Preventing severe necrotizing enterocolitis: Propensity score analysis of interventions associated with surgical NEC or NEC-associated death.

Purpose: To determine what modifiable interventions used in the neonatal intensive care unit (NICU) are associated with severe necrotizing enterocolitis (NEC) requiring surgical intervention.

Methods: A retrospective review of patients treated for NEC at a tertiary hospital from 1991 to 2016 was performed. Patient characteristics were used to calculate propensity scores for likelihood of exposure to seven interventions: enteral feeds, use of glucocorticoids, nonsteroidal anti-inflammatory drugs (NSAIDs), antacids, antibiotics, or umbilical arterial (UAC) and venous catheters (UVC). Conditional logistic regression was used to compare the odds of having surgical NEC if exposed to each treatment.

Results: We included 195 NEC patients: 69 severe NEC managed with surgery and 126 non-severe NEC managed medically. After propensity score matching based on birth characteristics, exposure to glucocorticoids (OR 5.21, 95%CI: 1.62, 16.70), NSAIDs (OR 4.87, 95%CI: 1.67, 14.17), UVC (OR 2.53 (95%CI: 1.19, 5.73), and UAC (OR 4.91, 95%CI: 2.12, 11.37) were associated with surgical NEC in separate conditional logistic regression analyses. Including these treatments in a second round of propensity matching and conditional logistic regression revealed that glucocorticoids (OR 2.99, 95%CI: 1.01, 8.88), NSAIDs (OR 3.97, 95%CI: 1.41, 11.19), UVC (OR 3.07, 95%CI: 1.46, 6.48), and UAC (OR 5.10, 95%CI: 2.10, 12.36) were still associated with surgical NEC.

Conclusion: After controlling for birth confounders and common NICU supportive interventions, use of glucocorticoids, NSAIDs and umbilical catheters independently increased the odds of developing severe NEC requiring surgical intervention.

Level-of-evidence Rating: Case-control, Level III evidence.