Early treatment versus expectant management of hemodynamically significant patent ductus arteriosus for preterm infants

Abstract

Background

Patent ductus arteriosus (PDA) is associated with significant morbidity and mortality in preterm infants. Nonsteroidal anti‐inflammatory drugs (NSAIDs) are used to prevent or treat a PDA. There are concerns regarding adverse effects of NSAIDs in preterm infants. Controversy exists on whether early targeted treatment of a hemodynamically significant (hs) PDA improves clinical outcomes.

Objectives

To assess the effectiveness and safety of early treatment strategies versus expectant management for an hs‐PDA in reducing mortality and morbidity in preterm infants.

Search methods

We used the standard search strategy of Cochrane Neonatal to search the Cochrane Central Register of Controlled Trials (CENTRAL 2019, Issue 6) in the Cochrane Library; MEDLINE via PubMed (1966 to 31 May 2019), Embase (1980 to 31 May 2019), and CINAHL (1982 to 31 May 2019). An updated search was run on 2 October 2020 in the following databases: CENTRAL via CRS Web and MEDLINE via Ovid. We searched clinical trial databases, conference proceedings, and the reference lists of retrieved articles for randomized controlled trials (RCT) and quasi‐randomized trials.

Selection criteria

We included RCTs in which early pharmacological treatment, defined as treatment initiated within the first seven days after birth, was compared to no intervention, placebo or other non‐pharmacological expectant management strategies for treatment of an hs‐PDA in preterm (< 37 weeks’ postmenstrual age) or low birth weight (< 2500 grams) infants.

Data collection and analysis

We performed data collection and analyses in accordance with the methods of Cochrane Neonatal. Our primary outcome was all‐cause mortality during hospital stay. We used the GRADE approach to assess the certainty of evidence for selected clinical outcomes.

Main results

We included 14 RCTs that enrolled 910 infants. Seven RCTs compared early treatment (defined as treatment initiated by seven days of age) versus expectant management and seven RCTs compared very early treatment (defined as treatment initiated by 72 hours of age) versus expectant management.

No difference was demonstrated between early treatment versus expectant management (no treatment initiated within the first seven days after birth) for an hs‐PDA for the primary outcome of ‘all‐cause mortality’ (6 studies; 500 infants; typical RR 0.80, 95% CI 0.46 to 1.39; typical RD ‐0.02; 95% CI ‐0.07 to 0.03; moderate‐certainty evidence), or other important outcomes such as surgical PDA ligation (4 studies; 432 infants; typical RR 1.08, 95% CI 0.65 to 1.80; typical RD ‐0.03; 95% CI ‐0.09 to 0.03; very low‐certainty evidence), chronic lung disease (CLD) (4 studies; 339 infants; typical RR 0.90, 95% CI 0.62 to 1.29; typical RD ‐0.03; 95% CI ‐0.10 to 0.03; moderate‐certainty evidence), severe intraventricular hemorrhage (IVH) (2 studies; 171 infants; typical RR 0.83,95% CI 0.32 to 2.16; typical RD ‐0.01; 95% CI ‐0.08 to 0.06; low‐certainty evidence), and necrotizing enterocolitis (NEC) (5 studies; 473 infants; typical RR 2.34,95% CI 0.86 to 6.41; typical RD 0.04; 95% CI 0.01 to 0.08; low‐certainty evidence). Infants receiving early treatment in the first seven days after birth were more likely to receive any PDA pharmacotherapy compared to expectant management (2 studies; 232 infants; typical RR 2.30, 95% CI 1.86 to 2.83; typical RD 0.57; 95% CI 0.48 to 0.66; low‐certainty evidence).

No difference was demonstrated between very early treatment versus expectant management (no treatment initiated within the first 72 hours after birth) for an hs‐PDA for the primary outcome of ‘all‐cause mortality’ (7 studies; 384 infants; typical RR 0.94, 95% CI 0.58 to 1.53; typical RD ‐0.03; 95% CI ‐0.09 to 0.04; moderate‐certainty evidence) or other important outcomes such as surgical PDA ligation (5 studies; 293 infants; typical RR 0.88, 95% CI 0.36 to 2.17; typical RD ‐0.01; 95% CI ‐0.05 to 0.02; moderate‐certainty evidence), CLD (7 studies; 384 infants; typical RR 0.83, 95% CI 0.63 to 1.08; typical RD ‐0.05; 95% CI ‐0.13 to 0.04; low‐certainty evidence), severe IVH (4 studies, 240 infants; typical RR 0.64, 95% CI 0.21 to 1.93; typical RD ‐0.02; 95% CI ‐0.07 to 0.04; moderate‐certainty evidence), NEC (5 studies; 332 infants; typical RR 1.08, 95% CI 0.53 to 2.21; typical RD 0.01; 95% CI ‐0.04 to 0.06; moderate‐certainty evidence) and neurodevelopmental impairment (1 study; 79 infants; RR 0.27, 95% CI 0.03 to 2.31 for moderate/severe cognitive delay at 18 to 24 months; RR 0.54, 95% CI 0.05 to 5.71 for moderate/severe motor delay at 18 to 24 months; RR 0.54, 95% CI 0.10 to 2.78 for moderate/severe language delay at 18 to 24 months; low‐certainty evidence). Infants receiving very early treatment in the first 72 hours after birth were more likely to receive any PDA pharmacotherapy compared to expectant management (4 studies; 156 infants; typical RR 1.64, 95% CI 1.31 to 2.05; typical RD 0.69; 95% CI 0.60 to 0.79; very low‐certainty evidence). Very early treatment, however, shortened the duration of hospitalization compared to expectant management (4 studies; 260 infants; MD ‐5.35 days; 95% CI ‐9.23 to ‐1.47; low‐certainty evidence).

Authors' conclusions

Early or very early pharmacotherapeutic treatment of an hs‐PDA probably does not reduce mortality in preterm infants (moderate‐certainty evidence). Early pharmacotherapeutic treatment of hs‐PDA may increase NSAID exposure (low‐certainty evidence) without likely reducing CLD (moderate‐certainty evidence), severe IVH or NEC (low‐certainty evidence). We are uncertain whether very early pharmacotherapeutic treatment of hs‐PDA also increases NSAID exposure (very low‐certainty evidence). Very early treatment probably does not reduce surgical PDA ligation, severe IVH or NEC (moderate‐certainty evidence), and may not reduce CLD or neurodevelopmental impairment (low‐certainty evidence). Additional large trials that specifically include preterm infants at the highest risk of PDA‐attributable morbidity, are adequately powered for patient‐important outcomes and are minimally contaminated by open‐label treatment are required to explore if early targeted treatment of hs‐PDA improves clinical outcomes. There are currently two trials awaiting classification and two ongoing trials exploring this question.

Plain language summary

Early treatment versus expectant management of symptomatic patent ductus arteriosus in preterm infants

Review question

Does early treatment (initiated within the first seven days after birth) or very early treatment (initiated within the first 72 hours after birth) of a symptomatic patent ductus arteriosus (PDA) with nonsteroidal anti‐inflammatory drugs (NSAIDs) compared to expectant management (without use of NSAIDs) improve outcomes in preterm infants?

Background

PDA is a common complication in preterm or low birth weight infants. PDA is an open vascular channel between the lungs and the heart which usually closes shortly after birth. In preterm infants, the PDA often remains open and may contribute to life‐threatening complications. Medications such as NSAIDs are used to prevent or treat a PDA before it becomes symptomatic. However, there are concerns regarding the side‐effects of NSAIDs in preterm infants. There is controversy on whether early treatment of a symptomatic PDA improves outcomes in preterm infants.

Study characteristics

We searched scientific databases for randomized controlled trials (clinical studies where people are randomly put into one of two or more treatment groups) in preterm (born at less than 37 weeks into pregnancy) or low‐birth‐weight (weighing less than 2500 grams) infants with a symptomatic PDA diagnosed using a combination of specific clinical features and ultrasound of the heart. The included studies compared early treatment, or very early treatment of a symptomatic PDA with NSAIDs compared to expectant management without the use of NSAIDs. The search is up to date as of 2 October 2020.

Key results

This review of 14 clinical trials (910 infants) found that early or very early treatment of a symptomatic PDA does not reduce death or other poor clinical outcomes in preterm infants. Early or very early treatment, on the other hand, appears to increase the number of preterm infants exposed to NSAIDs. There are currently two trials awaiting classification and two ongoing trials exploring this question.

Certainty of evidence

According to GRADE (a method to score the certainty of the trials supporting each outcome), the certainty of the evidence varied from very low to moderate but was moderate for the most important outcome of death.

Background

Description of the condition

The ductus arteriosus is a blood vessel that connects the two major arteries coming out of the heart (i.e. the aorta from the left ventricle and the pulmonary artery from the right ventricle), which plays an important role in maintaining fetal circulation (Gournay 2011). Following birth with the initiation of breathing and separation of the low‐resistance placenta, closure of the ductus arteriosus begins and functional closure continues over the next 24 to 72 hours (Benitz 2016). In preterm infants, the closure is often delayed leading to the ductus arteriosus remaining patent beyond the first few days of life. In healthy preterm neonates of more than 30 weeks' gestation, this patent ductus arteriosus (PDA) closes by day four in 90%, and by day seven in 98% of infants (Clyman 2012). In extremely preterm infants born at less than 24 weeks of gestation, the spontaneous PDA closure rates are only about 8% by day four, and 13% by day seven (Clyman 2012).

When the ductus arteriosus persists beyond the first few day of life, as pulmonary vascular resistance declines, blood starts flowing left‐to‐right from the aorta into the pulmonary arteries (Benitz 2016). With further decreases in pulmonary vascular resistance over the first several days after birth, the proportion of aortic blood flow that is diverted into the pulmonary circulation correspondingly increases (Benitz 2016). This 'ductal steal' results in excessive blood flow through the lungs, predisposing to development of pulmonary congestion, pulmonary edema, and worsening respiratory failure (Benitz 2016). At the same time, diversion of blood flow from the systemic circulation leads to systemic hypoperfusion, resulting in compromised perfusion to the bowel, kidney, and brain. When a PDA is associated with clinical or echocardiographic signs of pulmonary hyperperfusion and systemic hypoperfusion, it is defined as a hemodynamically significant PDA. A persistent hemodynamically significant PDA may be associated with numerous adverse outcomes, including prolongation of assisted ventilation and higher rates of death (Dice 2007), bronchopulmonary dysplasia (BPD) (Brown 1979), necrotizing enterocolitis (NEC) (Dollberg 2005), impaired renal function (Benitz 2016), intraventricular hemorrhage (IVH) (Ballabh 2010), periventricular leukomalacia (PVL) (Chung 2005), and cerebral palsy (Drougia 2007). However, the causal link between these associations has not been demonstrated (Benitz 2010).

Description of the intervention

Due to the above‐mentioned potential life‐threatening complications, nonsteroidal anti‐inflammatory drugs (NSAIDs) such as indomethacin and ibuprofen are used to close a hemodynamically significant PDA. NSAIDs act by inhibition of the cyclo‐oxygenase (COX) enzyme, thereby leading to down regulation of prostaglandin E₂ (PGE₂), a potent relaxant of the PDA (Mitra 2013). Acetaminophen (paracetamol) has also emerged as a potential pharmacotherapeutic option for PDA closure (Le 2015). Acetaminophen is postulated to exert its action through inhibition of the peroxidase enzyme thereby leading to down regulation of PGE₂ production (Grèen 1989). Use of indomethacin in preterm infants has been associated with transient or permanent derangement of renal function (Seyberth 1983), NEC (Coombs 1990), gastrointestinal hemorrhage or perforation (Wolf 1989), alteration of platelet function (Friedman 1976), and impairment of cerebral blood flow/cerebral blood flow velocity (Ohlsson 1993). Ibuprofen appears to be associated with a lower risk of NEC and transient renal insufficiency as compared to indomethacin (Ohlsson 2020b). Acetaminophen has no short‐term adverse effects. However, there are limited data on the long‐term neurodevelopmental effects of acetaminophen in preterm infants (Ohlsson 2020a). There is considerable variation in the timing of treatment where some neonatologists choose to treat a hemodynamically significant PDA early in the first few days of life to mitigate the adverse effects of persistent left–right shunting through the ductus, while some neonatologists choose to delay treatment awaiting spontaneous closure of the PDA. In this review, we define early treatment of a hemodynamically significant PDA as intention‐to‐treat the PDA using a pharmacotherapeutic agent within the first seven days of birth, and very early treatment of a hemodynamically significant PDA by 72 hours of age.

We define expectant management of the PDA as intention‐to‐manage a clinically or echocardiographically diagnosed (or both) hemodynamically significant PDA without pharmacological treatment within the time period defined as early treatment for the respective clinical trial. Expectant management could include non‐pharmacological interventions such as fluid restriction or modification of parameters on mechanical ventilation.

How the intervention might work

NSAIDs and acetaminophen are effective in closing a PDA compared to placebo (Mitra 2018). Ibuprofen appears as effective as indomethacin in closing a PDA while reducing the risk of NEC and transient renal insufficiency (Ohlsson 2020b). There is moderate‐quality evidence to suggest that acetaminophen is as effective as ibuprofen and low‐quality evidence to suggest that acetaminophen is as effective as indomethacin in closing a PDA (Ohlsson 2020a). However, there is an increasing concern that infants who require pharmacotherapy for a hemodynamically significant PDA may not respond to the standard treatment doses of NSAIDs if they are treated beyond the first 72 hours of life due to altered pharmacokinetics of the medications. To achieve optimal concentrations of ibuprofen for successful PDA closure, irrespective of gestational age, three doses of 10 mg/kg, 5 mg/kg, and 5 mg/kg are required at 24‐hour intervals for neonates younger than 70 hours; 14 mg/kg, 7 mg/kg, and 7 mg/kg for neonates between 70 and 108 hours; and 18 mg/kg, 9 mg/kg, and 9 mg/kg for neonates between 108 and 180 hours of life (Hirt 2008). Therefore, timely and effective treatment within the first few days of life may prevent potential complications arising from a persistent hemodynamically significant PDA.

Why it is important to do this review

Controversy exists on timing of PDA treatment or whether the PDA should be actively treated at all in preterm infants. Spontaneous closure of PDA occurs in 34% of infants with birth weight less than 1000 g and in 67% of infants with birth weight between 1000 g and1500 g in the first seven days of life (Clyman 2012). Therefore, expectant treatment of the PDA in the first seven days after birth may not be associated with adverse clinical outcomes in a significant proportion of preterm infants. Moreover, there is little evidence to support successful closing of the PDA leading to changes in clinical outcomes in preterm infants. One Bayesian network meta‐analysis, exploring pharmacotherapeutic treatment options for a hemodynamically significant PDA, found that placebo or no treatment did not significantly alter clinical outcomes such as mortality, NEC, or IVH, which suggests that expectant management of the PDA could also be an equally viable management option (Mitra 2018). However, there were too few data on timing of treatment in the included studies to draw a definite conclusion from the network meta‐regression analysis. One Cochrane Review explored early versus expectant management as a secondary outcome and included one randomized controlled trial (RCT) that defined early treatment as any treatment prior to 14 days of age (Ohlsson 2020b). There are no Cochrane Reviews that explore expectant versus early management of a hemodynamically significant PDA. Currently, there are two large, multicenter, RCTs underway that explore the proposed research question: the BeNeDuctus Trial (Belgium Netherlands Ductus Trial; NCT02884219) and the Baby-OSCAR Trial (Outcome after Selective early treatment for Closure of patent ductus ARteriosus in pre‐term babies; ISRCTN84264977). Therefore, a systematic review according to Cochrane methodology is justified as new trials have been identified.

Objectives

To assess the effectiveness and safety of early treatment strategies versus expectant management for a hemodynamically significant patent ductus arteriosus (PDA) in reducing mortality and morbidity in preterm infants.

Methods

Criteria for considering studies for this review

Types of studies

We included published and unpublished RCTs, quasi‐RCTs, and cluster‐RCTs comparing early treatment versus expectant management of hemodynamically significant PDA for preterm infants. Both superiority trials and non‐inferiority trials were eligible for inclusion.

Types of participants

Preterm (gestational age less than 37 weeks at birth) or low birth weight infants (less than 2500 grams) with a hemodynamically significant PDA diagnosed clinically or via echocardiography (or both) in the first seven days of life.

A hemodynamically significant PDA was defined clinically by the presence of a precordial murmur along with one or more of the following signs: hyperdynamic precordial impulse, tachycardia, bounding pulses, widened pulse pressure, worsening respiratory status, hypotension, or cardiac failure.

A hemodynamically significant PDA was defined echocardiographically by a moderate‐to‐large transductal diameter (PDA diameter greater than 1.5 mm with or without unrestrictive pulsatile flow, i.e. maximum systolic shunt velocity less than 2 m/second) with or without evidence of pulmonary over circulation (left atrium to aortic root ratio greater than 1.5 OR isovolumetric relaxation time less than 55 m seconds OR E:A ratio of 1.0 or greater or left ventricular output greater than 300 mL/kg/minute OR diastolic disturbance in the main pulmonary artery) with or without evidence of systemic hypoperfusion (absent/reversed diastolic flow in the postductal descending aorta or celiac trunk or middle cerebral artery) (El‐Khuffash 2013); or used a scoring system that incorporated one or more of the echocardiographic measures mentioned above (El‐Khuffash 2015).

Types of interventions

The interventions included early pharmacological treatment and expectant management for a hemodynamically significant PDA diagnosed clinically or via echocardiography (or both).

Early pharmacological treatment

Early pharmacological treatment included 'early treatment' defined as intention‐to‐treat a hemodynamically significant PDA within the first seven days of birth, and 'very early treatment' defined as intention‐to‐treat a hemodynamically significant PDA within the first 72 hours of life. The pharmacological treatments included indomethacin, ibuprofen, and acetaminophen. There were no restrictions on dose, route, or duration of treatment.

Expectant management

Expectant management was defined as intention‐to‐manage a clinically or echocardiographically (or both) diagnosed hemodynamically significant PDA without treatment within the time period defined as early treatment for the respective clinical trial. Beyond the defined 'early treatment' period, expectant management included no treatment, use of the same pharmacotherapeutic agent (i.e. same dose, route, and duration) as in the early treatment arm or surgical intervention for PDA closure.

Planned comparisons

The planned comparisons included the following:

1. early treatment (treatment of a hemodynamically significant PDA by seven days of age) versus expectant management (no treatment of a hemodynamically significant PDA in the first seven days after birth);

2. very early treatment (treatment of a hemodynamically significant PDA by 72 hours of age) versus expectant management (no treatment of a hemodynamically significant PDA in the first 72 hours after birth);

3. very early treatment (treatment of a hemodynamically significant PDA by 72 hours of age) versus early treatment (treatment of a hemodynamically significant PDA by seven days of age).

Types of outcome measures

Primary outcomes

Outcomes were selected based on their importance for decision‐making and included:

1. All‐cause mortality during hospital stay.

Secondary outcomes

1. Proportion of infants requiring surgical PDA ligation or transcatheter occlusion.

2. Proportion of infants receiving any pharmacotherapy for a hemodynamically significant PDA (defined as any receipt of pharmacotherapy for a hemodynamically significant PDA any time during hospital stay).

3. Proportion of infants receiving repeat courses of pharmacotherapy (repeated COX inhibitor or acetaminophen dosing, or both) for a persistent hemodynamically significant PDA.

4. Proportion of infants receiving rescue medical treatment (COX inhibitor or acetaminophen dosing, or both) for a hemodynamically significant PDA .

5. Chronic lung disease (CLD; defined as oxygen requirement at 36 weeks' postmenstrual age) (Ehrenkranz 2005).

6. Duration of ventilator support (days).

7. Duration of need for supplementary oxygen (days).

8. Postnatal corticosteroid use for CLD any time during hospital stay.

9. Pneumothorax.

10. Pulmonary hemorrhage.

11. Pulmonary hypertension (defined as hypoxemia refractory to oxygen therapy or to lung recruitment strategies (partial pressure of oxygen in arterial blood (PaO₂) less than 55 mmHg despite fraction of inspired oxygen (FiO₂) of 1.0) associated with a preductal‐to‐postductal oxygen gradient greater than 20 mmHg) (Roberts 1997; Walsh‐Sukys 2000).

12. Intraventricular hemorrhage (IVH) (grades I to IV) (Papile 1978).

13. Severe IVH (grades III and IV).

14. Periventricular leukomalacia (PVL; any grade) (De Vries 1992).

15. Necrotizing enterocolitis (NEC; stage 2 or greater) (Bell 1978).

16. Gastrointestinal bleeding within seven days of the first dose of pharmacotherapy.

17. Spontaneous intestinal perforation.

18. Time to full enteral feeds (postnatal age at time of achieving full enteral feeds).

19. Time to regain birth weight (days).

20. Severe retinopathy of prematurity (ROP) (stage 3 or greater) (according to the international classification of ROP) (ICCROP 2005).

21. Definite sepsis (clinical symptoms and signs of sepsis and a positive bacterial culture in a specimen obtained from normally sterile fluids or tissue obtained at postmortem).

22. Oliguria (defined as less than 1 mL/kg/hour).

23. Duration of hospitalization (total length of hospitalization from birth to discharge home or mortality) (days).

24. Neurodevelopmental outcome (assessed by a standardized and validated assessment tool, a child developmental specialist, or both) at any age reported (outcome data grouped at 12, 18, and 24 months, if available).

Search methods for identification of studies

Electronic searches

We initially conducted a comprehensive search including: Cochrane Central Register of Controlled Trials (CENTRAL 2019, issue 6) in the Cochrane Library; MEDLINE via PubMed (1996 to 31 May 2019); Embase via Ovid (1980 to 31 May 2019); and CINAHL via EBSCOhost (1982 to 31 May 2019) using the following search terms: patent ductus arteriosus, PDA, indomethacin, ibuprofen, paracetamol, acetaminophen, Tylenol, and applicable MeSH terms plus database‐specific limiters for RCTs and neonates. We further conducted a comprehensive updated search in October 2020 including: Cochrane Central Register of Controlled Trials (CENTRAL 2020, Issue 10) in the Cochrane Library and Ovid MEDLINE(R) and Epub Ahead of Print, In‐Process & Other Non‐Indexed Citations, Daily and Versions(R) (01 January 2019 to 02 October 2020).

We searched clinical trial registries for ongoing or recently completed trials. We searched the World Health Organization’s International Clinical Trials Registry Platform (ICTRP) (www.who.int/ictrp/search/en/) and the U.S. National Library of Medicine’s ClinicalTrials.gov (clinicaltrials.gov) via Cochrane CENTRAL. Additionally, we searched the ISRCTN Registry (http://www.isrctn.com/) for any unique trials not found through the Cochrane CENTRAL search.

We have included the updated search strategy for each database in Appendix 1. The initial search details are listed in Appendix 2. We did not apply language restrictions.

