Presented at the Neonatal Society 2016 Summer Meeting.
Battersby C, Longford N, Costeloe K, Modi N and the UK Neonatal Collaborative NEC Study Group
Imperial College London, UK
Background: Reliable surveillance requires a consistent case-definition. Gestational age affects both the risk of NEC and clinical presentation. The aim of this study was to develop a gestational age specific case definition suitable for NEC surveillance.
Methods: We conducted a national study (UK-NEC study) involving all 163 neonatal units in England from December 2011 to September 2014. We included infants of all gestational ages investigated for abdominal concerns with clinical and abdominal x-ray (AXR) findings available in the National Neonatal Research Database (NNRD). We included AXRs which indicated that an infant had the diagnosis of NEC confirmed or refuted by laparotomy; for those without a laparotomy, only AXRs with non-ambiguous diagnoses such as ‘definite NEC’ or ‘no NEC’ were included; those with ‘probable’ NEC were excluded. Gestational age, clinical and radiological findings were entered as candidate variables into a logistic regression model. Estimates of the model which best predicted ‘NEC’ (defined as NEC confirmed by laparotomy, or ‘definite NEC’ for those without a laparotomy) were used to develop a scoring scheme, from which we derived an ordinal case definition using cut-offs yielding the highest area under the ROC curve and highest Positive Predictive Value. We defined a NEC score as the sum of points allocated for clinical and radiological findings based on the fit of the logistic regression (three points to pneumatosis, two points to blood in stools, one point to abdominal tenderness, abdominal discolouration, the composite of increased and/or bilious aspirates AND abdominal distension; one or more of pneumoperitoneum, fixed loop or air in the liver).
Results: All clinical and radiological findings differed significantly between infants with and without NEC. More immature infants were less likely to present with pneumatosis, blood or mucous in stool, and were more likely to have a gasless appearance. The ordinal scoring scheme discriminated well between infants with and without NEC (area under ROC curve 87.1%), with a higher score corresponding to a higher probability of NEC.
Conclusion: We present a simple scoring scheme for NEC, from which we have derived a GA-specific definition. This provides a tool for surveillance using a consistent case-definition.
Corresponding author: email@example.com