Objective Lack of a universal case-definition hinders (NEC) surveillance and clinical trials. Bell’s staging, although widely used, was devised to guide surgical management after diagnosis¹. We aimed to develop an evidence-based NEC case-definition.

Methods We used the National Neonatal Research Database, that holds data extracted from the electronic records of babies admitted to NHS neonatal units, to obtain variables for infants receiving a laparotomy between November 2011–May 2013 (gestation, clinical signs (abdominal distension, tenderness, discolouration, mass, increased and/or bilious aspirates, bloody and/or mucousy stools), radiological signs (pneumatosis, air in the liver, pneumoperitoneum, fixed loop, gasless), confirmed NEC (at surgery or post-mortem). We developed case-definitions using logistic regression and classification trees², and calculated sensitivity, specificity and area under the receiver operating characteristic curve (AUC). Interim validation was conducted using a separate dataset covering June–December 2013.

Results We identified 142 infants from 40 neonatal units; 60 had confirmed NEC. The logistic regression definition comprised “pneumatosis OR two or more of (a) abdominal discolouration, (b) gasless X-ray (c) pneumoperitoneum” (sensitivity 0.72, specificity 0.88, AUC 0.8) and classification tree definition “pneumatosis OR abdominal discolouration OR [pneumoperitoneum AND aspirates]” (sensitivity 0.85, specificity 0.78, AUC 0.82). Validation showed logistic regression: sensitivity 0.64, specificity 0.95, AUC 0.85; classification tree sensitivity 0.67, specificity 0.91, AUC 0.79.

Conclusion Our interim analysis shows that the two methods derived similar definitions identifying a similar proportion of cases; pneumatosis was the strongest predictor. The final analysis will include data from June 2013-April 2014 and will investigate gestation-specific effects.

References

1. Bell MJ, Ternberg JL, Feigin RD , et al. Neonatal necrotizing enterocolitis. Therapeutic decisions based upon clinical staging. Ann Surg. 1978;187:1–7 Online First: 1978/01/01]

2. Crichton NJ, Hinde JP, Marchini J. Models for diagnosing chest pain: is CART helpful? Stat Med. 1997;16:717–27 Online