Oncel 2014

Methods	Randomised controlled trial conducted in the neonatal intensive care unit of Zekai Tahir Burak Maternity Teaching Hospital, Ankara,Turkey. Study period February to December 2012
Participants	90 infants with a gestational age ≤ 30 weeks, birth weight ≤ 1250 g, postnatal age 48 to 96 h, and 1 of the following echocardiographic criteria: a duct size > 1.5 mm, a left atrium‐to‐aorta ratio > 1.5, end diastolic reversal of blood flow in the aorta, or poor cardiac function in addition to clinical signs of a PDA Exclusion criteria were: the presence of major congenital abnormalities, right‐to‐left ductal shunting, life‐threatening infection, Grade III or Grade IV IVH, urine output of less than 1 mL/kg/h during the preceding 8 h, serum creatinine level > 1.6 mg/dL, platelet count < 60,000/mm³, liver failure, hyperbilirubinaemia requiring exchange transfusion, and persistent pulmonary hypertension
Interventions	45 infants received oral paracetamol at a dose of 15 mg/kg every 6 h for 3 days and 45 infants received oral ibuprofen at an initial dose of 10 mg/kg followed by 5 mg/kg at 24 and 48 h. Both paracetamol and ibuprofen were administered via an orogastric tube, which was flushed with 1 to 2 mL of sterile water to ensure delivery of the drug
Outcomes	Failure of PDA closure, all‐cause mortality, surgical closure of the PDA, duration of ventilator support, pulmonary haemorrhage, increase in grade of IVH, NEC, gastrointestinal bleed, ROP (requiring laser treatment), oliguria (not defined), sepsis (clinical symptoms and signs of sepsis and a positive blood bacterial culture), serum creatinine, bilirubin, AST and ALT, duration of hospitalisation. In 2017 the authors published neurodevelopmental outcomes of the infants enrolled in this trial; they reported on 30 children in the paracetamol group and 31 children in the ibuprofen group. They reported on neurodevelopmental impairment, MDI < 70, PDI < 70, moderate to severe cerebral palsy, blindness, deafness and MDI and PDI at 18 to 24 months corrected age.
Notes	We contacted Dr Oncel and he provided us with data for additional outcomes not reported in the published paper. In addition he provided outcome data for all 90 randomised infants
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Sequential numbers were generated at the computer centre of the NICU (information provided by the authors)
Allocation concealment (selection bias)	Low risk	The patients were randomly assigned to a treatment group by cards in sequentially numbered sealed opaque envelopes
Blinding of participants and personnel (performance bias) All outcomes	High risk	Paracetamol and ibuprofen were given according to different schedules and therefore it is likely that healthcare providers were not blinded to the drug the infant was given. The authors write: "....the intervention was not completely blinded because of the different number of doses per day of the drugs. However, the most important outcome—PDA closure—was made by a cardiologist who was blinded to the treatment groups. Second, safety outcomes should have been defined more clearly before the study started to prevent overestimation in evaluation"
Blinding of outcome assessment (detection bias) All outcomes	High risk	Paracetamol and ibuprofen were given according to different schedules and therefore it is likely that health care providers were not blinded to the drug the infant was given. The authors write: "....the intervention was not completely blinded because of the different number of doses per day of the drugs. However, the most important outcome—PDA closure—was made by a cardiologist who was blinded to the treatment groups. Second, safety outcomes should have been defined more clearly before the study started to prevent overestimation in evaluation"
Incomplete outcome data (attrition bias) All outcomes	Low risk	90 infants were randomised, and we received outcome data for the 10 infants (5 in ibuprofen group and 5 in paracetamol group) who died before the treatment was completed. Thus we received outcome data on an intention‐to‐treat basis for all 90 randomised infants
Selective reporting (reporting bias)	Low risk	The trial was registered at ClinicalTrials.gov — NCT01536158 — and there does not seem to be any deviations between the protocol and the full publication
Other bias	Low risk	Appears free of other bias