Infection rate among nutritional therapies for acute pancreatitis: A systematic review with network meta-analysis of randomized controlled trials

Background
Infection in acute pancreatitis (AP) is associated with nutritional therapies including naso-gastric (NG), naso-jejunal (NJ), and total parenteral nutrition (TPN). To examine infections among NG, NJ, TPN, and no nutritional support (NNS) in treating patients with AP.
Methods
The investigators completed comprehensive search in the Cochrane library, EMBASE, PubMed, Web of Science, and ClinicalTrials.gov without restriction on language and publication date before January 21, 2019. They also searched the reference lists of relevant studies for randomized controlled trials (RCTs) comparing NG, NJ, TPN, and NNS among patients with AP. Quantitative synthesis was conducted in a contrast-based network meta-analysis. To clarify effects, a network meta-analysis was conducted to calculate the surface under the cumulative ranking curve (SUCRA). Beside of overall infections, the event rates of infected pancreatic necrosis, bacteremia, line infection, pneumonia, urinary tract infection, and other types of infections were measured.
Results
The network meta-analysis of 16 RCTs showed that NJ had significantly lower overall infection rates compared with TPN (risk ratio: 0.59; 95% confidence interval: 0.38, 0.90); and NG had a larger effect size and higher rank probability compared with NJ, TPN, and NNS (mean rank = 1.7; SUCRA = 75.8). TPN was the least preferred (mean rank = 3.2; SUCRA = 26.6).
Conclusions
NG and NJ may be preferred therapies for treating patients with AP. Clinicians may consider NG as a first-line treatment for patients with AP (including severe AP) and even in patients receiving prophylactic antibiotics. In addition, we found that NNS should be avoided when treating patients with severe AP.