OBJECTIVES: To assess the influence of infection on mortality in necrotizing pancreatitis.

METHODS: Eligible prospective and retrospective studies were identified through manual and electronic searches (August 2015). The risk of bias was assessed using the Newcastle-Ottawa Scale (NOS). Meta-analyses were performed with subgroup, sensitivity, and meta-regression analyses to evaluate sources of heterogeneity.

RESULTS: We included 71 studies (n = 6970 patients). Thirty-seven (52%) studies used a prospective design and 25 scored ≥5 points on the NOS suggesting a low risk of bias. Forty studies were descriptive and 31 studies evaluated invasive interventions. In total, 801 of 2842 patients (28%) with infected necroses and 537 of 4128 patients (13%) with sterile necroses died with an odds ratio [OR] of 2.57 (95% confidence interval [CI], 2.00-3.31) based on all studies and 2.02 (95%CI, 1.61-2.53) in the studies with the lowest bias risk. The OR for prospective studies was 2.96 (95%CI, 2.51-3.50). In sensitivity analyses excluding studies evaluating invasive interventions, the OR was 3.30 (95%CI, 2.81-3.88). Patients with infected necrosis and organ failure had a mortality of 35.2% while concomitant sterile necrosis and organ failure was associated with a mortality of 19.8%. If the patients had infected necrosis without organ failure the mortality was 1.4%.

CONCLUSIONS: Patients with necrotizing pancreatitis are more than twice as likely to die if the necrosis becomes infected. Both organ failure and infected necrosis increase mortality in necrotizing pancreatitis.