Searching other resources

Additionally, we identified relevant studies not identified in the primary search through a review of the reference lists of all identified articles, correspondence with experts in the field, and a search of conference proceedings for the European Society for Pediatric Research and US Pediatric Academic Societies (1990 to 2020).

Data collection and analysis

We used the standard methods of Cochrane Neonatal.

Selection of studies

Three review authors (SM, AS and TD) independently screened the search results in duplicate by title and abstract for studies that potentially met the inclusion criteria. We obtained the full text of any articles that were potentially eligible, and independently performed full‐text assessments in duplicate. We resolved any disagreements through consensus. We recorded the selection process in sufficient detail to complete a PRISMA flow diagram (Moher 2009) (Figure 1), and 'Characteristics of included studies' and 'Characteristics of excluded studies' table. The study selection process was conducted on the Covidence platform.

1.

Study flow diagram

Data extraction and management

Three review authors (SM, AS and AvK) extracted, assessed, and coded all data for each study, using a standardized form developed on the Covidence platform. We replaced any standard error of the mean by the corresponding standard deviation. We used the formulae proposed by Hozo and colleagues, to estimate means and standard deviations (SD) from medians and ranges presented by the authors of some of the included studies (Hozo 2005). We resolved any disagreements through consensus. For each study, one review author (SM) entered the extracted data into Review Manager 5 (Review Manager 2020); a second review author (AS) checked data entry. All review authors assessed the protocol, analysis, and draft manuscript. We collected information regarding the method of randomization, blinding, drug intervention, and stratification for each included study. We noted the information regarding trial participants including gestational age criteria, birth weight criteria, and other inclusion or exclusion criteria. We analyzed the information on clinical outcomes of the primary and secondary outcomes.

Assessment of risk of bias in included studies

Two review authors (SM and AvK) independently assessed the risk of bias (low, high, or unclear) of all included trials using the Cochrane 'Risk of bias' tool (Higgins 2011), for the following domains (Higgins 2017):

1. sequence generation (selection bias);

2. allocation concealment (selection bias);

3. blinding of participants and personnel (performance bias);

4. blinding of outcome assessment (detection bias);

5. incomplete outcome data (attrition bias);

6. selective reporting (reporting bias);

7. any other bias.

We resolved any disagreements by discussion or with a third‐party assessor. See Appendix 3 for a more detailed description of risk of bias for each domain.

Measures of treatment effect

We performed statistical analyses using Review Manager 5 (Review Manager 2020). We used risk ratios (RRs) and risk differences (RDs) for categorical variables, and mean differences (MDs) for continuous variables. We reported the 95% confidence interval (CI) on all estimates.

Unit of analysis issues

The unit of analysis was the participating infant in individually randomized trials, and we conducted intention‐to‐treat analyses.

If included trials used cluster randomization and the analysis was adjusted for patient clustering, we extracted this effect estimate. If clustering was ignored, we conducted analyses by calculating an effective sample size for intervention and control. We used the intracluster correlation coefficient of the study if reported or a reasonable external estimate if it was not (Higgins 2017). If we identified both cluster‐RCTs and individually randomized controlled trials, we only combined the results from both if there was little heterogeneity between the study designs and the interaction between the effect of the intervention and the choice of randomization unit was unlikely.

In the event that included trials had more than two intervention groups, we combined the groups to create a single pairwise comparison as recommended in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2017).

Dealing with missing data

We contacted the study authors to request any missing data. If this was unsuccessful, we assumed that data were missing at random and conducted analysis using the data reported by study authors (e.g. the denominator being the number randomized minus participants whose outcomes were missing). We planned to conduct a sensitivity analysis assessing the impact of including these trials. If standard deviations for continuous outcomes were missing, we first looked for summary statistics that could be used to calculate these, contacted authors when these were not available, and finally used methods outlined in Chapter 16 of the Cochrane Handbook for Systematic Reviews of Interventions to impute missing standard deviations if the proportion of studies requiring imputation was small (Higgins 2017). We planned to conduct sensitivity analysis to assess the impact of including any imputed values.

Assessment of heterogeneity

Prior to synthesis, we assessed all studies for differences that may give rise to heterogeneity. These potential effect modifiers included: differences in study quality, participants, timing of the intervention, pharmacotherapeutic regimens, and outcome assessments. We additionally examined evidence of heterogeneity visually by inspecting the forest plots, and quantitatively using the I² statistic. We assessed the degree of heterogeneity as:

1. less than 25%: no heterogeneity;

2. 25% to 49%: low heterogeneity;

3. 50% to 74%: moderate heterogeneity; or

4. greater than 75%: substantial heterogeneity.

Assessment of reporting biases

We planned to assess possible publication bias through contour‐enhanced funnel plots if there were 10 or more clinical trials for a particular comparison (Higgins 2017). In addition, we evaluated whether results of published posters and available dissertations were subsequently published as full‐length manuscripts. We identified records in trial registries that had been terminated, listed as complete, or should have been feasibly completed given last updated status with regard to availability of results or subsequent publication. For preregistered trials or those with published protocols, we assessed the presence of reporting bias through comparison of their preplanned primary and secondary outcomes and analysis methods against those reported and used in the published report.

Data synthesis

When considered appropriate, we used Review Manager 5 to conduct meta‐analyses (Review Manager 2020). For categorical outcomes, we calculated the typical estimates of RR and RD, each with its 95% CI; and for continuous outcomes, we determined the MD with its 95% CI. All models were analyzed using fixed‐effect models. If meta‐analysis was considered to be inappropriate, we analyzed and interpreted individual trials separately.

Subgroup analysis and investigation of heterogeneity

Subgroup analyses

If the information was available, we planned to conduct the following subgroup analyses for the primary outcome.

1. Gestational age (less than 28 weeks, 28 to 32 weeks, 33 to 37 weeks).

2. Birth weight (less than 1000 g, 1000 g to 1500 g, 1501 g to 2500 g).

3. Method used to diagnose a hemodynamically significant PDA (by echocardiographic criteria or only by clinical criteria).

4. Degree of hemodynamic significance of the PDA (based on echocardiographic criteria):

   1. moderately hemodynamically significant PDA (defined by a moderate transductal diameter (PDA diameter 1.5 mm to 3 mm with or without unrestrictive pulsatile flow, i.e. maximum systolic shunt velocity less than 2 m/second) with or without evidence of pulmonary over circulation (left atrium‐to‐aortic root ratio 1.5 to 2.0 OR isovolumetric relaxation time 45 milliseconds to 55 milliseconds or E:A ratio 1.0 or left ventricular output 300 mL/kg/minute to 400 mL/kg/minute or main pulmonary artery diastolic velocity 0.3 m/second to 0.5 m/second) with or without evidence of systemic hypoperfusion (absent diastolic flow in the postductal descending aorta or celiac trunk or middle cerebral artery); or

   2. large hemodynamically significant PDA (defined by a large transductal diameter (PDA diameter greater than 3 mm with or without unrestrictive pulsatile flow, i.e. maximum systolic shunt velocity less than 2 m/second) with or without evidence of pulmonary over circulation (left atrium‐to‐aortic root ratio greater than 2.0 or isovolumetric relaxation time less than 45 milliseconds or E:A ratio greater than 1.0 or left ventricular output greater than 400 mL/kg/minute or main pulmonary artery diastolic velocity greater than 0.5 m/second) with or without evidence of systemic hypoperfusion (reversed diastolic flow in the postductal descending aorta or celiac trunk or middle cerebral artery) (El‐Khuffash 2013).

5. Use of prophylactic COX inhibitor therapy in the first 24 hours (yes/no).

6. Pharmacotherapeutic agent for PDA treatment (Appendix 4).

7. Co‐interventions (fluid intake restriction, diuretic medication, optimizing hematocrit levels, optimizing platelet counts, optimizing mechanical ventilation).

Sensitivity analysis

We planned to conduct sensitivity analyses to determine whether findings were affected by including only studies of adequate methodology (low risk of bias), defined as adequate randomization and allocation concealment, blinding of intervention and measurement, and up to and including a 10% loss to follow‐up.

Summary of findings and assessment of the certainty of the evidence

We used the GRADE approach, as outlined in the GRADE Handbook to assess the certainty of evidence for the following (clinically relevant) outcomes (Schünemann 2013):

1. All‐cause mortality during hospital stay.

2. Proportion of infants requiring surgical PDA ligation or transcatheter occlusion.

3. Proportion of infants receiving any pharmacotherapy for a hemodynamically significant PDA (defined as any receipt of pharmacotherapy for a hemodynamically significant PDA any time during hospital stay).

4. CLD (defined as oxygen requirement at 36 weeks' postmenstrual age).

5. Severe IVH (grades III and IV).

6. NEC (stage 2 or greater).

7. Duration of hospitalization (total length of hospitalization from birth to discharge home or mortality).

8. Neurodevelopmental outcome (assessed by a standardized and validated assessment tool, a child developmental specialist, or both) at any age reported (outcome data grouped at 12, 18, and 24 months, if available).

Two out of the following three review authors (SM, AS and TD) independently assessed the certainty of the evidence for each of the outcomes above. We considered evidence from RCTs as high certainty but downgraded the evidence one level for serious (or two levels for very serious) limitations based upon the following: design (risk of bias), consistency across studies, directness of the evidence, precision of estimates, and presence of publication bias. We used the GRADEpro GDT Guideline Development Tool to create two 'Summary of findings' tables to report the certainty of the evidence.

The GRADE approach results in an assessment of the certainty of a body of evidence in one of four grades.

1. High: we are very confident that the true effect lies close to that of the estimate of the effect.

2. Moderate: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.

3. Low: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.

4. Very low: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

Results

Description of studies

We identified 1456 potentially relevant studies through the electronic search. We identified 48 studies of relevance through title and abstract screening followed by full‐text screening. We excluded 30 studies through full‐text screening. We further identified two studies that are awaiting classification (ACTRN12616001517460; NCT01630278), and two ongoing studies (ISRCTN84264977, NCT02884219), leaving 14 studies which were included in the review.

Results of the search

The results of the updated search conducted in October 2020 are shown in Figure 1 (PRISMA flow diagram).

Included studies

We included a total of 14 studies that enrolled 910 participants.

The included studies compared early (seven studies) or very early (seven studies) medical treatment of PDA with expectant management. The individual study characteristics, inclusion criteria, treatment details, outcome details and author information can be found in the Characteristics of included studies table.

Early versus expectant management

Seven studies compared early treatment with expectant management of PDA (Bagnoli 2013; Gersony 1983; Ghanem 2010; Knight 2011; Krauss 1989; Sosenko 2012; Van Overmeire 2001). All seven studies were RCTs. For the Gersony 1983 study, data from only a subgroup of infants randomized within the first five days after birth was specifically included in the meta‐analysis. Knight 2011 was published as a conference abstract.

Of these seven studies, two used intravenous ibuprofen (Bagnoli 2013; Sosenko 2012) (ibuprofen‐lysine), four used intravenous indomethacin (Gersony 1983; Knight 2011; Krauss 1989; Van Overmeire 2001), and one used oral ibuprofen (Ghanem 2010). Both Bagnoli 2013 and Sosenko 2012 used three doses of IV ibuprofen: 10 mg/kg, followed by two 5 mg/kg doses on successive days. Ghanem 2010 used the same dose of ibuprofen, but via the oral route. Krauss 1989 and Van Overmeire 2001 used three doses of indomethacin or 200 micrograms/kg on successive days, whereas Gersony 1983 used either this dose, or 200 micrograms/kg followed by two 100 micrograms/kg doses. Knight 2011 used 200 micrograms/kg or 100 micrograms/kg of indomethacin on six successive days.

‘Early treatment’ was defined in this review as medication administered prior to seven days since birth. Each study used different age entry criteria. These can be found in the table Characteristics of included studies. The infants' age ranges varied: 24 hours to 14 days (Sosenko 2012); 48 to 96 hours (Ghanem 2010); 72 to 96 hours (Krauss 1989); 2 to 6 days (Knight 2011); > 72 hours (Bagnoli 2013); < 5 days (Gersony 1983); and 3 to 7 days (Van Overmeire 2001).

The studies varied in size between 26 participants (Knight 2011), and 134 participants (Bagnoli 2013). The studies were performed between 1981 (Gersony 1983), and 2010 (Sosenko 2012).

A weight cutoff for inclusion of < 1750 g was used by Gersony 1983, < 1500 g by three studies (Bagnoli 2013; Ghanem 2010; Krauss 1989), and < 1250 g by two studies (Knight 2011; Sosenko 2012). Infants of gestational age at birth < 32 weeks were included by five studies (Bagnoli 2013; Ghanem 2010; Sosenko 2012; Van Overmeire 2001). Knight 2011 included only infants born < 28 weeks.

Two studies were multicenter (Gersony 1983 ‐ 13 centers; Van Overmeire 2001 ‐ four centers), and the others were single‐center. Three studies were performed in the USA (Gersony 1983; Krauss 1989; Sosenko 2012); two in Europe (Bagnoli 2013; Van Overmeire 2001), one in Australia (Knight 2011), and one in Saudi Arabia (Ghanem 2010).

Very early versus expectant management

Seven studies compared very early treatment with expectant management of PDA (CTRI/2009/091/000041; DeWaal 2020; EL‐Khuffash 2020; Kaapa 1983; Kluckow 2014; Lin 2012; Merritt 1981). All seven studies were RCTs. CTRI/2009/091/000041 was published as a conference abstract. Varghese 2016 is a follow‐up study of the RCT by Kluckow 2014, using the same participants.

Of these seven RCTs, four used intravenous indomethacin (CTRI/2009/091/000041; Kaapa 1983; Kluckow 2014; Merritt 1981), one used intravenous ibuprofen (EL‐Khuffash 2020), one used intravenous ibuprofen‐lysine or indomethacin (DeWaal 2020), and one used oral ibuprofen (Lin 2012). The doses of intravenous indomethacin used were 200 micrograms/kg followed by two 100 micrograms/kg doses 24 hours apart in three RCTs (CTRI/2009/091/000041; DeWaal 2020; Kluckow 2014), 200 micrograms/kg repeated daily as necessary until ductal closure in one RCT (Kaapa 1983), and 200 micrograms/kg repeated daily up to three times in one RCT (Merritt 1981). The doses of intravenous ibuprofen used were 10 mg/kg, followed by two 5 mg/kg doses 24 hours apart by EL‐Khuffash 2020 and DeWaal 2020. The same dosing regimen was used via the oral route by Lin 2012.

‘Very early’ treatment was defined in this review as medication administered within 72 hours of birth. However, between the studies there was some variation. For full inclusion criteria, please see the Characteristics of included studies.

Infants were included if they weighed < 1500 g in one RCT (Lin 2012) and < 1350 g in a further RCT (Merritt 1981). Infants were included if born at < 29 weeks of gestation in three RCTs (DeWaal 2020; EL‐Khuffash 2020; Kluckow 2014). Infants born at < 32 weeks were included by Lin 2012, whereas CTRI/2009/091/000041 had gestational range of 28 to 32 weeks. Kaapa 1983 included all infants born preterm.

Included RCTs were performed between 1981 (Merritt 1981), and 2020 (EL‐Khuffash 2020). The RCTs varied in size between 24 participants (Merritt 1981), and 92 participants (Kluckow 2014). Two RCTs were multicenter (Kluckow 2014 ‐ three centers; DeWaal 2020 ‐ two centers) and the rest were single‐center. Two RCTs were performed in Australia (Kluckow 2014; DeWaal 2020); two in Europe (EL‐Khuffash 2020; Kaapa 1983); one in India (CTRI/2009/091/000041); one in China (Lin 2012); and one in the USA (Merritt 1981).

Excluded studies

We excluded 30 studies for the following reasons:

1. 11 studies were excluded because the study population did not match our inclusion criteria, which stipulated that presence of an hs‐PDA must be confirmed based on the above mentioned clinical and/or echocardiographic criteria prior to inclusion in the study (Couser 1996; Jannatdoust 2014; Mahony 1982; Mahony 1985; Ment 1985; Ment 1988; Ment 1994a; Ment 1994b; Mullett 1982; Rennie 1986; Sangtawesin 2008);

2. Nine studies were excluded for starting pharmacotherapy for PDA later than the seven days which we used to define 'early management' (Babaei 2018; Clyman 2019; Monset Couchard 1983; Nestrud 1980; Neu 1981; Rudd 1983; Yanagi 1981; Yeh 1981; Yeh 1982)

3. Three studies duplicated results from previous publications (Betkerur 1981 (renal outcomes, providing no information about PDA closure); Juujrvi 2019 ‐ results of Harkin 2016, which was excluded as below; Kluckow 2012 ‐ abstract of Kluckow 2014);

4. Three studies included infants treated for indications outside our criteria (Aranda 2009; Sangtawesin 2006; Weesner 1987);

5. One study included a comparator group which was not expectant management (Salama 2008);

6. One study used prophylactic treatment rather than diagnosing PDA prior to study inclusion (Harkin 2016);

7. One study was a cross‐over RCT, therefore effect of interventions on clinical outcomes could not be determined (Osborn 2003);

8. One study included no data on early management of PDA (Peckham 1984).

For further details see Characteristics of excluded studies

Risk of bias in included studies

For the summary of the authors' judgements on the risk of bias in individual studies, please see Figure 2 and Figure 3. One study was only available as a conference abstract with limited details and the corresponding author could not be contacted for further information (Knight 2011). Thus, we were unable to make a judgement on the risk of bias in this study due to limited information.

2.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

3.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Allocation

The overall risk of selection bias was low. Randomization and allocation procedures were clearly described in nine studies (CTRI/2009/091/000041; DeWaal 2020; EL‐Khuffash 2020; Gersony 1983; Kluckow 2014; Krauss 1989; Lin 2012; Merritt 1981; Sosenko 2012). There was insufficient information on randomization and allocation to allow us to make a judgement in four studies (Bagnoli 2013; Kaapa 1983; Knight 2011; Van Overmeire 2001). Ghanem 2010 included no details of the randomization process and appeared to randomize infants who had not yet been diagnosed with PDA, although they were only given treatment or placebo when a diagnosis of PDA was established.

Blinding

Blinding processes were clearly described in five studies (DeWaal 2020; EL‐Khuffash 2020; Gersony 1983; Kluckow 2014; Sosenko 2012). There was insufficient information in four studies (Bagnoli 2013; Ghanem 2010; Knight 2011; Lin 2012). Five studies provided either no information about blinding, or the staff giving treatments or assessing outcomes were not blinded (CTRI/2009/091/000041; Kaapa 1983; Krauss 1989; Merritt 1981; Van Overmeire 2001).

Incomplete outcome data

We judged 10 studies to be at low risk of attrition bias (Bagnoli 2013; DeWaal 2020; EL‐Khuffash 2020; Gersony 1983; Ghanem 2010; Kluckow 2014; Krauss 1989; Lin 2012; Sosenko 2012; Van Overmeire 2001). Information was insufficient in three studies (CTRI/2009/091/000041; Knight 2011; Merritt 1981). We judged Kaapa 1983 to be at high risk of attrition bias, and the study reported incomplete outcomes.

Selective reporting

We judged four studies to be at low risk of reporting bias (DeWaal 2020; EL‐Khuffash 2020; Kluckow 2014; Sosenko 2012). The other studies provided incomplete information to allow us to make a judgement to be made on the risk of reporting bias.

Other potential sources of bias

We judged that Merritt 1981 had a significant risk of bias from other sources as the exclusion criteria were separately applied to the treatment group, making the treatment group and control group systematically different.

Effects of interventions

See: Table 1; Table 2

Summary of findings 1. Early treatment compared to expectant management for preterm infants.

Early treatment compared to expectant management for preterm infants
Patient or population: preterm infants Setting: Neonatal Intensive Care Unit Intervention: early treatment (treatment of a hemodynamically significant PDA by seven days of age) Comparison: expectant management (no treatment of a hemodynamically significant PDA in the first seven days after birth)
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)
	Risk with Expectant Management	Risk with Early Treatment
All‐cause mortality during hospital stay	Study population		RR 0.80 (0.46 to 1.39)	500 (6 RCTs)	⊕⊕⊕⊝ MODERATE 1
	109 per 1,000	87 per 1,000 (50 to 151)
Proportion of infants requiring surgical PDA ligation or transcatheter occlusion	Study population		RR 1.08 (0.65 to 1.80)	432 (4 RCTs)	⊕⊝⊝⊝ VERY LOW 1 2 3
	145 per 1,000	156 per 1,000 (94 to 261)
Proportion of infants receiving any pharmacotherapy for a hemodynamically significant PDA	Study population		RR 2.30 (1.86 to 2.83)	232 (2 RCTs)	⊕⊕⊝⊝ LOW 4 5
	430 per 1,000	989 per 1,000 (799 to 1,000)
Chronic lung disease	Study population		RR 0.90 (0.62 to 1.29)	339 (4 RCTs)	⊕⊕⊕⊝ MODERATE 1
	263 per 1,000	237 per 1,000 (163 to 339)
Severe Intraventricular hemorrhage (grades III and IV)	Study population		RR 0.83 (0.32 to 2.16)	171 (2 RCTs)	⊕⊕⊝⊝ LOW 6
	95 per 1,000	79 per 1,000 (30 to 206)
Necrotizing enterocolitis (NEC; stage 2 or greater)	Study population		RR 2.34 (0.86 to 6.41)	473 (5 RCTs)	⊕⊕⊝⊝ LOW 2 7
	29 per 1,000	68 per 1,000 (25 to 185)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: Confidence interval; RR: Risk ratio; OR: Odds ratio
GRADE Working Group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect

¹ The CI included appreciable benefit and harm, therefore the certainty of evidence was rated down by one level for imprecision.

² > 50% of meta‐analytic weight to studies with high or unclear risk of bias in one of sequence generation, allocation concealment, or blinding. Therefore, the certainty of evidence was rated by one level for risk of bias.

³ I² value of 59% suggests there is moderate to substantial heterogeneity that cannot be explained by subgroup differences (test for subgroup differences P = 0.59). Therefore, the certainty of evidence was rated down by one level for inconsistency.

⁴ There was high risk of bias for blinding in one study and low risk of bias across multiple domains for one study.

⁵ Did not meet the optimal information size for detecting a 25% difference in benefit or harm (assuming a two‐sided alpha of 0.05 with 80% power). Therefore, the certainty of evidence was rated down by one level for imprecision.

⁶ As there were few events from two small‐sample RCTs and the CI included appreciable benefit and harm, the certainty of evidence was rated down by two levels for imprecision.

⁷ The CI included appreciable benefit favoring expectant management but crossed the threshold for no difference. Therefore, the certainty of evidence was rated down by one level for imprecision.

Summary of findings 2. Very early treatment compared to expectant management for preterm infants.

