BackgroundRandomized controlled trials (RCTs) have failed to demonstrate beneficial effects of the pharmacological treatment of patent ductus arteriosus (PDA) in preterm infants. We conducted a Bayesian model averaged (BMA) meta-analysis of RCTs comparing pharmacological treatment of PDA with placebo or expectant treatment.MethodsWe searched for RCTs including infants with gestational age (GA) ≤32 weeks and with a rate of open-label treatment of less than 25% in the control arm. Primary outcome was mortality and secondary outcomes included bronchopulmonary dysplasia (BPD). We calculated Bayes factors (BFs). The BF+/-is the ratio of the probability of the data under H+(pharmacological treatment is beneficial) over the probability of the data under H-(pharmacological treatment is harmful).ResultsFive RCTs were included (1341 infants). BMA showed strong evidence in favor of a harmful effect of medication for BPD (BF+/-=0.02) and BPD or death (BF+/-=0.03). When the two largest trials, which used early (<72 h) ibuprofen in infants with GA ≤28 weeks, were pooled, the BMA demonstrated moderate evidence in favor of higher mortality in the medication group (BF+/-=0.24).ConclusionsEarly ibuprofen treatment of a PDA in extremely preterm infants may result in more complications than clinical benefit.