An Exploratory Bayesian Network Analysis of Weight and Feeding Outcomes in 6-Month-Old Infants With Complex Congenital Heart Disease.

Purpose: We aimed to explore with a Bayesian network how clinically important factors related to feeding infants with complex congenital heart disease (CCHD) are associated with each other and influence feeding outcomes. Our goal was to raise questions for further study. Method: This descriptive study included data from 19 infants on severity of neonatal illness, early oral-motor (OM) and swallowing skills, feeding patterns, liquid and solid intake, and weight-for-age at 2 and 6 months. Bayesian network analysis was used to estimate the conditional probabilities of these variables in relation to each other and predict the 6-month weight-for-age z score and feeding outcomes (volume of liquid and solid food consumed) in the context of the other variables in the model. In clinically oriented scenarios, we manipulated the probability of specific predictor variables to 100% to examine the network effect on outcome variables. Results: Descriptive analyses revealed feeding and growth patterns consistent with prior literature. Bayesian network modeling identified three key themes: (a) feeding profiles may support risk stratification and guide targeted intervention, (b) OM skill development emerged as a foundational predictor of feeding and growth outcomes, and (c) clinical stability may obscure underlying feeding vulnerabilities. Conclusions: Bayesian network analysis provided insights into the conditional relationships among multiple factors, showing a method that could support clinical decision making. Further study with a larger, more diverse sample is needed to explore whether closer monitoring of intake and growth would promote better feeding outcomes, particularly for infants with less severe CCHD. [ABSTRACT FROM AUTHOR]