Very early treatment compared to expectant management for preterm infants
Patient or population: preterm infants Setting: Neonatal Intensive Care Unit Intervention: very early treatment (treatment of a hemodynamically significant PDA by 72 hours of age) Comparison: expectant management (no treatment of a hemodynamically significant PDA in the first 72 hours after birth)
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)
	Risk with Expectant Management	Risk with Very early treatment
All‐cause mortality during hospital stay	Study population		RR 0.94 (0.58 to 1.53)	384 (7 RCTs)	⊕⊕⊕⊝ MODERATE 1
	168 per 1,000	158 per 1,000 (98 to 258)
Proportion of infants requiring surgical PDA ligation or transcatheter occlusion	Study population		RR 0.88 (0.36 to 2.17)	293 (5 RCTs)	⊕⊕⊕⊝ MODERATE 1
	60 per 1,000	53 per 1,000 (22 to 130)
Proportion of infants receiving any pharmacotherapy for a hemodynamically significant PDA	Study population		RR 1.64 (1.31 to 2.05)	156 (4 RCTs)	⊕⊝⊝⊝ VERY LOW 2 3
	329 per 1,000	540 per 1,000 (431 to 675)
Chronic lung disease	Study population		RR 0.83 (0.63 to 1.08)	384 (7 RCTs)	⊕⊕⊝⊝ LOW 4 5
	378 per 1,000	313 per 1,000 (238 to 408)
Severe Intraventricular hemorrhage (grades III and IV)	Study population		RR 0.64 (0.21 to 1.93)	240 (4 RCTs)	⊕⊕⊕⊝ MODERATE 1
	66 per 1,000	42 per 1,000 (14 to 128)
Necrotizing enterocolitis (NEC; stage 2 or greater)	Study population		RR 1.08 (0.53 to 2.21)	332 (5 RCTs)	⊕⊕⊕⊝ MODERATE 1
	83 per 1,000	89 per 1,000 (44 to 183)
Duration of hospitalization		MD 5.35 lower (9.23 lower to 1.47 lower)	‐	240 (4 RCTs)	⊕⊕⊝⊝ LOW 6 7
Neurodevelopmental outcomes (moderate/severe cognitive delay at 18 to 24 months)	Study population		RR 0.27 (0.03 to 2.31)	79 (1 RCT)	⊕⊕⊝⊝ LOW 8
	98 per 1,000	26 per 1,000 (3 to 225)
Neurodevelopmental outcomes (moderate/severe motor delay at 18 to 24 months)	Study population		RR 0.54 (0.05 to 5.71)	79 (1 RCT)	⊕⊕⊝⊝ LOW 8
	49 per 1,000	26 per 1,000 (2 to 279)
Neurodevelopmental outcomes (moderate/severe language delay at 18 to 24 months)	Study population		RR 0.54 (0.10 to 2.78)	79 (1 RCT)	⊕⊕⊝⊝ LOW 8
	98 per 1,000	53 per 1,000 (10 to 271)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: Confidence interval; RR: Risk ratio; OR: Odds ratio;
GRADE Working Group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect

¹ The CI included appreciable benefit and harm, therefore, the certainty of evidence was rated down by one level for imprecision.

² > 50% of meta‐analytic weight from studies with high risk of bias in blinding personnel and outcome assessors

³ I² value of 90% suggesting substantial heterogeneity that could not be explained by subgroup differences. Therefore, the certainty of evidence was rated down by two levels for inconsistency.

⁴ I² value of 48% suggested there was moderate heterogeneity which is partly explained by subgroup differences (test for subgroup differences, P = 0.04). The certainty of evidence was therefore rated down by one level.

⁵ The CI included appreciable benefit favoring very early treatment but crossed the threshold for no difference. Therefore, the certainty of evidence was rated down by one level.

⁶ I² value of 57% suggested there was moderate to serious heterogeneity which is likely explained by subgroup differences. There was minimal overlap of CIs between the ibuprofen only subgroup vs the indomethacin/ibuprofen subgroup. The certainty of evidence was therefore rated down by one level.

⁷ The CI includes benefit that may not be clinically meaningful (1.5 days reduction in hospitalization) as well as benefit that may substantially meaningful (9 days reduction in hospitalization). Therefore the certainty of evidence was rated down by one level for imprecision

⁸ As there were few events in a small sample single RCT and the CI included appreciable benefit and harm, the certainty of evidence was rated down by two levels for imprecision.

Comparison 1. Early treatment versus expectant management

See Summary of findings table 1.

Primary outcome

All‐cause mortality during hospital stay

(Analysis 1.1)

1.1. Analysis.

Comparison 1: Early treatment vs expectant management, Outcome 1: All‐cause mortality during hospital stay

Six studies (n = 600) reported on this outcome. There were no statistically significant differences between the early treatment and expectant management groups in all‐cause mortality (typical RR 0.80, 95% CI 0.46 to 1.39; typical RD ‐0.02; 95% CI ‐0.07 to 0.03; I² = 0% for RR and for RD) (Figure 4). We rated the certainty of evidence as moderate.

4.

Forest plot of comparison: 1 Early treatment vs expectant management, outcome: 1.1 All‐cause mortality during hospital stay.

Subgroup analyses

Three out of the six studies used IV indomethacin, two studies used IV ibuprofen and one study used oral ibuprofen. No statistically significant differences in all‐cause mortality were noted in either the indomethacin (typical RR 0.95, 95% CI 0.45 to 1.99; typical RD ‐0.01; 95% CI ‐0.09 to 0.08; I² = 0% for RR and for RD), or the ibuprofen subgroups (typical RR 0.65, 95% CI 0.28, to 1.50; typical RD ‐0.03; 95% CI ‐0.08 to 0.03; I² = 0% for RR and for RD).

Secondary outcomes

Proportion of infants requiring surgical PDA ligation or transcatheter occlusion

(Analysis 1.2)

1.2. Analysis.

Comparison 1: Early treatment vs expectant management, Outcome 2: Surgical PDA ligation or transcatheter occlusion

Four studies (n = 432) reported on this outcome. There were no statistically significant differences between the early treatment and expectant management groups in surgical PDA ligation (typical RR 1.08, 95% CI 0.65 to 1.80; typical RD ‐0.03; 95% CI ‐0.09 to 0.03; I² = 59% for RR and 76% for RD, respectively). We rated the certainty of evidence as very low.

Subgroup analyses

One out of the four studies used IV indomethacin, two studies used IV ibuprofen and one study used oral ibuprofen. No statistically significant differences in surgical PDA ligation were noted in either the indomethacin (typical RR 0.74, 95% CI 0.17 to 3.17; typical RD ‐0.02; 95% CI ‐0.10 to 0.06; test for heterogeneity not applicable), or the ibuprofen subgroups (typical RR 1.14, 95% CI 0.66, to 1.96; typical RD ‐0.04; 95% CI ‐0.12 to 0.04; I² = 71% for RR and 84% for RD, respectively)

Proportion of infants receiving any pharmacotherapy for a hemodynamically significant PDA (defined as any receipt of pharmacotherapy for a hemodynamically significant PDA any time during hospital stay)

(Analysis 1.3)

1.3. Analysis.

Comparison 1: Early treatment vs expectant management, Outcome 3: Receipt of any pharmacotherapy for a hemodynamically significant PDA

Two studies (n = 232) reported on this outcome. There was a statistically significant difference between the early treatment and expectant management groups with regards to receipt of any pharmacotherapy for hs‐PDA favoring expectant management (typical RR 2.30, 95% CI 1.86 to 2.83; typical RD 0.57; 95% CI 0.48 to 0.66; I² = 0% for RR and for RD) (Figure 5). We rated the certainty of evidence as low.

5.

Forest plot of comparison: 1 Early treatment vs expectant management, outcome: 1.3 Receipt of any pharmacotherapy for a hemodynamically significant PDA.

Subgroup analyses

One study used IV indomethacin, and the other IV ibuprofen. Both studies demonstrated statistically significant differences in receipt of any pharmacotherapy for hs‐PDA favoring the expectant management group.

Proportion of infants receiving rescue medical treatment (COX inhibitor or acetaminophen dosing, or both) for a hemodynamically significant PDA

(Analysis 1.4)

1.4. Analysis.

Comparison 1: Early treatment vs expectant management, Outcome 4: Receipt of rescue medical treatment for a hemodynamically significant PDA

Two studies (n = 232) reported on this outcome. There were no statistically significant differences between the early treatment and expectant management groups (typical RR 0.63, 95% CI 0.27 to 1.48; typical RD ‐0.02; 95% CI ‐0.06 to 0.02; I² = 0% for RR and for RD).

Subgroup analyses

One study used IV indomethacin, and the other IV ibuprofen. None of the studies demonstrated statistically significant differences in receipt of rescue medical treatment for a hemodynamically significant PDA.

Chronic lung disease (CLD; defined as oxygen requirement at 36 weeks' postmenstrual age).

(Analysis 1.5)

1.5. Analysis.

Comparison 1: Early treatment vs expectant management, Outcome 5: Chronic lung disease

Four studies (n = 339) reported on this outcome. There were no statistically significant differences between the early treatment and expectant management groups in CLD (typical RR 0.90, 95% CI 0.62 to 1.29; typical RD ‐0.03; 95% CI ‐0.10 to 0.03; I² = 0% for RR and for RD). We rated the certainty of evidence as moderate.

Subgroup analyses

Two out of the four studies used IV indomethacin, one used IV ibuprofen, and one study used oral ibuprofen. No statistically significant differences in CLD were noted in either the indomethacin (typical RR 0.84, 95% CI 0.52 to 1.37; typical RD ‐0.06; 95% CI ‐0.20 to 0.07; I² = 0% for RR and for RD), or the ibuprofen subgroups (typical RR 0.97, 95% CI 0.56 to 1.69; typical RD ‐0.03; 95% CI ‐0.10 to 0.05; I² = 0% for RR and for RD).

Duration of ventilator support (days)

(Analysis 1.6)

1.6. Analysis.

Comparison 1: Early treatment vs expectant management, Outcome 6: Duration of ventilator support (days)

One study (n = 127) that used IV indomethacin reported on this outcome. There was no statistically significant difference between the early treatment and expectant management groups with respect to days on ventilator support (MD ‐13.40 days; 95% CI ‐30.15 to 3.35). The test for heterogeneity was not applicable.

Duration of need for supplementary oxygen (days)

(Analysis 1.7)

1.7. Analysis.

Comparison 1: Early treatment vs expectant management, Outcome 7: Duration of need for supplementary oxygen (days)

One study (n = 127) that used IV indomethacin reported on this outcome. There was no statistically significant difference between the early treatment and expectant management groups with respect to days on supplementary oxygen (MD ‐8.3 days; 95% CI ‐43.27 to 26.67). The test for heterogeneity was not applicable.

Postnatal corticosteroid use for CLD any time during hospital stay

(Analysis 1.8)

1.8. Analysis.

Comparison 1: Early treatment vs expectant management, Outcome 8: Postnatal corticosteroid use for CLD any time during hospital stay

One study (n = 105) that used IV ibuprofen reported on this outcome. There was no statistically significant difference between the early treatment and expectant management groups with respect to postnatal corticosteroid use for CLD (typical RR 1.13, 95% CI 0.37 to 3.49; typical RD 0.01; 95% CI ‐0.10 to 0.13). The test for heterogeneity was not applicable.

Pneumothorax

(Analysis 1.9)

1.9. Analysis.

Comparison 1: Early treatment vs expectant management, Outcome 9: Pneumothorax

Two studies (n = 146) reported on this outcome. There was no statistically significant difference between the early treatment and expectant management groups in pneumothorax (typical RR 0.88, 95% CI 0.32 to 2.42; typical RD 0.00; 95% CI ‐0.09 to 0.09; I² = 0% for RR and for RD).

Subgroup analyses

One study used IV indomethacin, and the other IV ibuprofen. None of the studies showed any significant differences in pneumothorax between the two groups.

Intraventricular hemorrhage (IVH) (grades I to IV)

(Analysis 1.10)

1.10. Analysis.

Comparison 1: Early treatment vs expectant management, Outcome 10: Intraventricular hemorrhage (IVH) (grades I to IV)

Three studies (n = 234) reported on this outcome. There were no statistically significant differences between the early treatment and expectant management groups in IVH (typical RR 0.91, 95% CI 0.34 to 2.48; typical RD 0.01; 95% CI ‐0.20 to 0.14; I² = 39% for RR and 73% for RD, respectively).

Subgroup analyses

Two out of the three studies used IV indomethacin and one used oral. No statistically significant differences in IVH were noted in either the indomethacin (typical RR 1.19, 95% CI 0.21 to 6.60; typical RD 0.02; 95% CI ‐0.05 to 0.08; I² = 68% for RR and 86% for RD, respectively) or the ibuprofen subgroups (typical RR 0.80, 95% CI 0.24 to 2.72; typical RD ‐0.03; 95% CI ‐0.20 to 0.14; test for heterogeneity not applicable).

Severe IVH (grades III and IV)

(Analysis 1.11)

1.11. Analysis.

Comparison 1: Early treatment vs expectant management, Outcome 11: Severe intraventricular hemorrhage IVH (grades III and IV)

Two studies (n = 171) reported on this outcome. One study used oral ibuprofen, and the other IV ibuprofen. There were no statistically significant differences between the early treatment and expectant management groups in severe IVH (typical RR 0.83, 95% CI 0.32 to 2.16; typical RD ‐0.01; 95% CI ‐0.08 to 0.06; I² = 0% for both RR and RD). We rated the certainty of evidence as low.

Periventricular leukomalacia (PVL; any grade)

(Analysis 1.12)

1.12. Analysis.

Comparison 1: Early treatment vs expectant management, Outcome 12: Periventricular leukomalacia (PVL; any grade)

Three studies (n = 298) reported on this outcome. There were no statistically significant differences between the early treatment and expectant management groups in PVL (typical RR 1.22, 95% CI 0.55 to 2.75; typical RD 0.03; 95% CI ‐0.03 to 0.08; I² = 0% for both RR and RD).

Subgroup analyses

One out of the three studies used IV indomethacin, one used oral ibuprofen, and one study used IV ibuprofen. No statistically significant differences in PVL were noted in either the indomethacin (typical RR 2.46, 95% CI 0.50 to 12.22; typical RD 0.05; 95% CI ‐0.03 to 0.13; test for heterogeneity not applicable) or the ibuprofen subgroups (typical RR 0.97, 95% CI 0.38 to 2.46; typical RD 0.00; 95% CI ‐0.08 to 0.09; I² = 0% for both RR and RD).

Necrotizing enterocolitis (NEC; stage 2 or greater)

(Analysis 1.13)

1.13. Analysis.

Comparison 1: Early treatment vs expectant management, Outcome 13: Necrotizing enterocolitis (NEC; stage 2 or greater)

Five studies (n = 473) reported on this outcome. There were no statistically significant differences between the early treatment and expectant management groups in NEC (typical RR 2.34, 95% CI 0.86 to 6.41; typical RD 0.04; 95% CI 0.01 to 0.08; I² = 10% for RR and 51% for RD, respectively). We rated the certainty of evidence as low.

Subgroup analyses

Two out of the five studies used IV indomethacin, two used IV ibuprofen, and one study used oral ibuprofen. No statistically significant differences in NEC were noted in either the indomethacin (typical RR 1.56, 95% CI 0.28 to 8.80; typical RD 0.02; 95% CI ‐0.04 to 0.08; I² = 49% for RR and 69% for RD, respectively) or the ibuprofen subgroups (typical RR 2.89, 95% CI 0.84 to 9.95; typical RD 0.06; 95% CI 0.01 to 0.11; I² = 7% for RR and 51% for RD, respectively).

Spontaneous intestinal perforation

(Analysis 1.14)

1.14. Analysis.

Comparison 1: Early treatment vs expectant management, Outcome 14: Spontaneous intestinal perforation

Three studies (n = 298) reported on this outcome. There were no statistically significant differences between the early treatment and expectant management groups in spontaneous intestinal perforation (typical RR 0.57, 95% CI 0.14 to 2.37; typical RD ‐0.01; 95% CI ‐0.04 to 0.03; I² = 0% for both RR and RD).

Subgroup analyses

One out of the three studies used IV indomethacin, one used oral ibuprofen, and one study used IV ibuprofen. No statistically significant differences in spontaneous intestinal perforation were noted in either the indomethacin (typical RR 0.98, 95% CI 0.06 to 15.40; typical RD ‐0.00; 95% CI ‐0.04 to 0.04; test for heterogeneity not applicable), or the ibuprofen subgroups (typical RR 0.47, 95% CI 0.09 to 2.47; typical RD ‐0.01; 95% CI ‐0.06 to 0.04; I² = 0% for both RR and RD).

Time to full enteral feeds (postnatal age at time of achieving full enteral feeds)

(Analysis 1.15)

1.15. Analysis.

Comparison 1: Early treatment vs expectant management, Outcome 15: Time to full enteral feeds (postnatal age at time of achieving full enteral feeds)

One study (n = 66) that used oral ibuprofen reported on this outcome. There was a statistically significant difference between the early treatment and expectant management groups with regards to time to full enteral feeds favoring early treatment (MD ‐4.8 days; 95% CI ‐8.55 to ‐1.05). The test for heterogeneity was not applicable.

Time to regain birth weight (days)

(Analysis 1.16)

1.16. Analysis.

Comparison 1: Early treatment vs expectant management, Outcome 16: Time to regain birth weight (days)

One study (n = 66) that used oral ibuprofen reported on this outcome. There was a statistically significant difference between the early treatment and expectant management groups with regards to time to regain birth weight favoring early treatment (MD ‐3.4 days; 95% CI ‐6.13 to ‐0.67). The test for heterogeneity was not applicable.

Severe retinopathy of prematurity (ROP) (stage 3 or greater) (according to the international classification of ROP)

(Analysis 1.17)

1.17. Analysis.

Comparison 1: Early treatment vs expectant management, Outcome 17: Severe retinopathy of prematurity (ROP)

Two studies (n = 146) reported on this outcome. There was no statistically significant difference between the early treatment and expectant management groups in severe ROP (typical RR 1.30, 95% CI 0.44 to 3.84; typical RD ‐0.01; 95% CI ‐0.10 to 0.09; I² = 14% for RR and 62% for RD, respectively).

Subgroup analyses

One study used IV indomethacin, and the other IV ibuprofen. None of the studies showed any statistically significant differences in severe ROP between the two groups.

Definite sepsis (clinical symptoms and signs of sepsis and a positive bacterial culture in a specimen obtained from normally sterile fluids or tissue obtained at postmortem)

(Analysis 1.18)

1.18. Analysis.

Comparison 1: Early treatment vs expectant management, Outcome 18: Definite sepsis

Four studies (n = 339) reported on this outcome. There were no statistically significant differences between the early treatment and expectant management groups in definite sepsis (typical RR 0.97, 95% CI 0.68 to 1.38; typical RD 0.00; 95% CI ‐0.08 to 0.09; I² = 0% for both RR and RD).

Subgroup analyses

Two of the four studies used IV indomethacin, one used oral ibuprofen, and one study used IV ibuprofen. No statistically significant differences in definite sepsis were noted in either the indomethacin (typical RR 1.16, 95% CI 0.56 to 2.41; typical RD 0.02; 95% CI ‐0.09 to 0.13; I² = 0% for both RR and RD) or the ibuprofen subgroups (typical RR 0.92, 95% CI 0.61 to 1.38; typical RD ‐0.02; 95% CI ‐0.15 to 0.11; I² = 0% for both RR and RD).

Oliguria (defined as less than 1 mL/kg/hour)

(Analysis 1.19)

1.19. Analysis.

Comparison 1: Early treatment vs expectant management, Outcome 19: Oliguria

Two studies (n = 261) reported on this outcome. There was a statistically significant difference between the early treatment and expectant management groups in oliguria favoring the expectant management group (typical RR 6.41, 95% CI 2.13 to 19.26; typical RD 0.23; 95% CI 0.15 to 0.31; I² = 48% for RR and 49% for RD, respectively).

Subgroup analyses

One study used IV indomethacin, and the other IV ibuprofen. Both studies demonstrated a statistically significant difference between the early treatment and expectant management groups in oliguria favoring the expectant management group (IV indomethacin: typical RR 4.59, 95% CI 1.39 to 15.21; typical RD 0.17; 95% CI 0.06 to 0.29; test for heterogeneity not applicable; IV ibuprofen: typical RR 39.00, 95% CI 2.40 to 633.01; typical RD 0.28; 95% CI 0.17 to 0.39; test for heterogeneity not applicable).

None of the studies included in this comparison reported on the proportion of infants receiving repeat courses of pharmacotherapy for a persistent hs‐PDA, pulmonary hemorrhage, pulmonary hypertension, gastrointestinal bleeding, duration of hospitalization or any neurodevelopmental outcomes.

Comparison 2. Very early treatment versus expectant management

See Summary of findings table 2.

Primary outcome

All‐cause mortality during hospital stay

(Analysis 2.1)

2.1. Analysis.

Comparison 2: Very early treatment vs expectant management, Outcome 1: All‐cause mortality during hospital stay

Seven studies (n = 384) reported on this outcome. There were no statistically significant differences between the very early treatment and expectant management groups in all‐cause mortality (typical RR 0.94, 95% CI 0.58 to 1.53; typical RD ‐0.03; 95% CI ‐0.09 to 0.04; I² = 16% for RR and 26% for RD, respectively) (Figure 6). We rated the certainty of evidence as moderate.

6.

Forest plot of comparison: 2 Very early treatment vs expectant management, outcome: 2.1 All‐cause mortality during hospital stay.

Subgroup analyses

Four out of the seven studies used indomethacin, two studies used ibuprofen and one study was a two‐center study that used indomethacin in one center and ibuprofen in the other. No significant differences in all‐cause mortality were noted in the indomethacin only subgroup (typical RR 0.92, 95% CI 0.47 to 1.80; typical RD ‐0.04; 95% CI ‐0.14 to 0.06; I² = 26% for RR and 37% for RD, respectively), the ibuprofen only subgroup (typical RR 1.46, 95% CI 0.58, to 3.67; typical RD 0.02; 95% CI ‐0.09 to 0.13; I² = 11% for RR and 45% for RD, respectively), or the subgroup with use of either indomethacin or ibuprofen (typical RR 0.53, 95% CI 0.17, to 1.60; typical RD ‐0.10; 95% CI ‐0.27 to 0.07; test for heterogeneity not applicable).

Secondary outcomes

Proportion of infants requiring surgical PDA ligation or transcatheter occlusion

(Analysis 2.2)

2.2. Analysis.

Comparison 2: Very early treatment vs expectant management, Outcome 2: Surgical PDA ligation or transcatheter occlusion

Five studies (n = 293) reported on this outcome. There were no statistically significant differences between the very early treatment and expectant management groups in surgical PDA ligation or transcatheter occlusion (typical RR 0.88, 95% CI 0.36 to 2.17; typical RD ‐0.01; 95% CI ‐0.05 to 0.02; I² = 0% for both RR and RD). We rated the certainty of evidence as moderate.

