Necrotizing enterocolitis (NEC) remains a major cause of neonatal morbidity and mortality. Some infants recover uneventfully with medical therapy whereas others develop severe disease (that is, NEC requiring surgery or resulting in death). Repeated attempts to identify clinical parameters that would reliably identify infants with NEC most likely to progress to severe disease have been unsuccessful. We hypothesized that comprehensive prospective data collection at multiple centers would allow us to develop a model which would identify those babies at risk for progressive NEC.This prospective, observational study was conducted at six university children's hospitals. Study subjects were neonates with suspected or confirmed NEC. Comprehensive maternal and newborn histories were collected at the time of enrollment, and newborn clinical data were collected prospectively, thereafter. Multivariate logistic regression analysis was used to develop a predictive model of risk factors for progression.Of 455 neonates analyzed, 192 (42%) progressed to severe disease, and 263 (58%) advanced to full feedings without operation. The vast majority of the variables studied proved not to be associated with progression to severe disease. A total of 12 independent predictors for progression were identified, including only 3 not previously described: having a teenaged mother (odds ratio, OR, 3.14; 95% confidence interval, CI, 1.45 to 6.96), receiving cardiac compressions and/or resuscitative drugs at birth (OR, 2.51; 95% CI, 1.17 to 5.48), and having never received enteral feeding before diagnosis (OR, 2.41; 95% CI, 1.08 to 5.52).Our hypothesis proved false. Rigorous prospective data collection of a sufficient number of patients did not allow us to create a model sufficiently predictive of progressive NEC to be clinically useful. It appears increasingly likely that further analysis of clinical parameters alone will not lead to a significant improvement in our understanding of NEC. We believe that future studies must focus on advanced biologic parameters in conjunction with clinical findings.
View details for DOI 10.1038/jp.2008.119
View details for Web of Science ID 000259675900003
View details for PubMedID 18784730