Systematic Review of Randomised Clinical Trials in Acute Pancreatitis

Introduction: This study aimed to conduct a systematic review of RCTs in relation to the management of acute pancreatitis with emphasis on risk of bias assessment (ROB), sample size calculation and SPIN. Methods: A systematic search of PUBMED, MEDLINE, CENTRAL databases was performed including RCTs across 3 predefined time periods (<1996,1996-2008, > 2008). RCT quality was analysed with Cochrane ROB-II, SPIN and sample size recalculation parameters. Results: 266 RCTs with 23,472 patients were included. The average RCTs per year for P1, P2, P3 were 1.3, 6.1 and 10.9. The clinical domains were nutrition (27%), drug treatment (21%), pancreatic necrosis management (13%), antibiotic prophylaxis (10%) and fluid management (7%). The ROB assessment showed low, some concerns and high in 20%, 24%, 55% of trials. Bias from selective results was lowest in P1 (9.5%), (p<0.005). Low bias in the measurement of outcomes was highest in P2 (76%) followed by P3 (82%) and P1(90%) (p=0.005). Bias arising from randomisation, allocation concealment and data completeness showed no association with time periods. Self-calculated sample sizes were reported in 40% of trials and significantly increased over time (P3 46% P<0.05). SPIN was identified in 26% with emphasis on non-significant trends in primary endpoint (31% of RCTs with SPIN). Conclusion: The quantity and quality of RCTs in AP have increased with time. Misleading reporting was identified in minority of trials (26%). ROB-II and sample size recalculations showed a trend towards more transparent reporting of methodology and SPIN was identified in third of studies.