Predictive Model of Early Spontaneous Ductus Arteriosus Closure Based on Neonatologist Performed Echocardiography in Preterm Infants

Background: Patent ductus arteriosus (PDA) treatment remains controversial. Modeling on the predictive capacity of early spontaneous PDA closure would help in decision-making. Aim: To design a predictive model of early spontaneous PDA closure. Methods: As part of a trial to assess efficacy and safety of two ibuprofen treatment schemes for PDA, infants below 29 weeks' gestation were scanned between 18 and 72 h of birth, and serially if indicated. PDA treatment was decided based on echocardiography signs of lung overflow or systemic hypoperfusion and clinical criteria. A PDA score that included the echocardiographic parameters significantly associated with treatment prescription was retrospectively applied. Perinatal variables and screening score were included in a backwards elimination model to predict early spontaneous closure. Results: Among 87 eligible infants (27 weeks' gestation; age at screening 45 h), 21 received ibuprofen at 69 h of life [screening score = 7 (IQR = 5-8.5); score at treatment = 9 (IQR = 8-9)], while 42 infants had conservative management, [screening score = 1 (IQR = 0-4)]. Twenty four infants were excluded (ibuprofen contraindication, declined consent or incomplete echocardiography). Screening score showed an AUC = 0.93 to predict early spontaneous PDA closure, [cut-off value = 4.5 (sensitivity = 0.90, specificity = 0.86)]. The predictive model for early spontaneous PDA closure followed the equation: Log (p/1-p) = -28.41 + 1.23^* gestational age -0.87^* PDA screening score. Conclusions: A predictive model of early spontaneous PDA closure that includes gestational age and the screening PDA score is proposed to help clinicians in the decision- making for PDA treatment. In addition, this model could be used in future intervention trials aimed to prevent PDA related morbidities to improve the eligibility criteria.