Abstract
BACKGROUND: Blunt pancreatic injury is frequently managed nonoperatively in children. Nutritional support practices - either enteral or parenteral - are heterogeneous and lack evidence-based guidelines. We hypothesized that use of parenteral nutrition (PN) in children with nonoperatively managed blunt pancreatic injury would 1) be associated with longer hospital stay and more frequent complications, and 2) differ in frequency by trauma center type. METHODS: We conducted a retrospective cohort study using the National Trauma Data Bank (2007-2016). Children (≤18 years) with blunt pancreatic injury were included. Patients were excluded for duodenal injury, mortality <4 days from admission, or laparotomy. We compared children that received versus those that did not receive PN. Logistic regression was used to model patient characteristics, injury severity, and trauma center type as predictors for propensity to receive PN. Treatment groups were balanced using the inverse probability of treatment weights. Outcomes included hospital length of stay, intensive care unit days, incidence of complications and mortality. RESULTS: 554 children with blunt pancreatic injury were analyzed. PN use declined in adult centers from 2012 to 2016, but remained relatively stable in pediatric centers. Propensity-weighted analysis demonstrated longer median length of stay in patients receiving PN (14 versus 4 days, rate ratio 2.19 [95% CI: 1.97, 2.43]). Children receiving PN also had longer ICU stay (rate ratio 1.73 [95% CI: 1.30, 2.30]). There was no significant difference in incidence of complications or mortality. CONCLUSIONS: Use of PN in children with blunt pancreatic injury that are managed nonoperatively differs between adult and pediatric trauma centers, and is associated with longer hospital stay. Early enteral feeding should be attempted first, with PN reserved for children with prolonged intolerance to enteral feeds. LEVEL OF EVIDENCE: III, Retrospective cohort.