(84) Predicting Patent Ductus Arteriosus Patency at 72-hours Postnatal Using Clinical Variables Available at Neonatal Intensive Care Unit Admission

Background: Patent ductus arteriosus (PDA) is the most common cardiac condition in neonatal patients and the incidence increases as the degree of prematurity increases. PDA requires an echocardiogram (echo) for diagnosis and many infants experience spontaneous closure. The early ability to predict which preterm infants are at highest risk for PDA patency might guide clinicians when deciding whether or not echo surveillance for PDA is warranted reduce unnecessary echos.

Objectives: To predict PDA patency in < 30-week gestation preterm infants between 48-72 hours postnatal using clinical predictors available at neonatal intensive care unit (NICU) admission.

Design/Methods: We prospectively consented parents and enrolled 23-296/7-weeks gestation neonates without congenital anomalies. Screening echos evaluated PDA patency between 48-72 hours postnatal. Multivariable logistic regression models with least absolute shrinkage and selection operator (LASSO) regularization using 10-fold cross validation predicted PDA patency. Performance was assessed with the area under the receiver-operating curve (AUC), sensitivity, specificity, and Brier score. 95% confidence intervals, internal validation, and optimism correction were implemented via bootstrapping with 2000 replicates.

Results: Of 526 consented infants, 523 survived to the 48–72 hours screening ECHO and 446 had complete data (Table 1). The distribution of demographic and clinical variables was similar between infants consented and infants with complete data (Figure 1). The final prediction model included gestational age, birth weight z-score, indomethacin within 48 hours of birth, antenatal steroids, maternal indication for delivery, delivery method, infant mean oxygen and lowest blood pressure in the first 12 hours. The optimism-adjusted model sensitivity for prediction of PDA at 48-72 hours was 0.64 (95% CI: 0.51, 0.86) and specificity was 0.64 (95% CI: 0.44, 0.82) (Table 2). The adjusted AUC for the predicting an open PDA was 0.68 (95% CI: 0.65, 0.75) and the Brier score was 0.19 (95% CI: 0.16 – 0.20) (Table 2). Bootstrap validation indicated decent calibration and generalizability (Figure 2).

Conclusions: Based on our results, clinical measures at NICU admission did not confidently predict open PDA at 72 hours. We were not able to identify infants at low risk for PDA, echos remain an important part of clinical care for infants who may have open PDA.