussed during the preparations of the study. About 30% of the hospitals participating in the study commonly used prophylactic antibiotics, once a patient was diagnosed with pancreatic necrosis. Because of the lack of evidence it was decided, even before the results of the most recent German trial were presented, not to administer prophylactic antibiotics in case of pancreatic necrosis without clinical suspicion of infected necrosis. When patients do receive antibiotics, for instance because of an urinary tract infection, effort is made to administer the study product with a 4-hour interval in order to minimise interference of the antibiotics with the probiotics.
The maximum of 72 hours between onset of symptoms and start of treatment was decided upon, based on the consideration that probiotics are expected to prevent infection of pancreatic necrosis. Therefore the probiotics should be administered prior to the stage that bacterial overgrowth starts and intraluminal bacteria migrate across the mucosal barrier. Experimental studies have shown that bacterial overgrowth occurs very early, within 24 hours after onset, in the course of acute pancreatitis and reduction of the bacterial load in the proximal small bowel by intraluminal antibiotics reduces the risk of infection of pancreatic necrosis [8, 33, 34].
The presence of pancreatic necrosis can only be detected reliably 5–7 days after onset . This is too late to effectively prevent bacterial overgrowth and translocation and therefore "predictive laboratory scores" are used as randomisation criteria and not diagnosis of pancreatic necrosis on CT scan. The scoring systems used are also simple and generally available. A major disadvantage though, is the limited positive predictive value and the high number of false positives [36, 37]. During interim-analysis the number of false positives will be calculated and the sample size may be adjusted.
All randomised patients will receive early enteral feeding by a jejunal feeding tube. Since patients with predicted severe pancreatitis would develop severe pancreatitis in only 50% of the cases, the fraction with a mild course would normally not receive a nasojejunal feeding tube. It is unavoidable to prevent this over treatment for patients with a mild course because the current scoring systems fail to select all patients with a severe course and early intervention is warranted.
The primary outcome parameter 'total of infectious complications', was chosen because it was shown in previous trials that also the number of pulmonary and urinary tract infections can be reduced by probiotics [19–21] This fits in with the concept that such complications are secondary to bacterial translocation.
All of the infectious complications are documented in the study period until the patient reaches one of the study-endpoints: infected necrosis, discharge or in-hospital death.