Three out of the five studies used indomethacin, one study used intravenous ibuprofen and one was a two‐center study that used indomethacin in one center and ibuprofen in the other. No significant differences in surgical PDA ligation or transcatheter occlusion were noted in the indomethacin only subgroup (typical RR 0.54, 95% CI 0.07 to 3.93; typical RD ‐0.02; 95% CI ‐0.07 to 0.03; I² = 0% for RR and for RD), the ibuprofen only subgroup (typical RR 1.00, 95% CI 0.36, to 2.75; typical RD 0.00; 95% CI ‐0.20 to 0.20; test for heterogeneity not applicable), or the subgroup with use of either indomethacin or ibuprofen (typical RR not estimable; typical RD 0.00; 95% CI ‐0.05 to 0.05; test for heterogeneity not applicable).

Proportion of infants receiving any pharmacotherapy for a hemodynamically significant PDA (defined as any receipt of pharmacotherapy for a hemodynamically significant PDA any time during hospital stay)

(Analysis 2.3)

2.3. Analysis.

Comparison 2: Very early treatment vs expectant management, Outcome 3: Receipt of any pharmacotherapy for a hemodynamically significant PDA

Four studies (n = 156) reported on this outcome. There was a statistically significant difference between the very early treatment and expectant management groups with respect to receipt of any pharmacotherapy for a hemodynamically significant PDA favoring the expectant management group (typical RR 1.64, 95% CI 1.31 to 2.05; typical RD 0.69; 95% CI 0.60 to 0.79; I² = 87% for RR and 90% for RD, respectively). We rated the certainty of evidence as very low.

Subgroup analyses

Three out of the four studies used indomethacin, and one study used intravenous ibuprofen. Statistically significant differences in receipt of any pharmacotherapy for a hemodynamically significant PDA were noted for both the indomethacin subgroup (typical RR 1.42, 95% CI 1.10 to 1.82; typical RD 0.70; 95% CI 0.59 to 0.81; I² = 88% for RR and 93% for RD, respectively) as well as the ibuprofen group (typical RR 2.90, 95% CI 1.77 to 4.76; typical RD 0.67; 95% CI 0.49 to 0.84; test for heterogeneity not applicable).

Proportion of infants receiving repeat courses of pharmacotherapy (repeated COX inhibitor or acetaminophen dosing, or both) for a persistent hemodynamically significant PDA

(Analysis 2.4)

2.4. Analysis.

Comparison 2: Very early treatment vs expectant management, Outcome 4: Receipt of repeat courses of pharmacotherapy for a persistent hemodynamically significant PDA

Three studies (n = 133) reported on this outcome. There were no statistically significant differences between the very early treatment and expectant management groups with respect to receipt of repeat courses of pharmacotherapy for a persistent hemodynamically significant PDA (typical RR 1.99, 95% CI 0.50 to 7.94; typical RD 0.05; 95% CI ‐0.03 to 0.14; I² = 12% for RR and 30% for RD, respectively).

Subgroup analyses

Two out of the three studies used indomethacin, and one study used ibuprofen. No significant differences in receipt of repeat courses of pharmacotherapy were noted in the indomethacin only (typical RR 1.39, 95% CI 0.29 to 6.66; typical RD 0.00; 95% CI ‐0.13 to 0.13; I² = 26% for RR and 44% for RD, respectively) or the ibuprofen only subgroups (typical RR 7.00, 95% CI 0.38, to 130.26; typical RD ‐0.03; 95% CI ‐0.02 to 0.21; test for heterogeneity not applicable).

Proportion of infants receiving rescue medical treatment (COX inhibitor or acetaminophen dosing, or both) for a hemodynamically significant PDA

(Analysis 2.5)

2.5. Analysis.

Comparison 2: Very early treatment vs expectant management, Outcome 5: Receipt of rescue medical treatment for a hemodynamically significant PDA

Two studies (n = 164) reported on this outcome. There were no statistically significant differences between the very early treatment and expectant management groups (typical RR 0.54, 95% CI 0.28 to 1.04; typical RD ‐0.03; 95% CI ‐0.10 to 0.04; I² = 0% for RR and 73% for RD, respectively).

Subgroup analyses

One study used IV indomethacin, and the other was a two‐center study that used indomethacin in one center and ibuprofen in the other. None of the studies demonstrated statistically significant differences in receipt of rescue medical treatment for a hemodynamically significant PDA.

Chronic lung disease (CLD; defined as oxygen requirement at 36 weeks' postmenstrual age)

(Analysis 2.6)

2.6. Analysis.

Comparison 2: Very early treatment vs expectant management, Outcome 6: Chronic lung disease

Seven studies (n = 384) reported on this outcome. There were no statistically significant differences between the very early treatment and expectant management groups in CLD (typical RR 0.83, 95% CI 0.63 to 1.08; typical RD ‐0.05; 95% CI ‐0.13 to 0.04; I² = 48% for RR and 68% for RD, respectively). We rated the certainty of evidence as low.

Four out of the seven studies used indomethacin, two studies used ibuprofen and one study was a two‐center study that used indomethacin in one center and ibuprofen in the other. There was a statistically significant difference between the very early treatment and expectant management groups in the ibuprofen only subgroup favoring very early treatment (typical RR 0.54, 95% CI 0.35, to 0.83; typical RD ‐0.26; 95% CI ‐0.42 to ‐0.09 I² = 0% for both RR and RD). No significant differences in CLD were noted in the indomethacin only subgroup (typical RR 1.06, 95% CI 0.61 to 1.83; typical RD 0.02; 95% CI ‐0.09 to 0.12; I² = 43% for RR and 70% for RD, respectively) or the subgroup with use of either indomethacin or ibuprofen (typical RR 1.12, 95% CI 0.71, to 1.75; typical RD 0.06; 95% CI ‐0.17 to 0.29; test for heterogeneity not applicable).

Duration of ventilator support (days)

(Analysis 2.7)

2.7. Analysis.

Comparison 2: Very early treatment vs expectant management, Outcome 7: Duration of ventilator support (days)

Six studies (n = 253) reported on this outcome. There was a statistically significant difference between the very early treatment and expectant management groups with respect to duration of ventilator support favoring very early treatment (MD ‐0.44 days; 95% CI ‐0.75 to ‐0.16; I² = 0%).

Subgroup analyses

Three out of the six studies used indomethacin, two studies used ibuprofen and one study was a two‐center study that used indomethacin in one center and ibuprofen in the other. Statistically significant differences in duration of ventilator support favoring very early treatment were noted in the indomethacin only subgroup (MD ‐1.37 days; 95% CI ‐2.70 to ‐0.04; I² = 0%), and the ibuprofen only subgroup (MD ‐0.41 days; 95% CI ‐0.71 to ‐0.11; I² = 0%) but not in the subgroup with use of either indomethacin or ibuprofen (MD 6.00 days; 95% CI ‐6.69 to 18.69; test for heterogeneity not applicable).

Duration of need for supplementary oxygen (days)

(Analysis 2.8)

2.8. Analysis.

Comparison 2: Very early treatment vs expectant management, Outcome 8: Duration of need for supplementary oxygen (days)

Three studies (n = 127) reported on this outcome. There were no statistically significant differences between the very early treatment and expectant management groups with respect to duration of need for supplementary oxygen (MD ‐1.73 days; 95% CI ‐3.99 to 0.53; I² = 0%).

Subgroup analyses

Two out of the three studies used indomethacin and one study used ibuprofen. There were no statistically significant differences between the very early treatment and expectant management groups with respect to duration of need for supplementary oxygen either in the indomethacin (MD ‐1.61 days; 95% CI ‐3.95 to 0.74; I² = 0%) or ibuprofen (MD ‐3.30 days; 95% CI ‐11.83 to 5.23; test for heterogeneity not applicable) subgroups.

Postnatal corticosteroid use for CLD any time during hospital stay

(Analysis 2.9)

2.9. Analysis.

Comparison 2: Very early treatment vs expectant management, Outcome 9: Postnatal corticosteroid use for CLD any time during hospital stay

Three studies (n = 224) reported on this outcome. There was a statistically significant difference between the very early treatment and expectant management groups with respect to postnatal corticosteroid use for CLD favoring expectant management (typical RR 2.11, 95% CI 1.12 to 3.97; typical RD 0.11; 95% CI 0.01 to 0.20; I² = 0% for both RR and RD).

Subgroup analyses

One study used indomethacin, one study used ibuprofen and the other was a two‐center study that used indomethacin in one center and ibuprofen in the other. None of the studies individually demonstrated a statistically significant difference between the very early treatment and expectant management groups with respect to postnatal corticosteroid use for CLD (Indomethacin only subgroup: typical RR 1.91, 95% CI 0.60 to 6.08; typical RD 0.08; 95% CI ‐0.06 to 0.21; test for heterogeneity not applicable; Ibuprofen only subgroup: typical RR 2.50, 95% CI 0.53 to 11.89; typical RD 0.10; 95% CI ‐0.06 to 0.26; test for heterogeneity not applicable; Indomethacin or ibuprofen subgroup: typical RR 2.11, 95% CI 0.89 to 5.02; typical RD 0.18; 95% CI ‐0.02 to 0.38; test for heterogeneity not applicable).

Pulmonary hemorrhage

(Analysis 2.10)

2.10. Analysis.

Comparison 2: Very early treatment vs expectant management, Outcome 10: Pulmonary hemorrhage

Five studies (n = 332) reported on this outcome. There were no statistically significant differences between the very early treatment and expectant management groups in pulmonary hemorrhage (typical RR 0.58, 95% CI 0.30 to 1.11; typical RD ‐0.03; 95% CI ‐0.08 to 0.02; I² = 0% for RR and 26% for RD, respectively).

Subgroup analyses

Two out of the five studies used indomethacin, two studies used ibuprofen and one study was a two‐center study that used indomethacin in one center and ibuprofen in the other. No significant differences in pulmonary hemorrhage were noted in the indomethacin only subgroup (typical RR 0.59, 95% CI 0.22 to 1.53; typical RD ‐0.04; 95% CI ‐0.15 to 0.07; I² = 62% for RR and 75% for RD, respectively), the ibuprofen only subgroup (typical RR 0.59, 95% CI 0.24, to 1.49; typical RD ‐0.02; 95% CI ‐0.10 to 0.06; I² = 0%), or the subgroup with use of either indomethacin or ibuprofen (typical RR 0.35, 95% CI 0.01, to 8.36; typical RD ‐0.03; 95% CI ‐0.10 to 0.05; test for heterogeneity not applicable).

Intraventricular hemorrhage (IVH) (grades I to IV)

(Analysis 2.11)

2.11. Analysis.

Comparison 2: Very early treatment vs expectant management, Outcome 11: Intraventricular hemorrhage (IVH) (grades I to IV)

Three studies (n = 200) reported on this outcome. There were no statistically significant differences between the very early treatment and expectant management groups in IVH (typical RR 0.70, 95% CI 0.44 to 1.10; typical RD ‐0.07; 95% CI ‐0.16 to 0.02; I² = 0% for RR and 2% for RD, respectively).

Subgroup analyses

Two out of the 3 studies used indomethacin and 1 study used ibuprofen. No significant differences in IVH were noted in the indomethacin only subgroup (typical RR 0.75, 95% CI 0.27 to 2.10; typical RD ‐0.05; 95% CI ‐0.15 to 0.05; I² = 34% for RR and 2% for RD, respectively) or the ibuprofen only subgroup (typical RR 0.68, 95% CI 0.41, to 1.14; typical RD ‐0.19; 95% CI ‐0.43 to 0.05; test for heterogeneity not applicable).

Severe IVH (grades III and IV)

(Analysis 2.12)

2.12. Analysis.

Comparison 2: Very early treatment vs expectant management, Outcome 12: Severe intraventricular hemorrhage IVH (grades III and IV)

Four studies (n = 240) reported on this outcome. There were no statistically significant differences between the very early treatment and expectant management groups in severe IVH (typical RR 0.64, 95% CI 0.21 to 1.93; typical RD ‐0.02; 95% CI ‐0.07 to 0.04; I² = 0% for both RR and RD). We rated the certainty of evidence as moderate.

Subgroup analyses

One out of the four studies used indomethacin, two studies used ibuprofen and one study was a two‐center study that used indomethacin in one center and ibuprofen in the other. No significant differences in severe intraventricular hemorrhage were noted in the indomethacin only subgroup (typical RR 1.00, 95% CI 0.07 to 15.00; typical RD 0.00; 95% CI ‐0.12 to 0.12; test for heterogeneity not applicable), the ibuprofen only subgroup (typical RR 0.67, 95% CI 0.11 to 3.98; typical RD ‐0.01; 95% CI ‐0.08 to 0.05; I² = 0%) or the subgroup with use of either indomethacin or ibuprofen (typical RR 0.53, 95% CI 0.10 to 2.71; typical RD ‐0.05; 95% CI ‐0.18 to 0.08; test for heterogeneity not applicable).

Periventricular leukomalacia (PVL; any grade)

(Analysis 2.13)

2.13. Analysis.

Comparison 2: Very early treatment vs expectant management, Outcome 13: Periventricular leukomalacia (PVL; any grade)

Five studies (n = 332) reported on this outcome. There were no statistically significant differences between the very early treatment and expectant management groups in PVL (typical RR 0.61, 95% CI 0.31 to 1.20; typical RD 0.00; 95% CI ‐0.04 to 0.04; I² = 61% for both RR and RD, respectively).

Subgroup analyses

Two out of the five studies used indomethacin, two used ibuprofen and one study was a two‐center study that used indomethacin in one center and ibuprofen in the other. No significant differences in periventricular leukomalacia were noted in the indomethacin only subgroup (typical RR 1.09, 95% CI 0.29 to 4.10; typical RD 0.00; 95% CI ‐0.07 to 0.07; test for heterogeneity not applicable for RR; I² = 0% for RD) or the ibuprofen only subgroup (typical RR 0.50, 95% CI 0.23 to 1.10; typical RD 0.02; 95% CI ‐0.09 to 0.14; I² = 76% for RR and 90% for RD, respectively). The effect estimates for the subgroup with use of either indomethacin or ibuprofen could not be computed as there were no events in either arm in that subgroup.

Necrotizing enterocolitis (NEC; stage 2 or greater)

(Analysis 2.14)

2.14. Analysis.

Comparison 2: Very early treatment vs expectant management, Outcome 14: Necrotizing enterocolitis (NEC; stage 2 or greater)

Five studies (n = 332) reported on this outcome. There were no statistically significant differences between the very early treatment and expectant management groups in NEC (typical RR 1.08, 95% CI 0.53 to 2.21; typical RD 0.01; 95% CI ‐0.04 to 0.06; I² = 0% for both RR and RD). We rated the certainty of evidence as moderate.

Subgroup analyses

Two out of the five studies used indomethacin, two used ibuprofen and one study was a two‐center study that used indomethacin in one center and ibuprofen in the other. No significant differences in NEC were noted in the indomethacin only subgroup (typical RR 0.80, 95% CI 0.18 to 3.49; typical RD ‐0.01; 95% CI ‐0.08 to 0.06; I² = 0% for both RR and RD), the ibuprofen only subgroup (typical RR 1.01, 95% CI 0.42, to 2.44; typical RD ‐0.01; 95% CI ‐0.13 to 0.11; I² = 2% for RR and 4% for RD, respectively) or the subgroup with use of either indomethacin or ibuprofen (typical RR 3.17, 95% CI 0.35, to 29.07; typical RD 0.06; 95% CI ‐0.05 to 0.17; test for heterogeneity not applicable).

Gastrointestinal bleeding within seven days of the first dose of pharmacotherapy

(Analysis 2.15)

2.15. Analysis.

Comparison 2: Very early treatment vs expectant management, Outcome 15: Gastrointestinal bleeding within seven days of the first dose of pharmacotherapy

Two studies (n = 164) reported on this outcome. One study used indomethacin and the other study was a two‐center study that used indomethacin in one center and ibuprofen in the other. None of the studies reported any event of gastrointestinal bleeding. Therefore, the effect estimates could not be computed for this outcome.

Spontaneous intestinal perforation

(Analysis 2.16)

2.16. Analysis.

Comparison 2: Very early treatment vs expectant management, Outcome 16: Spontaneous intestinal perforation

Four studies (n = 272) reported on this outcome. There were no statistically significant differences between the very early treatment and expectant management groups in spontaneous intestinal perforation (typical RR 0.96, 95% CI 0.15 to 6.36; typical RD 0.00; 95% CI ‐0.03 to 0.03; I² = 0% for both RR and RD).

Subgroup analyses

Two out of the four studies used indomethacin, one study used ibuprofen and one study was a two‐center study that used indomethacin in one center and ibuprofen in the other. No significant differences in spontaneous intestinal perforation were noted in the ibuprofen only subgroup (typical RR 0.50, 95% CI 0.05, to 5.24; typical RD ‐0.03; 95% CI ‐0.13 to 0.07; test for heterogeneity not applicable), or the subgroup with use of either indomethacin or ibuprofen (typical RR 3.17, 95% CI 0.13, to 75.24; typical RD 0.03; 95% CI ‐0.05 to 0.10; test for heterogeneity not applicable). The study using indomethacin did not report any event of spontaneous intestinal perforation. Therefore, the effect estimates could not be computed for this comparison.

Time to full enteral feeds (postnatal age at time of achieving full enteral feeds)

(Analysis 2.17)

2.17. Analysis.

Comparison 2: Very early treatment vs expectant management, Outcome 17: Time to full enteral feeds (postnatal age at time of achieving full enteral feeds)

Two studies (n = 136) reported on this outcome. Both studies used indomethacin. There were no statistically significant differences between the very early treatment and expectant management groups with respect to time to full enteral feeds (MD 2.49 days; 95% CI ‐0.67 to 5.64; I² = 0%).

Severe retinopathy of prematurity (ROP) (stage 3 or greater) (according to the international classification of ROP)

(Analysis 2.18)

2.18. Analysis.

Comparison 2: Very early treatment vs expectant management, Outcome 18: Severe retinopathy of prematurity (ROP)

Four studies (n = 268) reported on this outcome. There were no statistically significant differences between the very early treatment and expectant management groups in severe ROP (typical RR 1.02, 95% CI 0.41 to 2.55; typical RD ‐0.02; 95% CI ‐0.07 to 0.03; I² = 36% for RR and 21% for RD, respectively).

Subgroup analyses

Two out of the four studies used indomethacin, one used ibuprofen and one study was a two‐center study that used indomethacin in one center and ibuprofen in the other. No significant differences in severe ROP were noted in the indomethacin only subgroup (typical RR 0.16, 95% CI 0.01 to 2.93; typical RD ‐0.03; 95% CI ‐0.09 to 0.02; test for heterogeneity not applicable for RR; I² = 12% for RD), ibuprofen only subgroup (typical RR 0.80, 95% CI 0.24 to 2.69; typical RD ‐0.03; 95% CI ‐0.21 to 0.15; test for heterogeneity not applicable), or the subgroup with use of either indomethacin or ibuprofen (typical RR 2.64, 95% CI 0.55, to 12.75; typical RD 0.09; 95% CI ‐0.05 to 0.23; test for heterogeneity not applicable).

Definite sepsis (clinical symptoms and signs of sepsis and a positive bacterial culture in a specimen obtained from normally sterile fluids or tissue obtained at postmortem)

(Analysis 2.19)

2.19. Analysis.

Comparison 2: Very early treatment vs expectant management, Outcome 19: Definite sepsis

Four studies (n = 260) reported on this outcome. There were no statistically significant differences between the very early treatment and expectant management groups in definite sepsis (typical RR 0.94, 95% CI 0.63 to 1.41; typical RD 0.01; 95% CI ‐0.07 to 0.09; I² = 0% for both RR and RD).

Subgroup analyses

Two out of the four studies used indomethacin and two used ibuprofen. No significant differences in definite sepsis were noted in the indomethacin only subgroup (typical RR 0.90, 95% CI 0.57 to 1.42; typical RD ‐0.04; 95% CI ‐0.20 to 0.12; I² = 0% for both RR and RD), or the ibuprofen only subgroup (typical RR 1.10, 95% CI 0.47, to 2.58; typical RD 0.03; 95% CI ‐0.07 to 0.12; I² = 0% for both RR and RD).

Oliguria (defined as less than 1 mL/kg/hour)

(Analysis 2.20)

2.20. Analysis.

Comparison 2: Very early treatment vs expectant management, Outcome 20: Oliguria

One study (n = 44) that used indomethacin reported on this outcome. There was no statistically significant difference between the very early treatment and expectant management groups in oliguria (typical RR 5.00, 95% CI 0.63 to 39.39; typical RD 0.18; 95% CI ‐0.01 to 0.38; test for heterogeneity not applicable).

Duration of hospitalization (total length of hospitalization from birth to discharge home or mortality) (days)

(Analysis 2.21)

2.21. Analysis.

Comparison 2: Very early treatment vs expectant management, Outcome 21: Duration of hospitalization

Four studies (n = 240) reported on this outcome. There was a statistically significant difference between the very early treatment and expectant management groups with respect to duration of hospitalization favoring the very early treatment group (MD ‐5.35 days; 95% CI ‐9.23 to ‐1.47; I² = 57%). We rated the certainty of evidence as low.

Subgroup analyses

One out of the four studies used indomethacin, two used ibuprofen and one study was a two‐center study that used indomethacin in one center and ibuprofen in the other. No significant differences in duration of hospitalization were noted in the indomethacin only subgroup (MD ‐1.00 days; 95% CI ‐12.83 to 10.83; test for heterogeneity not applicable) or the subgroup with use of either indomethacin or ibuprofen (MD 38.00 days; 95% CI ‐5.66 to 81.66; test for heterogeneity not applicable). There was a statistically significant difference between the very early treatment and expectant management groups in the ibuprofen only subgroup with respect to duration of hospitalization favoring the very early treatment group (MD ‐6.27 days; 95% CI ‐10.39 to ‐2.14; I² = 59%).

Neurodevelopmental outcomes

Moderate/severe cognitive delay at 18 to 24 months

One study (n = 79) that used indomethacin reported on this outcome. There was no statistically significant difference between the very early treatment and expectant management groups in moderate/severe cognitive delay (typical RR 0.27, 95% CI 0.03 to 2.31; typical RD ‐0.07; 95% CI ‐0.18 to 0.03; test for heterogeneity not applicable; Analysis 2.22). We rated the certainty of evidence as low.

2.22. Analysis.

Comparison 2: Very early treatment vs expectant management, Outcome 22: Neurodevelopmental impairment (moderate/severe cognitive delay) at 18‐24 months

Moderate/severe motor delay at 18 to 24 months

One study (n = 79) that used indomethacin reported on this outcome. There was no statistically significant difference between the very early treatment and expectant management groups in moderate/severe motor delay (typical RR 0.54, 95% CI 0.05 to 5.71; typical RD ‐0.02; 95% CI ‐0.11 to 0.06; test for heterogeneity not applicable; Analysis 2.23). We rated the certainty of evidence as low.

