Meta-analysis and trial sequential analysis of randomized controlled trials comparing aggressive versus non-aggressive intravenous fluid therapy in acute pancreatitis: an insight into the existence of type 2 error.

Abstract

Background and aim: We aimed to evaluate comparative outcomes of aggressive versus non-aggressive intravenous fluid (IVF) therapy in patients with acute pancreatitis.

Methods: A systematic search of electronic data sources and bibliographic reference lists were conducted. All randomized controlled trials (RCTs) reporting outcomes of aggressive versus non-aggressive IVF therapy in acute pancreatitis were included and their risk of bias were assessed. Effect sizes were determined for overall mortality, systemic inflammatory response syndrome (SIRS), sepsis, respiratory failure, pancreatic necrosis, severe pancreatitis, clinical improvement, AKI, and length of stay using random-effects modeling. Trial sequential analysis was conducted to determine risk of types 1 or 2 errors.

Results: We included 10 RCTs reporting 993 patients with acute pancreatitis who received aggressive (n = 475) or non-aggressive (n = 518) IVF therapy. Aggressive IVF therapy was associated with significantly higher rate of sepsis (OR: 2.68, P = 0.0005) and longer length of stay (MD: 0.94, P < 0.00001) compared with the non-aggressive approach. There was no statistically significant difference in mortality (RD: 0.02, P = 0.31), SIRS (OR: 0.93, P = 0.89), respiratory failure (OR: 2.81, P = 0.07), pancreatic necrosis (OR: 1.98, P = 0.06), severe pancreatitis (OR: 1.31, P = 0.38), clinical improvement (OR: 1.12, P = 0.83) or AKI (OR: 1.06, P = 0.91) between the two groups. Sub-group analysis demonstrated higher morbidity and mortality associated with the aggressive approach in more severe disease. Trial sequential analysis detected risk of type 2 error.

Conclusions: Aggressive IVF therapy may be associated with higher morbidity in patients with acute pancreatitis compared with the non-aggressive approach, particularly in patients with more severe disease. It may also prolong length of hospital stay. The available evidence is subject to type 2 error indicating the need for adequately powered RCTs.