Adverse events associated with umbilical vascular catheters in neonatal intensive care: Development of a risk prediction model

Background: Adverse events associated with umbilical vascular catheters occur frequently in the neonatal intensive care unit. International guidelines recommend limiting catheter dwell time to reduce the risk of adverse events, and this drives clinical decision-making regarding catheter removal, yet other risk factors may also influence the risk of adverse events. 

Objectives: The aim of this study was to develop a clinically useful risk prediction model that could be utilised in the neonatal intensive care unit to identify infants at a greater risk of developing an adverse event associated with umbilical vascular catheters. 

Methods: A risk prediction model was developed for the umbilical venous catheter and umbilical arterial catheter based on the dataset of a published retrospective cohort study in a South Australian neonatal intensive care unit. Least absolute shrinkage and selection operator regression was used to develop the model. Deviance was used to evaluate the model's goodness of fit, and the Hosmer–Lemeshow test and calibration plot were used to assess calibration. The area under the receiver operating characteristic curve evaluated the model's discrimination. 

Results: For adverse events associated with umbilical venous catheters, the least absolute shrinkage and selection operator model selected none of the potential predictor variables. Five predictors of adverse events were identified for umbilical arterial catheters: thrombocytopaenia, intrauterine growth restriction/small for gestational age, congenital heart disease/defects including patent ductus arteriosus, maternal diabetes, and a dwell time of >7 days. The area under the receiver operating characteristic curve was 0.68 (95% confidence interval: 0.61–0.74). A link test found that the model was properly specified, and a Hosmer–Lemeshow test demonstrated that the model was well calibrated (p = 0.104). 

Conclusions: A risk prediction model has been developed to identify infants at a greater risk of an adverse event associated with umbilical arterial catheters. The model needs to be externally validated before it can be implemented into clinical practice.