2.23. Analysis.

Comparison 2: Very early treatment vs expectant management, Outcome 23: Neurodevelopmental impairment (moderate/severe motor delay) at 18‐24 months

Moderate/severe language delay at 18 to 24 months

One study (n = 79) that used indomethacin reported on this outcome. There was no statistically significant difference between the very early treatment and expectant management groups in moderate/severe language delay (typical RR 0.54, 95% CI 0.10 to 2.78; typical RD ‐0.04; 95% CI ‐0.16 to 0.07; test for heterogeneity not applicable; Analysis 2.24 ). We rated the certainty of evidence as low.

2.24. Analysis.

Comparison 2: Very early treatment vs expectant management, Outcome 24: Neurodevelopmental impairment (moderate/severe language delay) at 18‐24 months

None of the studies included in this comparison reported on pneumothorax, pulmonary hypertension or time to regain birth weight. Neurodevelopmental outcomes for cognitive, motor and language functions were reported separately (Analysis 2.22; Analysis 2.23; Analysis 2.24).

Comparison 3. Very early treatment versus early treatment

None of the trials included in this review reported on very early treatment versus early treatment

Other subgroup and sensitivity analyses

Subgroup analyses based on choice of pharmacotherapy have been outlined within comparison 1 and 2. None of the trials were stratified based on the degree of hemodynamic significance of the PDA, hence this subgroup analysis was not possible. None of the trials specifically included infants less than 28 weeks or less than 1000 grams as predefined in our subgroup analyses. Three trials ( DeWaal 2020; EL‐Khuffash 2020; Kluckow 2014), included in the meta‐analysis enrolled infants < 29 weeks of gestation. Since, each trial used a different pharmacotherapeutic option (indomethacin only in Kluckow 2014, ibuprofen only in EL‐Khuffash 2020 and both indomethacin and ibuprofen in DeWaal 2020), we did not attempt to combine these three trials to create a new subgroup. Similarly given the small number of trials within each comparison and further within each subgroup, we refrained from conducting additional sensitivity analyses based on adequacy of trial methdology.

Discussion

Summary of main results

Fourteen RCTs completed to date have reported on 910 infants. Seven RCTs compared early treatment (defined as treatment initiated by seven days of age) versus expectant management and seven RCTs compared very early treatment (defined as treatment initiated by 72 hours of age) versus expectant management. Based on the choice of pharmacotherapy, the comparison of early treatment versus expectant management had two subgroups: intravenous/oral indomethacin (four studies; Gersony 1983; Knight 2011; Krauss 1989; Van Overmeire 2001), and intravenous/oral ibuprofen (three studies; Bagnoli 2013; Ghanem 2010; Sosenko 2012); the comparison of very early treatment versus expectant management had three subgroups: intravenous/oral indomethacin only (four studies; CTRI/2009/091/000041; Kaapa 1983; Kluckow 2014; Merritt 1981), and intravenous/oral ibuprofen only (two studies; EL‐Khuffash 2020; Lin 2012), and intravenous/oral indomethacin or intravenous/oral ibuprofen (one RCT; DeWaal 2020). The study by Knight 2011 did not contribute any data to the meta‐analysis.

There was no significant difference between early treatment versus expectant management (no treatment initiated within the first seven days after birth) for a hemodynamically significant PDA for the primary outcome of ‘all‐cause mortality’ (moderate‐certainty evidence) or other clinically relevant outcomes such as surgical PDA ligation (very low‐certainty evidence), chronic lung disease (moderate‐certainty evidence), or severe IVH and NEC (low‐certainty evidence). Infants receiving expectant management in the first seven days after birth appeared to be significantly less likely to receive any PDA pharmacotherapy compared to early initiation of treatment (low‐certainty evidence). Among other outcomes, early treatment of hs‐PDA appeared to significantly shorten time to full enteral feeds as well as time to regain birth weight while significantly increasing the risk of oliguria compared to expectant management. There were no notable subgroup differences observed for this comparison.

There was no significant difference between very early treatment versus expectant management (no treatment initiated within the first 72 hours after birth) for a hemodynamically significant PDA for the primary outcome of ‘all‐cause mortality’ (moderate‐certainty evidence) or other clinically relevant outcomes such as surgical PDA ligation, severe IVH and NEC (moderate‐certainty evidence), or chronic lung disease and neurodevelopmental impairment (low‐certainty evidence). Infants receiving expectant management in the first 72 hours after birth appeared to be significantly less likely to receive any PDA pharmacotherapy compared to very early initiation of treatment (very low‐certainty evidence). Very early treatment, however, appeared to significantly shorten the duration of hospitalization compared to expectant management (low‐certainty evidence). Among other outcomes, very early treatment of hs‐PDA appeared to significantly shorten the duration of mechanical ventilation while concomitantly increasing the exposure to postnatal corticosteroids compared to expectant management. The only clinically notable subgroup effect in this comparison was for the outcome of CLD. Very early treatment with ibuprofen appeared to significantly lower CLD while similar effects were not demonstrated in the other two subgroups.

Overall completeness and applicability of evidence

This is the first review comparing early treatment versus expectant management of a hemodynamically significant PDA in preterm infants. Though overall, expectant management appears to be a prudent option for the initial management of an hs‐PDA, being associated with less exposure to nonsteroidal anti‐inflammatory drugs without likely worsening relevant clinical outcomes, we recommend cautious interpretation and application of these findings due to several reasons.

First, it is important to consider the variation in expectant management practices across trials before applying the results in clinical practice. For the expectant management arm, in this review, we did not differentiate between no treatment (or placebo) versus active non‐pharmacological measures to reduce the PDA shunt volume. Six out of the 14 trials used only placebo or no treatment in the expectant management arm and did not specify any other active non‐pharmacological measures (CTRI/2009/091/000041; DeWaal 2020; EL‐Khuffash 2020; Kluckow 2014; Knight 2011; Sosenko 2012). Three out of the 14 trials actively restricted fluid intake as a part of their conservative management strategy (Bagnoli 2013; Ghanem 2010; Van Overmeire 2001), while the remaining five trials used medications such as loop diuretics and/or digoxin in addition to fluid restriction as a part of their conservative management strategy (Gersony 1983; Kaapa 1983; Krauss 1989; Lin 2012; Merritt 1981). Given the overall paucity of trials and the use of multiple pharmacotherapeutic interventions we decided against attempts at further teasing out the effects of different non‐pharmacological expectant management strategies on clinical outcomes. It has been shown by Mitra 2020 that PDA closure rates in the placebo/no treatment arms of the placebo‐controlled trials of hs‐PDA treatment vary substantially (ranging from 0% to 78%) highlighting the wide variation in expectant management strategies used across trials. Thus, we acknowledge that variation in expectant management strategies across trials could potentially be an important contributor to heterogeneity thus lowering our certainty in the estimates for relevant clinical outcomes.

Second, none of the trials were stratified based on the degree of hemodynamic significance of the PDA. There was wide variation in the definition of a hemodynamically significant PDA across trials. Hs‐PDA was defined in most trials based on characteristic clinical signs along with echocardiographic evidence of increased PDA shunt volume. However, the trials did not attempt to differentiate between PDAs with moderate versus high shunt volume based on any clinical or echocardiographic criteria. Hence, inclusion of all infants with PDA who fulfilled the current definition of hs‐PDA used in this review might have led to inclusion of a highly heterogeneous population in terms of their PDA shunt volume, especially more mature infants with a moderate volume PDA shunt where spontaneous PDA closure was highly likely. This could possibly explain why only a fraction of infants randomized to the expectant management groups required subsequent pharmacotherapy. Only four trials exclusively enrolled infants at less than 29 weeks of gestation (DeWaal 2020; EL‐Khuffash 2020; Kluckow 2014; Knight 2011). Therefore, clinicians should exercise caution in applying these results to extremely preterm infants born at less than 28 weeks of gestation with a large PDA shunt documented on echocardiography.

Early treatment appeared to shorten duration of parenteral nutrition and time to regain birth weight, while very early treatment appeared to reduce duration of hospitalization and mechanical ventilation. Such findings could be explained by early modulation of the PDA shunt volume using pharmacotherapy thereby improving systemic hypoperfusion as well as pulmonary hyperperfusion, respectively. However, early treatment was associated with increased risk of oliguria likely due to the effect of NSAIDs on glomerular filtration (Antonucci 2009). It is unclear from the review whether this NSAID‐induced oliguria translated into longer‐term kidney injury in preterm infants.

Infants receiving early treatment appeared to have a significantly higher exposure to postnatal corticosteroids. The trials included in this comparison were mostly at low risk of bias. Hence, higher use of postnatal corticosteroids in the very early treatment group is likely to be an unbiased co‐intervention secondary to physiological changes following very early treatment with NSAIDs. Schmidt 2006 had demonstrated that prophylactic indomethacin independently increased need for supplemental oxygen and reduced postnatal weight loss thereby leading to worsened ventilatory status by the end of the first week of age. This could possibly explain increased use of postnatal corticosteroids in the very early treatment group.

Subgroup analysis based on the choice of pharmacotherapy showed that very early treatment (within first 72 hours) with ibuprofen for an hs‐PDA significantly lowered CLD while such effects were not demonstrated with early treatment (initiated within the first seven days). Though these differences were observed in the respective subgroup analyses of two different comparisons, and therefore could have occurred by chance, one probable physiological explanation for this finding could be variation in treatment effectiveness in relation to ibuprofen pharmacokinetics. All ibuprofen trials included in this review used standard dose ibuprofen (10 mg/kg initial dose followed by two doses of 5mg/kg at 24 hour intervals). Hirt 2008 demonstrated that to achieve therapeutic serum concentrations of ibuprofen beyond the first 72 hours of age, increasing doses of ibuprofen are required irrespective of gestational age. Therefore infants receiving treatment with standard dose ibuprofen beyond the first 72 hours might not have received optimal pharmacotherapy to close an hs‐PDA. In addition, Schena 2015 and Liebowitz 2019 have shown that prolonged exposure to an hs‐PDA might increase the risk of CLD. This hypothesis was further tested in a post hoc analysis of the results of the PDA RCT (EL‐Khuffash 2020), included in this review. EL‐Khuffash 2020 showed that infants who achieved successful PDA closure had lower risk of CLD and CLD/death compared to infants receiving placebo. Therefore, whether ‘early treatment’ should be defined as ensuring PDA shunt elimination using an appropriate pharmacotherapeutic agent versus mere provision of pharmacotherapy requires further research.

Quality of the evidence

The certainty of evidence for the primary outcome ‘all‐cause mortality’ was moderate according to GRADE for both comparisons (early treatment versus expectant management; very early treatment versus expectant management). Evidence for other clinical outcomes were of very low‐ to moderate‐certainty. The evidence was rated down for the following reasons.

First, the overall risk of bias was judged as high for three trials (Ghanem 2010; Krauss 1989; Van Overmeire 2001) contributing to the early treatment versus expectant management comparison. The overall risk of bias was judged as low for two trials (Gersony 1983; Sosenko 2012), and unclear for two trials (Bagnoli 2013; Knight 2011). For the comparison of very early treatment versus expectant management, the overall risk of bias was judged as low for four (DeWaal 2020; EL‐Khuffash 2020; Kluckow 2014; Lin 2012), out of the seven trials and as high for the remaining three trials (CTRI/2009/091/000041; Kaapa 1983; Merritt 1981). Therefore, for some comparisons, more than 50% of the meta‐analytic weight was contributed by studies with unclear or high risk of bias. Hence, the certainty of evidence was rated down by one level for such comparisons.

Second, substantial heterogeneity was noted for outcomes such as surgical PDA ligation, receipt of any pharmacotherapy and NEC (for the comparison of early treatment versus expectant management) as well as for the outcomes such as receipt of any pharmacotherapy, CLD and duration of hospitalization (for the comparison of very early treatment versus expectant management). The heterogeneity could only be partly explained by choice of pharmacotherapy. Given the possibility of other sources of heterogeneity unaccounted for in the analysis, the certainty of evidence was rated down for these outcomes for inconsistency.

Third, most trials included in this review were small sample pilot trials with insufficient power to detect statistically significant differences in the rates of clinically relevant outcomes thereby leading to serious or very serious imprecision on GRADE assessment.

Potential biases in the review process

We are not aware of any potential biases in the review process. The authors were not involved with any of the included trials.

Agreements and disagreements with other studies or reviews

Some of the studies included in this review (Bagnoli 2013; Ghanem 2010; Lin 2012; Sosenko 2012), have been included in a previous Cochrane Neonatal Review on use of ibuprofen for treatment of PDA (Ohlsson 2020b). Many of the studies have also been included in a previous systematic review and network meta‐analysis comparing pharmacotherapeutic options for treatment of a symptomatic PDA (Mitra 2018). Previous meta‐analyses have shown that PDA pharmacotherapy is associated with significantly better PDA closure rates. However, improved rates of PDA closure did not appear to translate into longer‐term clinical benefit (Mitra 2018; Ohlsson 2020b).

Previous reviews, as mentioned above, have primarily focused on evaluating the effectiveness of therapy without taking into account the timing of treatment. It has been postulated that substantial use of open‐label pharmacotherapy in the placebo or no treatment group of included trials likely pulled the effect estimates for clinical outcomes towards the null thereby resulting in no statistically significant difference (Smith 2020). Acknowledging the fact that the pragmatic design of the trials allowing open‐label treatment might have prevented us from drawing firm conclusions on the effectiveness of treatment on clinical outcomes, an important unanswered clinical question was whether early expectant management for an hs‐PDA can reduce exposure to any NSAID without worsening clinical outcomes. Therefore, in this review we specifically explored this clinical question. We found that early expectant management for an hs‐PDA reduced exposure to any NSAID without worsening clinical outcomes. However, our confidence in the certainty of these estimates is moderate to low, due to the factors mentioned previously. Our results are in keeping with the results of two recent RCTs on early treatment versus conservative management in preterm infants with a persistent hemodynamically significant PDA beyond the first week of age (Clyman 2019; Sung 2020). Both trials found no difference in CLD or death with conservative management as compared to provision of pharmacotherapy in the second week of life. However, a post hoc secondary analysis of eligible infants treated outside the PDA‐TOLERATE trial (Clyman 2019) due to lack of physician equipoise showed that the infants who were treated prior to six days of age had a significantly lower incidence of CLD (odds ratio (OR) 0.28; 95% CI: 0.11 to 0.68) and combined outcome CLD or death (OR 0.26; 95% CI: 0.11 to 0.63) in spite of having a significantly lower gestational age (Liebowitz 2019). Hence, further large trials are required to conclusively establish that early non‐pharmacological management is the approach of choice irrespective of the gestational age and degree of hemodynamic significance of the PDA.

Authors' conclusions

Implications for practice.

Moderate‐certainty evidence suggests that early or very early pharmacotherapeutic treatment of an hs‐PDA probably does not reduce mortality in preterm infants. Low‐certainty evidence suggests that early pharmacotherapeutic treatment of hs‐PDA may lead to increased NSAID exposure in preterm infants. In addition, early pharmacotherapeutic treatment of hs‐PDA probably does not reduce CLD (moderate‐certainty evidence) and may not reduce severe IVH or NEC (low‐certainty evidence).

We are uncertain whether very early pharmacotherapeutic treatment of hs‐PDA leads to increased NSAID exposure in preterm infants (very low‐certainty evidence). Very early treatment probably does not reduce surgical PDA ligation, severe IVH or NEC (moderate‐certainty evidence), and may not reduce CLD or neurodevelopmental impairment (low‐certainty evidence). However, low‐certainty evidence does suggest that very early treatment of hs‐PDA may reduce the duration of hospitalization in preterm infants.

Implications for research.

Additional large trials adequately powered for patient‐important outcomes are required to explore if early targeted treatment of an hs‐PDA improves clinical outcomes such as mortality and neurodevelopmental impairment in preterm infants. Researchers should pay special attention to the inclusion criteria to maximize the chances of detecting a clinically meaningful signal in future trials. Extremely preterm infants born less than 28 weeks of gestational age with a large shunt volume are likely at the highest risk of PDA‐attributable morbidity (El‐Khuffash 2013; Mitra 2020). Therefore, future trials should attempt to restrict their inclusion criteria to this subgroup to avoid unnecessary exposure of more mature infants to NSAIDs where the likelihood of spontaneous PDA closure is high.Two large ongoing trials (ISRCTN84264977; NCT02884219) comparing very early treatment versus conservative management of hs‐PDA in extremely preterm infants will hopefully provide further evidence on very early treatment of hs‐PDA in this gestational age group.

The current review essentially compares early (or very early) treatment versus early conservative management. Given the high rate of open‐label treatment in the expectant management group, the current review was unable to explore if true conservative management without any exposure to NSAIDs during the hospital stay is as effective as early pharmacological treatment. If researchers attempt to answer the question whether PDA treatment truly alters clinically meaningful endpoints in preterm infants at the highest risk of PDA‐attributable morbidity, then any form of open‐label treatment in the control arm during hospital stay should be disallowed in future trials.

History

Protocol first published: Issue 2, 2019
Review first published: Issue 12, 2020

Appendices

Appendix 1. 2020 Search methods

The RCT filters have been created using Cochrane's highly sensitive search strategies for identifying randomised trials (Higgins 2017). The neonatal filters were created and tested by the Cochrane Neonatal Information Specialist.

CENTRAL via CRS Web:

Date ranges: 01 January 2019 to 02 October 2020
Terms:
1 MESH DESCRIPTOR Indomethacin EXPLODE ALL AND CENTRAL:TARGET
2 MESH DESCRIPTOR Ibuprofen EXPLODE ALL AND CENTRAL:TARGET
3 MESH DESCRIPTOR Acetaminophen EXPLODE ALL AND CENTRAL:TARGET
4 MESH DESCRIPTOR Cyclooxygenase Inhibitors EXPLODE ALL AND CENTRAL:TARGET
5 MESH DESCRIPTOR Anti‐Inflammatory Agents, Non‐Steroidal EXPLODE ALL AND CENTRAL:TARGET
6 indomethacin or indometacin or indocid or indocin or ibuprofen or advil or paracetamol or acetaminophen or tylenol or cyclooxygenase inhibitor* or cyclo oxygenase inhibitor* or non‐steroidal anti‐inflammatory drug* or non‐steroidal anti‐inflammatory agent* or NSAID* AND CENTRAL:TARGET
7 #1 OR #2 OR #3 OR #4 OR #5 OR #6 AND CENTRAL:TARGET
8 MESH DESCRIPTOR Ductus Arteriosus, Patent EXPLODE ALL AND CENTRAL:TARGET
9 patent ductus arteriosus or PDA or (patency adj3 ductus arteriosus) AND CENTRAL:TARGET
10 #8 OR #9 AND CENTRAL:TARGET
11 MESH DESCRIPTOR Infant, Newborn EXPLODE ALL AND CENTRAL:TARGET16104
12 infant or infants or infant's or "infant s" or infantile or infancy or newborn* or "new born" or "new borns" or "newly born" or neonat* or baby* or babies or premature or prematures or prematurity or preterm or preterms or "pre term" or premies or "low birth weight" or "low birthweight" or VLBW or LBW or ELBW or NICU AND CENTRAL:TARGET
13 #12 OR #11 AND CENTRAL:TARGET
14 #7 AND #10 AND #13 AND CENTRAL:TARGET
15 2019 TO 2020:YR AND CENTRAL:TARGET
16 #15 AND #14 AND CENTRAL:TARGET

MEDLINE via Ovid:

Date ranges: 01 January 2019 to 02 October 2020
Terms:
1. exp Indomethacin/
2. exp Ibuprofen/
3. exp Acetaminophen/
4. exp Cyclooxygenase Inhibitors/
5. exp Anti‐Inflammatory Agents, Non‐Steroidal/
6. (indomethacin or indometacin or indocid or indocin or ibuprofen or advil or paracetamol or acetaminophen or tylenol or cyclooxygenase inhibitor* or cyclo oxygenase inhibitor* or non‐steroidal anti‐inflammatory drug* or non‐steroidal anti‐inflammatory agent* or NSAID*).mp.
7. 1 or 2 or 3 or 4 or 5 or 6
8. exp Ductus Arteriosus, Patent/
9. (patent ductus arteriosus or PDA or (patency adj3 ductus arteriosus)).mp.
10. 8 or 9
11. exp infant, newborn/
12. (newborn* or new born or new borns or newly born or baby* or babies or premature or prematurity or preterm or pre term or low birth weight or low birthweight or VLBW or LBW or infant or infants or 'infant s' or infant's or infantile or infancy or neonat*).ti,ab.
13. 11 or 12
14. randomized controlled trial.pt.
15. controlled clinical trial.pt.
16. randomized.ab.
17. placebo.ab.
18. drug therapy.fs.
19. randomly.ab.
20. trial.ab.
21. groups.ab.
22. or/14‐21
23. exp animals/ not humans.sh.
24. 22 not 23
25. 13 and 24
26. randomi?ed.ti,ab.
27. randomly.ti,ab.
28. trial.ti,ab.
29. groups.ti,ab.
30. ((single or doubl* or tripl* or treb*) and (blind* or mask*)).ti,ab.
31. placebo*.ti,ab.
32. 26 or 27 or 28 or 29 or 30 or 31
33. 12 and 32
34. limit 33 to yr="2019 ‐Current"
35. 25 or 34
36. 7 and 10 and 35
37. limit 36 to yr="2019 ‐Current"

ISRCTN:

Date ranges: 2019 to 2020
Terms:
indomethacin within Condition: Patent ductus arteriosus AND Participant age range: Neonate
ibuprofen within Condition: Patent ductus arteriosus AND Participant age range: Neonate
paracetamol within Condition: Patent ductus arteriosus AND Participant age range: Neonate
acetaminophen within Condition: Patent ductus arteriosus AND Participant age range: Neonate
cyclooxygenase inhibitors AND Condition: Patent ductus arteriosus AND Participant age range: Neonate
non‐steroidal anti‐inflammatory drugs AND Condition: Patent ductus arteriosus AND Participant age range: Neonate

Appendix 2. Initial search methods

We used the criteria and standard methods of Cochrane and Cochrane Neonatal (see the Cochrane Neonatal search strategy for specialized register). We searched for errata or retractions from included studies published in full‐text on PubMed (www.ncbi.nlm.nih.gov/pubmed), and reported the date this was done within the review, if applicable.

