Novel Tool for Discrimination of Feeding Intolerance and Necrotizing Enterocolitis Risk in the Preterm Infant

Abstract

BackgroundPreterm infants are at increased risk of developing feeding intolerance and necrotizing enterocolitis (NEC). Comprehensive nursing assessments can improve infant outcomes through early identification of these conditions.ObjectiveTo validate a novel nursing tool for assessment of neonatal feeding intolerance and NEC risk.MethodsThe tool was developed based on published literature and implemented in a University of Illinois‐affiliated 48‐ bed level III neonatal intensive care unit following Institutional Review Board approval. Data were collected from the electronic medical record of all preterm infants with parental consent during the study period. Tool error rate, consistency (Cronbach's alpha), and inter‐rater reliability (intraclass correlation coefficient [ICC] and Fleiss’ kappa) were assessed. Risk factors significant to the development of feeding intolerance (days with emesis, abdominal distention, or gastric residuals > 50% of previous feeding volume) and NEC (ICD‐9 557.0, 777.50, 777.51, 777.52, or 777.53) were identified through χ2 testing. Tool discrimination was evaluated via receiver operating characteristic (ROC) curve. Pearson correlation coefficient and independent t‐test were used to assess construct validity. Following these assessments, the tool was optimized and reassessed. Optimization was achieved by removing factors not significant to ≥ 2 outcomes of interest and weighting point values assigned to each factor.ResultsFrom August 2013 to December 2014, 49 nurses scored 499 tools on the 133 enrolled infants. Cronbach's alpha (0.71), ICC (0.99), and Fleiss’ kappa (1.00) were in acceptable ranges. Scoring error rate was 9.2%, and inversely related to the number of tools a given nurse scored (R2 = 0.53, P = 0.01). Total optimized tool score was associated with all outcome measures (area under the ROC curve; diagnostic odds ratio [95% CI]) for the number of days with emesis (0.67; 1.12 [1.05, 1.21]), abdominal distention (0.81; 1.27 [1.17, 1.41]), or gastric residuals > 50% of previous feeding volume (0.62; 1.10 [1.03, 1.18]), and NEC (0.91; 1.30 [1.13, 1.59]). Correlation coefficients for each of these outcomes were 0.50, 0.60, 0.49, and 0.36, respectively. Optimized tool scores of infants who did and did not develop each of the outcome measures were significantly different (P < 0.05).ConclusionsThe tool is a clinically efficacious means to discriminate infants at risk of feeding intolerance and NEC. Implementation at additional institutions and further refinement will improve its clinical utility.Support or Funding InformationInternally funded.