Objective
to compare existing outcome prediction models and create a novel model to predict death or intestinal failure (IF) in infants with surgical necrotizing enterocolitis (NEC).
Study design
A retrospective, observational cohort study conducted in a two-campus health system in Atlanta, GA from September 2009 to May 2015. Participants included all infants <=37 weeks’ gestation with surgical NEC. Logistic regression was used to model the probability of death or IF, as a composite outcome, using preoperative variables defined by specifications from three existing prediction models: American College of Surgeons National Surgical Quality Improvement Program Pediatric (NSQIP-P), Score for Neonatal Acute Physiology Perinatal Extension (SNAPP-II), and Vermont Oxford Risk Adjustment Tool (VON-RA). A novel preoperative hybrid prediction model was also derived and validated against a patient cohort from a separate campus.
Results
Among 147 patients with surgical NEC, discrimination in predicting death or IF was highest with NSQIP-P (Area under curve [AUC] 0.84 [95% CI 0.77–0.91]) when compared with SNAPPE-II (0.60; 95% CI 0.48–0.72) and VON-RA (0.74; 95% CI 0.65–0.83). A hybrid model was developed using four pre-operative variables: 1 minute Apgar score, inotrope use, mean blood pressure, and sepsis. The hybrid model AUC was 0.85 (95% CI 0.78–0.92) in the derivation cohort and 0.77 (95% CI 0.66–0.86) in the validation cohort.
Conclusions
Pre-operative prediction of death or IF among infants with surgical NEC is possible using existing prediction tools and, to a greater extent, using a newly proposed 4 variable hybrid model.