We conducted a comprehensive search including: Cochrane Central Register of Controlled Trials (CENTRAL 2019, issue 6) in the Cochrane Library; MEDLINE via PubMed (1996 to 31 May 2019); Embase via Ovid (1980 to 31 May 2019); and CINAHL via EBSCOhost (1982 to 31 May 2019) using the following search terms: patent ductus arteriosus, PDA, indomethacin, ibuprofen, paracetamol, acetaminophen, Tylenol, and applicable MeSH terms plus database‐specific limiters for RCTs and neonates (see below for the full search strategies for each database). We did not apply language restrictions. We searched clinical trials registries for ongoing or recently completed trials (clinicaltrials.gov; the World Health Organization's International Trials Registry and Platform, and the ISRCTN Registry).

CENTRAL

infant or infants or infantile or infancy or newborn* or "new born" or "new borns" or "newly born" or neonat* or baby* or babies or premature or prematures or prematurity or preterm or preterms or "pre term" or premies or "low birth weight" or "low birthweight" or VLBW or LBW or ELBW or NICU

PubMed

((infant, newborn[MeSH] OR newborn*[TIAB] OR "new born"[TIAB] OR "new borns"[TIAB] OR "newly born"[TIAB] OR baby*[TIAB] OR babies*[TIAB] OR premature[TIAB] OR prematurity[TIAB] OR preterm[TIAB] OR "pre term"[TIAB] OR “low birth weight”[TIAB] OR "low birthweight"[TIAB] OR VLBW[TIAB] OR LBW[TIAB] OR infan*[TIAB] OR neonat*[TIAB]) AND (randomized controlled trial [pt] OR controlled clinical trial [pt] OR randomized [tiab] OR placebo [tiab] OR drug therapy [sh] OR randomly [tiab] OR trial [tiab] OR groups [tiab]) NOT (animals [mh] NOT humans [mh]))

Embase

#1 (infant, newborn or newborn or neonate or neonatal or premature or very low birth weight or low birth weight or VLBW or LBW or Newborn or infan* or neonat*).mp

#2 exp infant

#3 (#1 OR #2)

#4 (human not animal).mp

#5 (randomized controlled trial or controlled clinical trial or randomized or placebo or clinical trials as topic or randomly or trial or clinical trial).mp

#6 (#3 and #4 and #5)

CINAHL

(infant, newborn OR newborn OR neonate OR neonatal OR premature OR low birth weight OR VLBW OR LBW or Newborn or infan* or neonat*) AND (randomized controlled trial OR controlled clinical trial OR randomized OR placebo OR clinical trials as topic OR randomly OR trial OR PT clinical trial)

Appendix 3. 'Risk of bias' tool

1. Sequence generation (checking for possible selection bias). Was the allocation sequence adequately generated?

For each included study, we categorised the method used to generate the allocation sequence as:

1. low risk (any truly random process e.g. random number table; computer random number generator);

2. high risk (any non‐random process e.g. odd or even date of birth; hospital or clinic record number); or

3. unclear risk.

2. Allocation concealment (checking for possible selection bias). Was allocation adequately concealed?

For each included study, we categorised the method used to conceal the allocation sequence as:

1. low risk (e.g. telephone or central randomisation; consecutively numbered sealed opaque envelopes);

2. high risk (open random allocation; unsealed or non‐opaque envelopes, alternation; date of birth); or

3. unclear risk.

3. Blinding of participants and personnel (checking for possible performance bias). Was knowledge of the allocated intervention adequately prevented during the study?

For each included study, we categorised the methods used to blind study participants and personnel from knowledge of which intervention a participant received. Blinding was assessed separately for different outcomes or class of outcomes. We categorised the methods as:

1. low risk, high risk or unclear risk for participants; and

2. low risk, high risk or unclear risk for personnel.

4. Blinding of outcome assessment (checking for possible detection bias). Was knowledge of the allocated intervention adequately prevented at the time of outcome assessment?

For each included study, we categorised the methods used to blind outcome assessment. Blinding was assessed separately for different outcomes or class of outcomes. We categorised the methods as:

1. low risk for outcome assessors;

2. high risk for outcome assessors; or

3. unclear risk for outcome assessors.

5. Incomplete outcome data (checking for possible attrition bias through withdrawals, dropouts, protocol deviations). Were incomplete outcome data adequately addressed?

For each included study and for each outcome, we described the completeness of data including attrition and exclusions from the analysis. We noted whether attrition and exclusions were reported, the numbers included in the analysis at each stage (compared with the total randomised participants), reasons for attrition or exclusion where reported, and whether missing data were balanced across groups or were related to outcomes. Where sufficient information was reported or supplied by the trial authors, we re‐included missing data in the analyses. We categorised the methods as:

1. low risk (< 20% missing data);

2. high risk (≥ 20% missing data); or

3. unclear risk.

6. Selective reporting bias. Are reports of the study free of suggestion of selective outcome reporting?

For each included study, we described how we investigated the possibility of selective outcome reporting bias and what we found. For studies in which study protocols were published in advance, we compared prespecified outcomes versus outcomes eventually reported in the published results. If the study protocol was not published in advance, we contacted study authors to gain access to the study protocol. We assessed the methods as:

1. low risk (where it is clear that all of the study's prespecified outcomes and all expected outcomes of interest to the review have been reported);

2. high risk (where not all the study's prespecified outcomes have been reported; one or more reported primary outcomes were not prespecified outcomes of interest and are reported incompletely and so cannot be used; study fails to include results of a key outcome that would have been expected to have been reported); or

3. unclear risk.

7. Other sources of bias. Was the study apparently free of other problems that could put it at a high risk of bias?

For each included study, we described any important concerns we had about other possible sources of bias (for example, whether there was a potential source of bias related to the specific study design or whether the trial was stopped early due to some data‐dependent process). We assessed whether each study was free of other problems that could put it at risk of bias as:

1. low risk;

2. high risk; or

3. unclear risk.

If needed, we explored the impact of the level of bias through undertaking sensitivity analyses.

Appendix 4. Subgroups based on pharmacotherapeutic agents used for patent ductus arteriosus treatment

1. Intravenous (IV) or oral ibuprofen (10 mg/kg to 20 mg/kg followed by 5 mg/kg to 10 mg/kg 24 and 48 hours later).

2. IV ibuprofen (10 mg/kg to 20 mg/kg IV followed by 5 mg/kg to 10 mg/kg IV 24 and 48 hours later).

3. Standard‐dose IV ibuprofen (10 mg/kg IV followed by 5 mg/kg IV 24 and 48 hours later).

4. High‐dose IV ibuprofen (15 mg/kg to 20 mg/kg IV followed by 7.5 mg/kg to 10 mg/kg IV 24 and 48 hours later).

5. Oral ibuprofen (10 mg/kg to 20 mg/kg orally followed by 5 mg/kg to 10 mg/kg orally 24 and 48 hours later).

6. Standard‐dose oral ibuprofen (10 mg/kg orally followed by 5 mg/kg orally 24 and 48 hours later).

7. High‐dose oral ibuprofen (15 mg/kg to 20 mg/kg orally followed by 7.5 mg/kg to 10 mg/kg orally 24 and 48 hours later).

8. IV or oral indomethacin (0.1 mg/kg to 0.3 mg/kg at 12 to 24‐hour intervals for three doses).

9. IV indomethacin (0.1 mg/kg to 0.3 mg/kg IV at 12 to 24‐hour intervals for three doses).

10. Oral indomethacin (0.1 mg/kg to 0.3 mg/kg orally at 12 to 24‐hour intervals for three doses).

11. IV or oral or rectal acetaminophen (15 mg/kg at six‐hour intervals for three to seven days).

12. Oral acetaminophen (15 mg/kg at six‐hour intervals for three to seven days).

13. IV acetaminophen (15 mg/kg at six‐hour intervals for three to seven days).

14. Rectal acetaminophen (15 mg/kg at six‐hour intervals for three to seven days).

Data and analyses

Comparison 1. Early treatment vs expectant management.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 All‐cause mortality during hospital stay	6	500	Risk Ratio (M‐H, Fixed, 95% CI)	0.80 [0.46, 1.39]
1.1.1 Intravenous/oral indomethacin	3	195	Risk Ratio (M‐H, Fixed, 95% CI)	0.95 [0.45, 1.99]
1.1.2 Intravenous/oral ibuprofen	3	305	Risk Ratio (M‐H, Fixed, 95% CI)	0.65 [0.28, 1.50]
1.2 Surgical PDA ligation or transcatheter occlusion	4	432	Risk Ratio (IV, Fixed, 95% CI)	1.08 [0.65, 1.80]
1.2.1 Intravenous/oral indomethacin	1	127	Risk Ratio (IV, Fixed, 95% CI)	0.74 [0.17, 3.17]
1.2.2 Intravenous/oral ibuprofen	3	305	Risk Ratio (IV, Fixed, 95% CI)	1.14 [0.66, 1.96]
1.3 Receipt of any pharmacotherapy for a hemodynamically significant PDA	2	232	Risk Ratio (IV, Fixed, 95% CI)	2.30 [1.86, 2.83]
1.3.1 Intravenous/oral indomethacin	1	127	Risk Ratio (IV, Fixed, 95% CI)	2.40 [1.79, 3.21]
1.3.2 Intravenous/oral ibuprofen	1	105	Risk Ratio (IV, Fixed, 95% CI)	2.19 [1.62, 2.96]
1.4 Receipt of rescue medical treatment for a hemodynamically significant PDA	2	232	Risk Ratio (IV, Fixed, 95% CI)	0.63 [0.27, 1.48]
1.4.1 Intravenous/oral indomethacin	1	127	Risk Ratio (IV, Fixed, 95% CI)	0.33 [0.01, 7.91]
1.4.2 Intravenous/oral ibuprofen	1	105	Risk Ratio (IV, Fixed, 95% CI)	0.66 [0.27, 1.60]
1.5 Chronic lung disease	4	339	Risk Ratio (IV, Fixed, 95% CI)	0.90 [0.62, 1.29]
1.5.1 Intravenous/oral indomethacin	2	168	Risk Ratio (IV, Fixed, 95% CI)	0.84 [0.52, 1.37]
1.5.2 Intravenous/oral ibuprofen	2	171	Risk Ratio (IV, Fixed, 95% CI)	0.97 [0.56, 1.69]
1.6 Duration of ventilator support (days)	1	127	Mean Difference (IV, Fixed, 95% CI)	‐13.40 [‐30.15, 3.35]
1.7 Duration of need for supplementary oxygen (days)	1	127	Mean Difference (IV, Fixed, 95% CI)	‐8.30 [‐43.27, 26.67]
1.8 Postnatal corticosteroid use for CLD any time during hospital stay	1	105	Risk Ratio (IV, Fixed, 95% CI)	1.13 [0.37, 3.49]
1.9 Pneumothorax	2	146	Risk Ratio (IV, Fixed, 95% CI)	0.88 [0.32, 2.42]
1.9.1 Intravenous/oral indomethacin	1	41	Risk Ratio (IV, Fixed, 95% CI)	0.62 [0.15, 2.56]
1.9.2 Intravenous/oral ibuprofen	1	105	Risk Ratio (IV, Fixed, 95% CI)	1.26 [0.30, 5.35]
1.10 Intraventricular hemorrhage (IVH) (grades I to IV)	3	234	Risk Ratio (IV, Fixed, 95% CI)	0.91 [0.34, 2.48]
1.10.1 Intravenous/oral indomethacin	2	168	Risk Ratio (IV, Fixed, 95% CI)	1.19 [0.21, 6.60]
1.10.2 Intravenous/oral ibuprofen	1	66	Risk Ratio (IV, Fixed, 95% CI)	0.80 [0.24, 2.72]
1.11 Severe intraventricular hemorrhage IVH (grades III and IV)	2	171	Risk Ratio (IV, Fixed, 95% CI)	0.83 [0.32, 2.16]
1.12 Periventricular leukomalacia (PVL; any grade)	3	298	Risk Ratio (IV, Fixed, 95% CI)	1.22 [0.55, 2.75]
1.12.1 Intravenous/oral indomethacin	1	127	Risk Ratio (IV, Fixed, 95% CI)	2.46 [0.50, 12.22]
1.12.2 Intravenous/oral ibuprofen	2	171	Risk Ratio (IV, Fixed, 95% CI)	0.97 [0.38, 2.46]
1.13 Necrotizing enterocolitis (NEC; stage 2 or greater)	5	473	Risk Ratio (IV, Fixed, 95% CI)	2.34 [0.86, 6.41]
1.13.1 Intravenous/oral indomethacin	2	168	Risk Ratio (IV, Fixed, 95% CI)	1.56 [0.28, 8.80]
1.13.2 Intravenous/oral ibuprofen	3	305	Risk Ratio (IV, Fixed, 95% CI)	2.89 [0.84, 9.95]
1.14 Spontaneous intestinal perforation	3	298	Risk Ratio (IV, Fixed, 95% CI)	0.57 [0.14, 2.37]
1.14.1 Intravenous/oral indomethacin	1	127	Risk Ratio (IV, Fixed, 95% CI)	0.98 [0.06, 15.40]
1.14.2 Intravenous/oral ibuprofen	2	171	Risk Ratio (IV, Fixed, 95% CI)	0.47 [0.09, 2.47]
1.15 Time to full enteral feeds (postnatal age at time of achieving full enteral feeds)	1	66	Mean Difference (IV, Fixed, 95% CI)	‐4.80 [‐8.55, ‐1.05]
1.16 Time to regain birth weight (days)	1	66	Mean Difference (IV, Fixed, 95% CI)	‐3.40 [‐6.13, ‐0.67]
1.17 Severe retinopathy of prematurity (ROP)	2	146	Risk Ratio (IV, Fixed, 95% CI)	1.30 [0.44, 3.84]
1.17.1 Intravenous/oral indomethacin	1	41	Risk Ratio (IV, Fixed, 95% CI)	0.30 [0.02, 5.34]
1.17.2 Intravenous/oral ibuprofen	1	105	Risk Ratio (IV, Fixed, 95% CI)	1.65 [0.51, 5.31]
1.18 Definite sepsis	4	339	Risk Ratio (IV, Fixed, 95% CI)	0.97 [0.68, 1.38]
1.18.1 Intravenous/oral indomethacin	2	168	Risk Ratio (IV, Fixed, 95% CI)	1.16 [0.56, 2.41]
1.18.2 Intravenous/oral ibuprofen	2	171	Risk Ratio (IV, Fixed, 95% CI)	0.92 [0.61, 1.38]
1.19 Oliguria	2	261	Risk Ratio (IV, Fixed, 95% CI)	6.41 [2.13, 19.26]
1.19.1 Intravenous/oral indomethacin	1	127	Risk Ratio (IV, Fixed, 95% CI)	4.59 [1.39, 15.21]
1.19.2 Intravenous/oral ibuprofen	1	134	Risk Ratio (IV, Fixed, 95% CI)	39.00 [2.40, 633.01]

Comparison 2. Very early treatment vs expectant management.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 All‐cause mortality during hospital stay	7	384	Risk Ratio (IV, Fixed, 95% CI)	0.94 [0.58, 1.53]
2.1.1 Intravenous/oral indomethacin	4	188	Risk Ratio (IV, Fixed, 95% CI)	0.92 [0.47, 1.80]
2.1.2 Intravenous/oral ibuprofen	2	124	Risk Ratio (IV, Fixed, 95% CI)	1.46 [0.58, 3.67]
2.1.3 Intravenous/oral indomethacin or intravenous/oral ibuprofen	1	72	Risk Ratio (IV, Fixed, 95% CI)	0.53 [0.17, 1.60]
2.2 Surgical PDA ligation or transcatheter occlusion	5	293	Risk Ratio (IV, Fixed, 95% CI)	0.88 [0.36, 2.17]
2.2.1 Intravenous/oral indomethacin	3	161	Risk Ratio (IV, Fixed, 95% CI)	0.54 [0.07, 3.93]
2.2.2 Intravenous/oral ibuprofen	1	60	Risk Ratio (IV, Fixed, 95% CI)	1.00 [0.36, 2.75]
2.2.3 Intravenous/oral indomethacin or intravenous/oral ibuprofen	1	72	Risk Ratio (IV, Fixed, 95% CI)	Not estimable
2.3 Receipt of any pharmacotherapy for a hemodynamically significant PDA	4	156	Risk Ratio (IV, Fixed, 95% CI)	1.64 [1.31, 2.05]
2.3.1 Intravenous/oral indomethacin	3	96	Risk Ratio (IV, Fixed, 95% CI)	1.42 [1.10, 1.82]
2.3.2 Intravenous/oral ibuprofen	1	60	Risk Ratio (IV, Fixed, 95% CI)	2.90 [1.77, 4.76]
2.4 Receipt of repeat courses of pharmacotherapy for a persistent hemodynamically significant PDA	3	133	Risk Ratio (IV, Fixed, 95% CI)	1.99 [0.50, 7.94]
2.4.1 Intravenous/oral indomethacin	2	69	Risk Ratio (IV, Fixed, 95% CI)	1.39 [0.29, 6.66]
2.4.2 Intravenous/oral ibuprofen	1	64	Risk Ratio (IV, Fixed, 95% CI)	7.00 [0.38, 130.26]
2.5 Receipt of rescue medical treatment for a hemodynamically significant PDA	2	164	Risk Ratio (IV, Fixed, 95% CI)	0.54 [0.28, 1.04]
2.5.1 Intravenous/oral indomethacin	1	92	Risk Ratio (IV, Fixed, 95% CI)	0.52 [0.26, 1.02]
2.5.2 Intravenous/oral indomethacin or intravenous/oral ibuprofen	1	72	Risk Ratio (IV, Fixed, 95% CI)	1.06 [0.07, 16.26]
2.6 Chronic lung disease	7	384	Risk Ratio (IV, Fixed, 95% CI)	0.83 [0.63, 1.08]
2.6.1 Intravenous/oral indomethacin	4	188	Risk Ratio (IV, Fixed, 95% CI)	1.06 [0.61, 1.83]
2.6.2 Intravenous/oral ibuprofen	2	124	Risk Ratio (IV, Fixed, 95% CI)	0.54 [0.35, 0.83]
2.6.3 Intravenous/oral indomethacin or intravenous/oral ibuprofen	1	72	Risk Ratio (IV, Fixed, 95% CI)	1.12 [0.71, 1.75]
2.7 Duration of ventilator support (days)	6	253	Mean Difference (IV, Fixed, 95% CI)	‐0.45 [‐0.75, ‐0.16]
2.7.1 Intravenous/oral indomethacin	3	92	Mean Difference (IV, Fixed, 95% CI)	‐1.37 [‐2.70, ‐0.04]
2.7.2 Intravenous/oral ibuprofen	2	89	Mean Difference (IV, Fixed, 95% CI)	‐0.41 [‐0.71, ‐0.11]
2.7.3 Intravenous/oral indomethacin or intravenous/oral ibuprofen	1	72	Mean Difference (IV, Fixed, 95% CI)	6.00 [‐6.69, 18.69]
2.8 Duration of need for supplementary oxygen (days)	3	127	Mean Difference (IV, Fixed, 95% CI)	‐1.73 [‐3.99, 0.53]
2.8.1 Intravenous/oral indomethacin	2	67	Mean Difference (IV, Fixed, 95% CI)	‐1.61 [‐3.95, 0.74]
2.8.2 Intravenous/oral ibuprofen	1	60	Mean Difference (IV, Fixed, 95% CI)	‐3.30 [‐11.83, 5.23]
2.9 Postnatal corticosteroid use for CLD any time during hospital stay	3	224	Risk Ratio (IV, Fixed, 95% CI)	2.11 [1.12, 3.97]
2.9.1 Intravenous/oral indomethacin	1	92	Risk Ratio (IV, Fixed, 95% CI)	1.91 [0.60, 6.08]
2.9.2 Intravenous/oral ibuprofen	1	60	Risk Ratio (IV, Fixed, 95% CI)	2.50 [0.53, 11.89]
2.9.3 Intravenous/oral indomethacin or intravenous/oral ibuprofen	1	72	Risk Ratio (IV, Fixed, 95% CI)	2.11 [0.89, 5.02]
2.10 Pulmonary hemorrhage	5	332	Risk Ratio (IV, Fixed, 95% CI)	0.58 [0.30, 1.11]
2.10.1 Intravenous/oral indomethacin	2	136	Risk Ratio (IV, Fixed, 95% CI)	0.59 [0.22, 1.53]
2.10.2 Intravenous/oral ibuprofen	2	124	Risk Ratio (IV, Fixed, 95% CI)	0.59 [0.24, 1.49]
2.10.3 Intravenous/oral indomethacin or intravenous/oral ibuprofen	1	72	Risk Ratio (IV, Fixed, 95% CI)	0.35 [0.01, 8.36]
2.11 Intraventricular hemorrhage (IVH) (grades I to IV)	3	200	Risk Ratio (IV, Fixed, 95% CI)	0.70 [0.44, 1.10]
2.11.1 Intravenous/oral indomethacin	2	136	Risk Ratio (IV, Fixed, 95% CI)	0.75 [0.27, 2.10]
2.11.2 Intravenous/oral ibuprofen	1	64	Risk Ratio (IV, Fixed, 95% CI)	0.68 [0.41, 1.14]
2.12 Severe intraventricular hemorrhage IVH (grades III and IV)	4	240	Risk Ratio (IV, Fixed, 95% CI)	0.64 [0.21, 1.93]
2.12.1 Intravenous/oral indomethacin	1	44	Risk Ratio (IV, Fixed, 95% CI)	1.00 [0.07, 15.00]
2.12.2 Intravenous/oral ibuprofen	2	124	Risk Ratio (IV, Fixed, 95% CI)	0.67 [0.11, 3.98]
2.12.3 Intravenous/oral indomethacin or intravenous/oral ibuprofen	1	72	Risk Ratio (IV, Fixed, 95% CI)	0.53 [0.10, 2.71]
2.13 Periventricular leukomalacia (PVL; any grade)	5	332	Risk Ratio (IV, Fixed, 95% CI)	0.61 [0.31, 1.20]
2.14 Necrotizing enterocolitis (NEC; stage 2 or greater)	5	332	Risk Ratio (IV, Fixed, 95% CI)	1.08 [0.53, 2.21]
2.14.1 Intravenous/oral indomethacin	2	136	Risk Ratio (IV, Fixed, 95% CI)	0.80 [0.18, 3.49]
2.14.2 Intravenous/oral ibuprofen	2	124	Risk Ratio (IV, Fixed, 95% CI)	1.01 [0.42, 2.44]
2.14.3 Intravenous/oral indomethacin or intravenous/oral ibuprofen	1	72	Risk Ratio (IV, Fixed, 95% CI)	3.17 [0.35, 29.07]
2.15 Gastrointestinal bleeding within seven days of the first dose of pharmacotherapy	2	164	Risk Ratio (IV, Fixed, 95% CI)	Not estimable
2.15.1 Intravenous/oral indomethacin	1	92	Risk Ratio (IV, Fixed, 95% CI)	Not estimable
2.15.2 Intravenous/oral indomethacin or intravenous/oral ibuprofen	1	72	Risk Ratio (IV, Fixed, 95% CI)	Not estimable
2.16 Spontaneous intestinal perforation	4	272	Risk Ratio (IV, Fixed, 95% CI)	0.96 [0.15, 6.36]
2.16.1 Intravenous/oral indomethacin	2	136	Risk Ratio (IV, Fixed, 95% CI)	Not estimable
2.16.2 Intravenous/oral ibuprofen	1	64	Risk Ratio (IV, Fixed, 95% CI)	0.50 [0.05, 5.24]
2.16.3 Intravenous/oral indomethacin or intravenous/oral ibuprofen	1	72	Risk Ratio (IV, Fixed, 95% CI)	3.17 [0.13, 75.24]
2.17 Time to full enteral feeds (postnatal age at time of achieving full enteral feeds)	2	136	Mean Difference (IV, Fixed, 95% CI)	2.49 [‐0.67, 5.64]
2.17.1 Intravenous/oral indomethacin	2	136	Mean Difference (IV, Fixed, 95% CI)	2.49 [‐0.67, 5.64]
2.18 Severe retinopathy of prematurity (ROP)	4	268	Risk Ratio (IV, Fixed, 95% CI)	1.02 [0.41, 2.55]
2.18.1 Intravenous/oral indomethacin	2	136	Risk Ratio (IV, Fixed, 95% CI)	0.16 [0.01, 2.93]
2.18.2 Intravenous/oral ibuprofen	1	60	Risk Ratio (IV, Fixed, 95% CI)	0.80 [0.24, 2.69]
2.18.3 Intravenous/oral indomethacin or intravenous/oral ibuprofen	1	72	Risk Ratio (IV, Fixed, 95% CI)	2.64 [0.55, 12.75]
2.19 Definite sepsis	4	260	Risk Ratio (IV, Fixed, 95% CI)	0.94 [0.63, 1.41]
2.19.1 Intravenous/oral indomethacin	2	136	Risk Ratio (IV, Fixed, 95% CI)	0.90 [0.57, 1.42]
2.19.2 Intravenous/oral ibuprofen	2	124	Risk Ratio (IV, Fixed, 95% CI)	1.10 [0.47, 2.58]
2.20 Oliguria	1	44	Risk Ratio (IV, Fixed, 95% CI)	5.00 [0.63, 39.39]
2.20.1 Intravenous/oral indomethacin	1	44	Risk Ratio (IV, Fixed, 95% CI)	5.00 [0.63, 39.39]
2.21 Duration of hospitalization	4	240	Mean Difference (IV, Fixed, 95% CI)	‐5.35 [‐9.23, ‐1.47]
2.21.1 Intravenous/oral indomethacin	1	44	Mean Difference (IV, Fixed, 95% CI)	‐1.00 [‐12.83, 10.83]
2.21.2 Intravenous/oral ibuprofen	2	124	Mean Difference (IV, Fixed, 95% CI)	‐6.27 [‐10.39, ‐2.14]
2.21.3 Intravenous/oral indomethacin or intravenous/oral ibuprofen	1	72	Mean Difference (IV, Fixed, 95% CI)	38.00 [‐5.66, 81.66]
2.22 Neurodevelopmental impairment (moderate/severe cognitive delay) at 18‐24 months	1	79	Risk Ratio (IV, Fixed, 95% CI)	0.27 [0.03, 2.31]
2.23 Neurodevelopmental impairment (moderate/severe motor delay) at 18‐24 months	1	79	Risk Ratio (IV, Fixed, 95% CI)	0.54 [0.05, 5.71]
2.24 Neurodevelopmental impairment (moderate/severe language delay) at 18‐24 months	1	79	Risk Ratio (IV, Fixed, 95% CI)	0.54 [0.10, 2.78]

