Root cause analysis and Bayesian networks of Down syndrome and necrotizing enterocolitis

This study demonstrates the use of root cause analysis and Bayesian networks in assessing risk of Down syndrome and infant mortality due to necrotizing enterocolitis (NEC).

The contribution of maternal age, ethnicity, smoking, and infant's comorbidities on mortality associated with NEC (83 cases) was investigated using data obtained from the North Carolina linked birth/infant death files from 1999 to 2003. The data related to Down syndrome, which included 747 infants born with Down syndrome between the years of 1999 and 2003, were provided by the North Carolina Birth defects Monitoring Program. Flowcharts were built to identify potential risk factors and their associations, while the Bayesian network methodology was utilized to encode probabilistic relationships among these variables.

On the basis of the NEC model, the 3 most common causes of NEC infant mortality were respiratory tract conditions, cardiac, and infection-related problems. For the second application, prior live births (at least 1 prior birth) and infant's gender (male) were found to be the most prevalent causes of Down syndrome. Bayesian belief networks constitute an excellent tool for explorative and causal data analysis, and can assist health care providers in gaining insight into a complex problem.