Eight RCTs (n=502) were included in the meta-analysis. Two trials received a Jadad score of 5, three scored 3, two scored 2 and one scored 1. Six studies were single-blinded. Two studies were double-blinded.
Overall, antibiotic treatment did not significantly improve mortality rates in severe acute pancreatitis patients in comparison with placebo (RR 0.76, 95% CI 0.49 to 1.16). No significant heterogeneity was measured between the studies (I2=8.8%).
Antibiotic treatment did not significantly protect against infected necrosis (seven studies, n= 429) or surgical interventions (seven studies, n=476). Antibiotic treatment significantly reduced non-pancreatic infections (RR 0.6, 95% CI 0.44 to 0.82; six studies, n=407). However, there was significant heterogeneity (I2= 50.1%). Relative risk reduction (RRR) was 40% (95% CI 18% to 56%). Absolute risk reduction (ARR) was 15% (95% CI 6% to 23%). Number needed to treat was 7 (95% CI 4 to 17).
Sensitivity analysis of β-lactam antibiotics (four studies, n=305) found that they did not protect against mortality, infected necrosis and surgical interventions, but they provided significant protection against non pancreatic infections (RR 0.38, 95% CI 0.22 to 0.68, RRR 62%, 95% CI 32 to 78, ARR 12%, 95% CI 3 to 20, NNT=8, 95% CI 5 to 33).
Subanalysis of high-quality studies (five studies, n=381) found that use of antibiotics in severe acute pancreatitis patients did not protect against mortality, infected necrosis, surgical interventions and non-pancreatic infections. Publication bias was found for the five high-quality studies through uses of funnels plots, Egger tests (p=0.031) and Begg and Mazumdar tests (p=0.027).