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Bagnoli 2013.

Study characteristics
Methods	Single‐center, randomized controlled trial conducted in Siena, Italy Study period: January 2006 to December 2010
Participants	Inclusion criteria: gestational age ≤ 32 weeks; birth weight ≤ 1500 g; PDA with evidence of ductal shunting documented by echocardiography, postnatal age > 72 hours; and informed consent signed by a parent/legal guardian. The criterion for hemodynamically significant PDA was the presence of at least 2 of the following three parameters: LA/AO ratio of ≥ 1.4:1; LV/AO ratio of 2.1:1; and/or narrowest ductal diameter > 1.5 mm, left‐to‐right shunting of blood and diastolic reversal of blood flow in the aorta. Exclusion criteria: congenital heart malformation; major congenital malformations or chromosomal anomalies, or both; antenatal renal malformation on fetal ultrasound; platelet count < 75,000/mm3; clinical bleeding tendency; renal failure; congenital bacterial infection; treatment with a steroid at any time since birth.
Interventions	Intervention characteristics Early treatment (n = 67) Medication: IV ibuprofen Dose and frequency: ibuprofen was given intravenously for 10 minutes using 10 mg/kg loading doses, followed by 5 mg/kg/d on the 2nd and 3rd study day Timing of treatment: echocardiograms were performed in the first 72 to 96 hours of life and infants diagnosed with hs‐PDA based on these echocardiograms were randomized to IV ibuprofen or placebo. Therefore, all infants were treated within the first 4 days of life. Expectant management (n = 67) Placebo (0.9% NaCl) given IV Additional comments on expectant management: From day 1 to day 7, the amount of fluid administered to the population was 70, 90, 110, 130, 140, 140 and 150 mL/kg respectively.
Outcomes	Relevant outcomes for this study included: need for surgical ligation of the PDA; oliguria; NEC; mortality at 28 days of life.
Identification	Sponsorship source: None declared Country: Italy Setting: Neonatal Intensive Care Unit at Siena University Hospital (Italy) between January 2006 and December 2010 Authors name: Franco Bagnoli Institution: Department of Pediatrics, Obstetrics and Reproductive Medicine, Neonatal Intensive Care, University of Siena, Siena, Italy Email: annalisa.rossetti1985@gmail.com Address: Annalisa Rossetti, Department of Pediatrics, Obstetrics and Reproductive Medicine, Neonatal Intensive Care, University of Siena, Siena, Italy.
Notes	Additional study details were obtained from the previously published Cochrane review "Ibuprofen for the treatment of patent ductus arteriosus in preterm or low birth weight (or both) infants" (Ohlsson 2020b).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Sequence generation method not specified
Allocation concealment (selection bias)	Unclear risk	Allocation concealment method not specified
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	The methods section mentioned that this was a double‐blind study, but method of blinding not specified
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Method of blinding not specified
Incomplete outcome data (attrition bias) All outcomes	Low risk	Outcome data reported for all randomized infants
Selective reporting (reporting bias)	Unclear risk	Study protocol was unavailable. Unclear if there were any deviations from the protocol
Other bias	Low risk	Appeared free of other bias

CTRI/2009/091/000041.

Study characteristics
Methods	Single‐center, randomized controlled trial conducted in New Delhi, India Study period: 2007 to 2009
Participants	Inclusion criteria: infants born between 28 and 32 weeks; on assisted ventilation in NICU within 24 hours of birth with a hemodynamically significant PDA Hemodynamically significant PDA was defined as: PDA diameter 2 mm or larger and LA:AO > 1.4 in first 24 hours after birth on echocardiogram. Exclusion criteria: additional cardiac defects: ASD, VSD
Interventions	Intervention characteristics Very early treatment (n = 22) Medication: IV indomethacin Dose and frequency: 1st dose was 0.2 mg/kg within the first 24 hours after birth followed by 2 doses of 0.1 mg/kg/dose 24 hours apart on 2nd and 3rd day. If any additional course was warranted, schedule for repeat course was as follows: 5 doses, each 0.1 mg/kg 24 hours apart for 5 consecutive days. Timing of treatment: treatment initiated within the first 24 hours after birth Expectant management (n = 22) No treatment
Outcomes	Relevant outcomes for this study included: mortality; surgical PDA ligation; receipt of any pharmacotherapy for hs‐PDA; receipt of repeat courses of pharmacotherapy for a persistent hs‐PDA; CLD; pulmonary hemorrhage; IVH, severe IVH; PVL; NEC; spontaneous intestinal perforation; severe ROP; sepsis; oliguria; duration of ventilatory support; duration of supplementary oxygen; time to full enteral feeds; duration of hospitalization.
Identification	Sponsorship source: not mentioned Country: India Setting: level III NICU, Sir Ganga Ram Hospital, New Delhi Authors name: Anil Batra Institution: Department of Neonatology Sir Ganga Ram Hospital Email: drbatra77@gmail.com Address: New Delhi, India
Notes	The study abstract was published as a conference proceeding. All relevant data were obtained from personal communication with the contact author.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Stratified block randomization (in blocks of 6 each) was done by computer‐generated random number sequence.
Allocation concealment (selection bias)	Low risk	Numbers were stored in sealed opaque envelopes.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Not blinded
Blinding of outcome assessment (detection bias) All outcomes	High risk	Not blinded
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Results for all randomized infants were reported.
Selective reporting (reporting bias)	Unclear risk	The trial was registered with the Clinical Trials Registry of India (CTRI) (CTRI/2009/091/000041). There does not seem to be any obvious protocol deviation but recruitment was apparently completed when the trial was registered. So, unclear if there was any selective outcome reporting
Other bias	Unclear risk	Given this was an unblinded trial, difficult to assess if there were other sources of bias. Furthermore, the trial was retrospectively registered following complete recruitment.

DeWaal 2020.

Study characteristics
Methods	Multicenter (2 sites), randomized controlled trial Study period: 2016 to 2019
Participants	Inclusion criteria: preterm infant < 29 week gestational age at birth; presence of hs‐PDA on echocardiogram. HS‐PDA defined as: PDA diameter > 1.5 mm with signs of pulmonary volume load (LA/AO ratio > 1.5 and/or left atrium volume > 1.00 mL/kg) Exclusion criteria: significant (> 33%) right‐to‐left shunting over the PDA; active major hemorrhage (pulmonary, IVH grade 3 or 4, other); critical congenital heart disease or other major congenital abnormalities; absolute contraindications for NSAIDs.
Interventions	Intervention characteristics Very early treatment (n = 35) Medication: site A used IV ibuprofen‐lysine and site B used IV indomethacin. Dose and frequency: IV ibuprofen‐lysine was used at a dose of 10 mg/kg on day 1 followed by 5 mg/kg on days 2 and 3; IV indomethacin was used at a dose of 0.2 mg/kg on day 1 followed by 0.1 mg/kg on days 2 and 3. Timing of treatment: treatment initiated within 72 hours after birth Expectant management (n = 37) Normal saline given IV
Outcomes	Relevant outcomes for this study included: mortality; surgical PDA ligation; receipt of any pharmacotherapy for hs‐PDA; receipt of rescue PDA pharmacotherapy; CLD; pulmonary hemorrhage; severe IVH, PVL, NEC; spontaneous intestinal perforation; gastrointestinal bleed; ROP requiring treatment; sepsis; duration of ventilatory support; postnatal steroid use; duration of hospitalization.
Identification	Sponsorship source: financial support was obtained through a grant of the John Hunter Hospital Charitable Trust, the Hunter Children’s Research Foundation, Mill House Foundation and Heart Research Australia. Country: Australia Setting: Department of Neonatology, John Hunter Children’s Hospital and University of Newcastle, NSW, Australia; Department of Neonatology, Royal North Shore Hospital and University of Sydney, NSW, Australia Authors name: Dr. Koert de Waal Institution: Department of Neonatology, John Hunter Children’s Hospital and University of Newcastle, NSW, Australia Email: koert.dewaal@hnehealth.nsw.gov.au Address: Department of Neonatology, John Hunter Children’s Hospital and University of Newcastle, NSW, Australia
Notes	The study abstract was published as a conference proceeding. All relevant data were obtained from personal communication with the contact author.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	A random number was generated with a variable block size for each site.
Allocation concealment (selection bias)	Low risk	The list was held by pharmacy and each new recruit was randomly allocated using sealed envelopes to receive either NSAID therapy or an equal volume of normal saline.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	The intervention group received indomethacin and the placebo group received an identical volume of normal saline. Only the pharmacy was aware of the treatment allocation. The bedside staff were blinded to the intervention.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Medical staff performing cardiac ultrasounds and pathology staff were blinded to the treatment allocation. A blinded research nurse performed the oxygen challenge test at 36 weeks post‐conceptual age.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Results for all randomized infants were reported.
Selective reporting (reporting bias)	Low risk	The trial was prospectively registered with the Australia New Zealand Clinical Trials Registry (Registration number: ACTRN12616000195459). There did not seem to be any deviations between the protocol and the final report as available from the authors. The 2‐year neurodevelopmental outcomes are not yet available.
Other bias	Low risk	Appeared free of other bias

EL‐Khuffash 2020.

Study characteristics
Methods	Single‐center, randomized controlled trial Study period: 2016 to 2020
Participants	Inclusion criteria: all infants < 29 weeks' gestation (up to 28 weeks and 6 days) with a hemodynamically significant PDA (defined as being identified on comprehensive echocardiography between 36 and 48 hours of life) were considered eligible for inclusion. The PDA risk score was calculated based on the following formula: (Gestation in weeks × ‐1.304) + (PDA diameter in mm × 0.781) + (left ventricular output in mL/kg/min × 0.008) + (maximum PDA velocity in m/s × ‐1.065) + (LV a wave in cm/s × ‐0.470) + 41, where 41 is the constant of the formula. Infants with a risk score ≥ 5.0 were deemed to be at high risk of developing CLD/Death and were randomized to either arm. Infants with a low risk score (< 5.0) were not randomized but followed to discharge in a similar fashion to randomized infants. Exclusion criteria: potentially eligible infants were excluded from the study if any of the following criteria were identified at screening: lack of consent of study investigators to carry out echocardiogram examination; lethal congenital abnormality or obvious syndrome; suspected pulmonary hypoplasia; known or suspected NEC; thrombocytopenia: platelet count < 100/mm2; impaired renal function with a creatinine > 100 µmol/L; oliguria < 1 ml/kg/hour; culture positive sepsis; congenital heart disease other than a PDA or a patent foramen ovale (PFO); active bleeding including grade 3 or higher IVH or gastrointestinal hemorrhage.
Interventions	Intervention characteristics Very early treatment (n = 30) Medication: intravenous Ibuprofen Dose and frequency: 10 mg/kg, followed by 2 doses of 5 mg/kg 24 hours apart administered as a short infusion over 15 minutes. The patency of the ductus was assessed 24 hours after the last Ibuprofen dose using echocardiography. If the PDA remained open (PDA diameter > 1.5 mm), then a 2nd course of Ibuprofen was given. No further doses of Ibuprofen were administered. Timing of treatment: All infants were randomized between 36 to 48 hours of age. Expectant management (n = 30) IV placebo: infants in the expectant management group received intravenous dose of placebo (normal saline), at a volume equivalent to that in the intervention group (2 mL/kg 1st dose; 1 mL/kg 2nd & 3rd doses), which was also administered as a short infusion over 15 minutes. Following completion of the course, the patency of the ductus was assessed by echocardiography 24 hours after the last placebo dose. If the PDA remained open (defined as any identifiable flow), then a 2nd course of placebo was given.
Outcomes	Relevant outcomes for this study included: chronic lung disease; mortality; surgical PDA ligation; receipt of any pharmacotherapy for hs‐PDA; receipt of open label PDA pharmacotherapy; pulmonary hemorrhage; severe IVH, PVL, NEC, ROP requiring treatment; sepsis; duration of ventilatory support and supplemental oxygen; postnatal steroid use; duration of hospitalization.
Identification	Sponsorship source: this Study was funded by a Research Grant from the Temple Street Hospital Foundation (Ref: RPAC 16‐03). The National Children’s Research Centre funded PhD fees associated with this project. Country: Ireland Setting: Neonatal Intensive Care Unit Authors name: Dr. Afif EL‐Khuffash Institution: Department of Neonatology, The Rotunda Hospital, Dublin, Ireland Email: afifelkhuffash@rcsi.com
Notes	The unpublished study results were obtained from personal communication with the contact author.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	A computer‐generated central randomization scheme was used to assign the infants to the 2 arms in a 1:1 ratio.
Allocation concealment (selection bias)	Low risk	Quote: "The study pharmacist received a file containing the sequence of treatment group assignments for the cohort from a statistician who was not otherwise involved in the study. Access to the file was restricted to selected pharmacy personnel and was both encrypted and password protected on the pharmacy server".
Blinding of participants and personnel (performance bias) All outcomes	Low risk	The trial pharmacist (not involved in recruitment, allocation, and data collection) prepared the trial drug or placebo and issued the syringe for infusion to the trial investigator team for administration. Ibuprofen preparation is colorless, odorless and was indistinguishable from the saline preparation used for the placebo arm. Trial participants and their families, healthcare providers, data abstractors, echocardiographers, primary outcome assessors, and data analysts remained blind, throughout to study, to the randomized allocation.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	As above. Furthermore, the primary and secondary outcome assessors were also blinded to the PDA status of the infants during data collection.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Results for all randomized infants were reported.
Selective reporting (reporting bias)	Low risk	The trial was registered with the ISRCTN (13281214) and the European Union Drug Regulating Authorities Clinical Trials Database (2015‐004526‐33). There did not seem to be any deviations between the protocol and the final report as available from the authors.
Other bias	Low risk	Appeared free of other bias

Gersony 1983.

Study characteristics
Methods	Multicenter (13 centers), randomized controlled trial Study period: April 1979 to April 1981
Participants	Inclusion criteria: infants with a birth weight < 1750 grams with a hemodynamically significant PDA in the first 14 days after birth. This review specifically looks at the subgroup of infants who were treated within the first 5 days after birth. Criteria for diagnosis of hs‐PDA: 1st‐order clinical criteria included the presence of a continuous murmur, a systolic murmur, or when no murmur was present, need for ventilatory support for at least 48 hours. Given one of these, the infant was evaluated for the presence of 2nd‐order criteria, which consisted of clinical signs such as hyperactive precordium, increased pulse pressure (or bounding pulses), tachycardia (heart rate ≥ 170 bpm), tachypnea (respiratory rate ≥ 70/min), hepatomegaly (≥ 3 cm below right costal margin), and the need for varying levels of ventilatory support, as well as noninvasive findings: LA/AO by echocardiography ≥ 1.15 and cardiomegaly with pulmonary plethora on chest radiograph. Exclusion criteria: weight less than or equal to 500 grams; congenital anomalies; chromosomal abnormalities; death within the first 24 hours; admission to study facility after 14 days of age; Contraindications to indomethacin: BUN greater than or equal to 30 mg/dL serum creatinine clearance greater than or equal to 1.8 mg/dL total urine output less than or equal to 0.6 mL/kg/hour over the preceding 8 hours platelet count less than 60000/mm3 stool Hematest greater than or equal to 3+ (or "moderate" to "large") evidence of bleeding diathesis clinical or radiographic evidence of necrotizing enterocolitis clinical or noninvasive evidence of intracranial hemorrhage within preceding 7 days.
Interventions	Intervention characteristics Early treatment (n = 13) Medication: IV Indomethacin Dose and frequency: 0.2 mg/kg body weight for the initial dose. Infants less than 48 hours of age at the time of trial entry received 0.1 mg/kg body weight for the 2nd and 3rd doses of the drug. Infants who were 2 to 5 days of age at the time of the first dose received 0.2 mg/kg for the 2nd and 3rd doses. Timing of treatment: early treatment initiated within 5 days in this subgroup of interest Expectant management (n = 28) IV placebo The comparison group received usual medical therapy, which included fluid restriction, diuretic use, and occasionally, digoxin.
Outcomes	Relevant outcomes for this study included: death; surgical ligation; BPD; pneumothorax; intracranial hemorrhage; other bleeding; NEC; sepsis; elevated creatinine clearance; retrolental fibroplasia grades 3 and 4; duration of mechanical ventilation; age at home discharge.
Identification	Sponsorship source: Grant HL 23121 from the National Heart, Lung and Blood Institute Country: USA Setting: 13 centers across USA Comments: Dr Morton Rosenberg of Merck, Sharp & Dohme Research Laboratories were acknowledged for furnishing indomethacin and placebo. Authors name: Welton M. Gersony Institution: Columbia University College of Physicians and Surgeons, New York Address: Columbia University College of Physicians and Surgeons, Pediatric Cardiology, 620 West 168th St., New York, NY 10032
Notes	This review specifically looked at the subgroup of infants with birth weight < 1000 grams and who were treated within the first 5 days after birth.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Assignment of the initial treatment (1‐third indomethacin and 2‐thirds placebo in trial A) was done by random number generation, within blocks of 9 (3 indomethacin, 6 placebo).
Allocation concealment (selection bias)	Low risk	Opaque vials containing indomethacin or placebo, which formed colorless solutions, were provided to each clinical center by the coordinating center and were administered sequentially to infants on trial entry.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Physicians were blinded as to the therapy given.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Although not explicitly stated, blinding of treatment seemed adequate to ensure that outcome assessment was accomplished in a blinded fashion.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All outcomes reported for all infants randomized to this specific group under consideration (birth weight < 1000 g who were treated before 5 days of age)
Selective reporting (reporting bias)	Unclear risk	We could not judge if there were any deviations from the original protocol.
Other bias	Low risk	Appeared free of other bias

Ghanem 2010.

Study characteristics
Methods	Single center, randomized controlled trial Study period: November 2006 to April 2008
Participants	Inclusion criteria: gestational age < 32 weeks and birth weight < 1500 g; Postnatal age between 48 and 96 hours; RDS on chest radiograph necessitating treatment with surfactant; mechanical ventilation, and need for oxygen supplementation above 25%; echocardiographic evidence of hemodynamically significant PDA (left atrium/aortic root diameter ratio > 1.4 or ductal size > 1.5 mm). Exclusion criteria: presence of major congenital anomalies; IVH of grade 3 according to the classification by Papile 1978. within the previous 24 hours; serum creatinine level greater than or equal to 1.5 mg%; serum urea nitrogen concentration > 50 mg%; platelet count less than or equal to 60,000/cc; a tendency to bleed (defined by the presence of hematuria, blood in the endotracheal aspirate, gastric aspirate, or stools and/or oozing from puncture sites); hyperbilirubinemia necessitating exchange transfusion.
Interventions	Intervention characteristics Early treatment (n = 33) Medication: oral Ibuprofen Dose and frequency: initial 10 mg/kg followed by 2 doses of 5 mg/kg at 24 hours interval. The 2nd and 3rd doses were given if after the 1st dose PDA was still hemodynamically significant, as demonstrated by echocardiography, and there was no evidence of deterioration in brain ultrasonography. Timing of treatment: treatment initiated within 48 to 96 hours after birth following echocardiographic‐Doppler ultrasound evaluation. Expectant management (n = 33) Medication: oral (feeding tube) placebo (unspecified) Fluid intake began at 70 to 80 mL/kg/d and was increased by 20 to 30 mL/kg/d to a maximum of 160 mL/kg/d by the end of the 1st week, adjusted according to body weight. Furosemide was not used during the 1st week of life.
Outcomes	Relevant outcomes for this study included: mortality; CLD; severe IVH, PVL, NEC; spontaneous intestinal perforation; sepsis; duration of ventilatory support; time to regain birth weight; time to full enteral feeds; duration of hospitalization.
Identification	Sponsorship source: none Country: Saudi Arabia Setting: NICU at Al‐Jedaani Hospital, Saudi Arabia Authors name: Dr. Sabry Ghanem Institution: Al‐Jedaani Hospital, Saudi Arabia Email: sabryghanem1@yahoo.com Address: Department of Pediatrics, Alazhar University, Cairo, Egypt
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	The authors mentioned that the infants were divided into 2 groups after written parental consent. The authors did not mention if the groups were randomized.
Allocation concealment (selection bias)	High risk	Allocation concealment seemed unlikely as it did not appear that the infants were randomized.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	It was unclear if the personnel were blinded during the study.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	The echocardiographic assessment was done by a cardiologist blinded to the treatment allocation. However, it was unclear if the assessors for all reported outcomes were blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Outcomes reported for all enrolled infants
Selective reporting (reporting bias)	Unclear risk	We could not judge if there were any deviations from the original protocol.
Other bias	Low risk	Appeared free of other bias

Kaapa 1983.

Study characteristics
Methods	Single‐center, randomized controlled trial Study period: not mentioned
Participants	Inclusion criteria: preterm infants with idiopathic respiratory distress syndrome (IRDS); presence of an umbilical artery catheter; clinical features of hs‐PDA (murmur, bounding pulses and hyperactive precordium) along with documentation of open PDA through serial retrograde aortogram Exclusion criteria: infants whose PDA was found to be closed on the aortogram prior to institution of the intervention were excluded from the analysis.
Interventions	Intervention characteristics Very early treatment (n = 13) Medication: oral indomethacin Dose and frequency: 0.2 mg/kg once, repeated if necessary at 24‐hour intervals Timing of treatment: all infants were treated prior to 72 hours of age. Expectant management (n = 14) No pharmacological treatment When cardiac failure due to ductal shunting was established, treatment with fluid restriction, diuretics and digitalis was instituted.
Outcomes	Relevant outcomes for this study included: mortality; receipt of any pharmacotherapy for a hemodynamically significant PDA; CLD; oliguria; duration of ventilatory support; duration of supplementary oxygen.
Identification	Sponsorship source: this study was supported by the Alma and K. A. Snellman Foundation. Country: Finland Setting: Departments of Paediatrics, University of Oulu, Oulu, Finland Authors name: Dr Pekka Kääpä Institution: Departments of Paediatrics, University of Oulu, Oulu, Finland Address: Department of Paediatrics University of Oulu SF‐902 20 OUIU 22 Finland
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Infants were randomly allocated to 2 groups. However, method of sequence generation was not specified.
Allocation concealment (selection bias)	Low risk	Patients allocated to treatment using unmarked envelopes
Blinding of participants and personnel (performance bias) All outcomes	High risk	The treatment group was known as the treatment group and received oral medications; the control group received nothing.
Blinding of outcome assessment (detection bias) All outcomes	High risk	The treatment group was known as the treatment group and received oral medications; the control group received nothing.
Incomplete outcome data (attrition bias) All outcomes	High risk	Data on duration of ventilatory support and duration of supplemental oxygen not provided for all randomized infants
Selective reporting (reporting bias)	Unclear risk	Could not judge if there were any deviations from the original protocol as study conducted in pre‐registration era
Other bias	Unclear risk	Given this was an unblinded trial, unclear if there were other sources of bias

Kluckow 2014.

Study characteristics
Methods	Multicenter (3 centers), randomized controlled trial Study period: January 2007 to December 2010
Participants	Inclusion criteria: infants born < 29 weeks; born or transferred to one of the 3 hospitals in the study; initial screening cardiac and cranial ultrasound before 12 hours after birth; documented hs‐PDA on echocardiogram. Hemodynamically significant PDA defined as: PDA diameter > 1.8 mm at postnatal age 3 to 5 hours, > 1.6 mm at post‐natal age 6 to 8 hours and > 1.3 mm at postnatal age 9 to 12 hours Exclusion criteria: infants had a grade 2 to 4 PIVH; PH; critically unstable; not expected to survive; significant congenital anomaly; structural health disease; specific contraindications to indomethacin treatment (thrombocytopenia (< 50000/mm3), raised creatinine (> 120 mmol/L), severe growth restriction (< 3rd centile) with abnormal umbilical Doppler arterial flows, infants with ductal right to left shunting for more than 33% of the cardiac cycle, indicative of pulmonary hypertension, if not resolved by 12 hours)
Interventions	Intervention characteristics Very early treatment (n = 44) Medication: IV indomethacin (indocid PDA, Merck, Sharpe and Dome at a concentration of 1 mg/2 mL) Dose and frequency: loading dose of 0.2 mg/kg followed by 0.1 mg/kg for 2 further doses at 24‐hour intervals administered by 30‐minute infusion Infants started treatment between 3 to 12 hours. Expectant management (n = 48) Placebo (normal saline)
Outcomes	Relevant outcomes for this study included: mortality; surgical PDA ligation; receipt of rescue pharmacotherapy for hs‐PDA; CLD; postnatal corticosteroid use; pulmonary hemorrhage; IVH; PVL; NEC; spontaneous intestinal perforation; gastrointestinal bleeding; severe ROP; sepsis. Follow‐up study published as a conference proceeding by Varghese 2016 reported the following outcomes at 2 to 3 years of age: moderate/severe cognitive delay; moderate/severe motor delay; moderate/severe language delay.
Identification	Sponsorship source: in part funded by Heart Research Australia Country: Australia Setting: 3 neonatal intensive care units in Australia (1) Royal North Shore Hospital, Sydney Australia (2) Centre for Neonatal Research and Education, University of Western Australia, Perth (3) Royal Prince Alfred Women and Babies Hospital, Sydney, Australia Authors name: Dr Martin Kluckow Institution: Department of Neonatology, Royal North Shore Hospital, Sydney, Australia Email: martin.kluckow@ sydney.edu.au Address: Department of Neonatology Royal North Shore Hospital. Level 5, Douglas Building, St. Leonards, Sydney, NSW 2065 Australia
Notes	Follow‐up data as reported by Varghese 2016 were available for 79 out of the 92 randomized infants.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random number‐generated table
Allocation concealment (selection bias)	Unclear risk	Sequence generated using random‐number generated table. No comment made on allocation concealment
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Syringes containing either the study drug or a saline placebo were prepared by pharmacy. Clinicians, investigators and nursing staff were blinded.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	As above
Incomplete outcome data (attrition bias) All outcomes	Low risk	All outcomes assessed during hospital stay
Selective reporting (reporting bias)	Low risk	The trial was registered with the Australia New Zealand Clinical Trials Registry 12 days after the first participant enrollment. The results have been reported as per the registered protocol. Hence a major protocol deviation affecting a substantial proportion of infants was unlikely to have occurred.
Other bias	Unclear risk	The desired sample size for the trial was 170 in each arm. However, due to lack of availability of the study drug, the trial was terminated early after randomizing 92 infants.

Knight 2011.

Study characteristics
Methods	Randomized controlled trial
Participants	Inclusion criteria: infants < 28 weeks GA or < 1250 g between 2 to 6 chronological days with a hemodynamically significant PDA hs‐PDA defined by echocardiography as: PDA diameter 1.5 mm or greater and reversed descending aortic diastolic flow Exclusion criteria: not explicitly stated
Interventions	Intervention characteristics Early treatment (n = 13) Medication: IV indomethacin Dose and frequency: 200/100 mcg/kg IV 24 hourly × 6. The course could be repeated, with strict criteria for further PDA treatment. Timing of treatment: Infants randomized between 2 to 6 days of age Expectant management (n = 13) Received placebo. Details not provided
Outcomes	Relevant outcomes included: open‐label indomethacin treatment; peak inspiratory pressure; fraction inspired oxygen during treatment.
Identification	Country: Australia Setting: Neonatal Intensive Care Unit. Unclear if single‐center or multicenter trial Authors name: Dr D Knight Institution: Mater Mothers’ Hospital, Brisbane, Australia
Notes	This study was published as a conference abstract.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	It was not stated how allocation was completed.
Allocation concealment (selection bias)	Unclear risk	Unclear if allocation was concealed
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	It was not clear if staff were blinded to drugs administered.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Unclear if outcome assessors were blinded to the intervention
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Unclear if all randomized infants were accounted for
Selective reporting (reporting bias)	Unclear risk	Unable to judge; study protocol not found
Other bias	Unclear risk	Limited report to judge

Krauss 1989.

Study characteristics
Methods	Single‐center, randomized controlled trial Study period: not described
Participants	Inclusion criteria: preterm infants with birth weight < 1500 grams, ventilator dependent at 24 hours after birth, preterm infants and diagnosed with hs‐PDA by clinical and echocardiographic examination No definition of hs‐PDA provided Exclusion criteria: serum creatinine level greater than 110 μmol/L; serum urea nitrogen level greater than 10.5 mol/L; platelet count less than 60 X 109/L; urinary output less than 0.6 mL/kg/h over 8 hours preceding evaluation; laboratory evidence of infection or of bleeding; NEC; severe (stage 4) intraventricular hemorrhage; congenital anomalies.
Interventions	Intervention characteristics Early treatment (n = 12) Medication: IV indomethacin Dose and frequency: 3 doses, 0.2 mg/kg per dose Timing of treatment: treatment initiated between 72 hours and 96 hours of age Expectant management (n = 15) No treatment provided The infants received usual medical management.
Outcomes	Relevant outcomes for this study included: mortality; surgical PDA ligation; receipt of any pharmacotherapy for hs‐PDA.
Identification	Sponsorship source: none reported Country: USA Setting: New York Hospital, New York Comments: Authors name: Dr Alfred Krauss Institution: Cornell University Medical College Email: not available Address: New York Hospital, 1300 York Ave, NewYork, NY
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Assigned to group by random number
Allocation concealment (selection bias)	Unclear risk	No information provided on allocation concealment
Blinding of participants and personnel (performance bias) All outcomes	High risk	The treatment group was known. The treatment group received a dose of indomethacin while the control group did not.
Blinding of outcome assessment (detection bias) All outcomes	High risk	The treatment group was known. The treatment group received a dose of indomethacin while the control group did not.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Results for all randomized infants were reported.
Selective reporting (reporting bias)	Unclear risk	Unable to judge; study protocol not found
Other bias	Low risk	Free of other bias

Lin 2012.

Study characteristics
Methods	Single‐center, randomized controlled trial Study period: January 2009 to December 2011
Participants	Inclusion criteria: preterm infants ≤ 32 weeks GA and BW < 1500 grams with a hemodynamically significant PDA Definition of hemodynamically significant PDA: presence of clinical signs of PDA (precordial murmur, active precordium, heart rate > 160, bounding pulse, wide pulse pressure > 25 mmHg, cardiomegaly and increased pulmonary vasculature on chest X‐ray) with confirmation on echocardiogram (PDA diameter ≥ 1.5 mm with left to right shunt) within 24 hours after birth. Exclusion criteria: chromosome anomaly; other congenital anomalies of heart, kidneys, and GI tract; death or discharge within 7 days of life; grade 3 or 4 IVH on cranial us done within 24 hours of life; contraindication for ibuprofen: BUN > 8.9 mmol/L or creatinine > 141 umol/L; urine output < 0.5 mL/kg/hr; platelet count < 80 x 109/L; has bleeding tendency or DIC suspected or confirmed NEC.
Interventions	Intervention characteristics Very early treatment (n = 32) Medication: oral ibuprofen Dose and frequency: initial dose of 10 mg/kg followed by 2 doses of 5 mg/kg at 24 hour intervals Timing of treatment: initiated within 24 hours of birth Expectant management (n = 32) Oral placebo (normal saline) Dose and frequency: 1st dose 1 mL/kg, 2nd dose 24 hours later of 0.5 mL/kg, 3rd dose 48 hours later of 0.5 mL/kg Regular treatment of PDA such as oxygen administration, fluid restriction, acidosis correction, diuretics and antibiotics continued.
Outcomes	Relevant outcomes for this study included: mortality; receipt of repeat courses of pharmacotherapy for a persistent hs‐PDA; CLD; pulmonary hemorrhage; IVH; severe IVH; PVL; NEC; spontaneous intestinal perforation; sepsis; duration of ventilatory support; duration of hospitalization.
Identification	Sponsorship source: not reported Country: China Setting: Hospital of Xiamen City Authors name: Dr Xin‐Zhu Lin Institution: Xiamen Women and Children’s Hospital Email: hqchen_996@yahoo.com Address: Hospital of Xiamen City, Xiamen, Fujian 361000, China
Notes	Article translated from Chinese
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer random‐number generator
Allocation concealment (selection bias)	Low risk	Reported as done but no method specified
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Not enough information reported; unclear
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not enough information reported; unclear
Incomplete outcome data (attrition bias) All outcomes	Low risk	Outcomes reported on all infants.
Selective reporting (reporting bias)	Unclear risk	We could not comment on selective outcome reporting as we did not have access to any study protocol.
Other bias	Low risk	Appeared free of other sources of bias

Merritt 1981.

Study characteristics
Methods	Single‐center, randomized controlled trial Study period: not specified
Participants	Inclusion criteria: infants ≤ 1.35 kg at birth who had severe respiratory distress syndrome (defined as increased oxygen requirement, tachypnea, retractions, expiratory grunting, and radiographic patterns with air bronchograms) and ventilatory failure (FiO2% > 0.6 with PaO2 or tcvO2 < 50 torr, or apnea) requiring institution of IMV during the first hours of life with evidence of hs‐PDA hs‐PDA defined as: clinical signs of PDA characterized by a systolic murmur, increased brachial and femoral pulses, and precordial hyperkinesis, in combination with a cardio‐thoracic ratio > 0.6 on X‐ray; or, left atrium to aortic root ration > 1.2; or, demonstration of PDA on retrograde aortogram. Exclusion criteria: total serum bilirubin > 10 mg/dL; blood urea nitrogen > 25 mg/dL; serum creatinine > 1.2 mg/dL; platelet count < 50,000/mm3; signs of necrotizing enterocolitis; clinical evidence of bleeding from the umbilicus, needle pricks, or the finding of occult blood in the stool. Exclusion criteria were defined only for the treatment group.
Interventions	Intervention characteristics Very early treatment (n = 11) Medication: IV indomethacin (lyophilized indomethacin sodium) Dose and frequency: initial dose 0.2 mg/kg of actual weight. This dose was repeated up to 3 times at 24‐hour intervals if the PDA persisted. Timing of treatment: treatment initiated within 72 hours after birth. Expectant management (n = 13) Fluid restriction and/or furosemide 1 to 2 mg/kg every 12 hours
Outcomes	Relevant outcomes for this study included: mortality; CLD; surgical PDA ligation; duration of assisted ventilation.
Identification	Sponsorship source: none reported Country: USA Setting: University of Rochester School of Medicine and Dentistry Authors name: Dr T Allen Merritt Institution: University of Rochester School of Medicine and Dentistry Address: Department of Pediatrics, University Hospital, 225 W. Dickinson, San Diego, CA 92103
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomized using the last digit of 6 numbers generated by the hospital at the time of the infant's admission
Allocation concealment (selection bias)	Unclear risk	It was unclear how randomization numbers were distributed.
Blinding of participants and personnel (performance bias) All outcomes	High risk	This was not a placebo‐controlled trial. So, the personnel were not blinded to the treatment allocation.
Blinding of outcome assessment (detection bias) All outcomes	High risk	The study staff were not blinded to the treatment arms.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Unclear how many infants were eventually randomized. Therefore we could not judge if all randomized infants were accounted for.
Selective reporting (reporting bias)	Unclear risk	It was unclear if there were deviations from the original protocol.
Other bias	High risk	Exclusion criteria separately applied to treatment group, making treatment group and control group systematically different

Sosenko 2012.

Study characteristics
Methods	Single‐center, randomized controlled trial Study period: January 2008 to August 2010
Participants	Inclusion criteria: preterm infants birth weights between 500 and 1250 grams and gestational ages between 23 and 32 weeks who were > 24 hours old but ≤ 14 days old and were diagnosed to have clinical symptoms of PDA (metabolic acidosis (base excess < ‐7), mean blood pressure less than weeks in gestation or requiring vasopressor support, bounding pulses/hyperactive precordium, or systolic murmur) that was confirmed on color Doppler echocardiogram (presence of a PDA with either predominantly left‐to‐right or bidirectional shunt) Exclusion criteria: severely small for gestational age (> 3 SD less than the mean birth weight for gestational age); major congenital malformations; proven sepsis (positive blood culture results); serum creatinine level > 1.75 mg/dL; oliguria (urine output < 1 cc/kg/hr); pulmonary hypertension (with right‐to‐left PDA shunt); abdominal pathology (abdominal distension, discoloration, abnormal abdominal radiograph); bleeding diathesis; terminal condition (intractable respiratory failure, intractable hypotension, no expectation of survival beyond 48 hours); symptoms of a highly significant PDA at study enrollment (defined as signs of PDA plus the presence of pulmonary hemorrhage alone, or signs of PDA plus cardiomegaly and pulmonary edema on chest radiography plus one of the following: either hypotension not caused by an identifiable cause other than PDA requiring vasopressor dose > 10 mcg/kg/min or respiratory failure (not caused by an identifiable cause other than PDA))
Interventions	Intervention characteristics Early treatment (n = 54) Medication: IV ibuprofen lysine (NeoProfen, Ovation (now Lundbeck) Pharmaceuticals, Deerfield, Illinois) Dose and frequency: initial dose of 10 mg/kg, followed by 2 doses of 5 mg/kg each, every 24 hours by slow intravenous infusion Additional comments on dose and frequency: if an enrolled infant had persistent signs of PDA without meeting hemodynamic significance (as defined in the study), a prolonged course of blinded assigned study drug (2 additional doses of blinded 5 mg/kg ibuprofen or equivalent volume of blinded placebo) or a repeat course of blinded assigned study drug was administered at the discretion of the clinical team. Timing of treatment: most of the study infants were treated within the 1st 5 days. Expectant management (n = 51) IV placebo (5% dextrose)
Outcomes	Relevant outcomes for this study included: mortality; surgical PDA ligation; receipt of any pharmacotherapy for hs‐PDA; receipt of rescue pharmacotherapy for hs‐PDA; CLD; postnatal steroid use; pneumothorax; severe IVH, PVL, NEC; spontaneous intestinal perforation; severe ROP; sepsis; duration of ventilatory support; duration of supplementary oxygen.
Identification	Sponsorship source: supported by an unrestricted grant from Ovation (now Lundbeck) Pharmaceuticals and University of Miami. (Project: new born) Country: USA Setting: tertiary level Neonatal Intensive Care Unit at Holtz Children’s/University of Miami/Jackson Memorial Medical Center (Miami, Florida) Authors name: Ilene R. S. Sosenko Institution: Division of Neonatology, Department of Pediatrics, University of Miami/Miller School of Medicine, Miami, FL Email: isosenko@miami.edu Address: Department of Pediatrics/Neonatology (R131), University of Miami Miller School of Medicine, PO Box 016960, Miami, FL33101
Notes	The study was included due to the following reasons: most infants in the early treatment group initiated treatment within 5 days (median 3 days (10th to 90th percentile: 2 to 5 days)); their definition for PDA was consistent with the definition of hemodynamically significant PDA as defined in the protocol.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomly assigned by a random number generator.
Allocation concealment (selection bias)	Low risk	Allocation at pharmacy level, in sealed envelopes
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Clinicians, investigators and nursing staff were blinded to the study group to which the baby was assigned and the medication the baby was receiving. Only the neonatal pharmacists were aware of the study group of each baby and were responsible for preparing the ‘‘blinded’’ ibuprofen or ‘‘blinded’’ placebo study drug.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	As mentioned above
Incomplete outcome data (attrition bias) All outcomes	Low risk	Outcomes were reported on all randomized infants.
Selective reporting (reporting bias)	Low risk	This study was registered (#NCT00802685). There were no deviations from the protocol, except that the study had to be stopped early after two‐thirds of the planned enrollment when the study drug was no longer available.
Other bias	Unclear risk	After two‐thirds of the authors planned enrollment (105 of 168 infants), NeoProfen was recalled and was not available for use. Authors did not enroll the full anticipated number of participants.

Van Overmeire 2001.

Study characteristics
Methods	Multicenter (4 centers) randomized controlled trial Study period: not specified
Participants	Inclusion criteria: gestational age < 32 weeks; respiratory distress syndrome for which ventilatory support was needed by means of mechanical ventilation (either conventional or prophylactic high‐frequency oscillatory ventilation) or continuous positive airway pressure with additional oxygen requirements above 30% fraction of inspired oxygen; age of 3 days; echocardiographic evidence of a PDA. Hemodynamically significant PDA was defined as PDA with moderate or severe shunt on echocardiography. The PDA shunt was defined as “moderate,” if a disturbed diastolic flow was easily detected in the main pulmonary artery with a diastolic reversed flow in the aorta beneath the ductus and a forward flow above the ductal insertion. PDA shunt was defined as “severe" if a diastolic backflow in the aorta was straightforward and if dilatation of the left atrium was present (LA/Ao > 1.5). Exclusion criteria: major congenital malformation; severe pulmonary hypertension with predominant right‐to‐left ductal shunting; life‐threatening infection or hydrops fetalis; contraindications for the use of indomethacin such as recent intraventricular hemorrhage (< 24 hours); clinical bleeding tendency as revealed by hematuria or blood in the endotracheal aspirate, gastric aspirate, or stools; oozing from venous or capillary puncture sites or platelet count < 60 × 109/L; urine output < 1 mL/kg/h during the preceding 8 hours; blood urea nitrogen > 14 mmol/L (40 mg/dL) or serum creatinine concentration > 140 μmol/L (1.6 mg/dL); and hyperbilirubinemia with impending exchange transfusion
Interventions	Intervention characteristics Early treatment (n = 64) Medication: IV Indomethacin Dose and frequency: 3 doses of 0.2 mg/kg each given during a 15‐minute infusion at 12‐hour intervals Timing of treatment: treatment initiated on day 3 Additional comments on timing of treatment: if there was still moderate to severe shunting through the ductus on day 7 in infants receiving respiratory support, the same indomethacin treatment was applied. Expectant management (n = 63) No placebo was given. Daily fluid intake was titrated to allow a 10% weight loss during the first 5 to 6 postnatal days. No pharmacological treatment in the first 7 days. If ventilatory support needed to be increased by at least 10% (FIO2 or mean airway pressure) from day 3 to day 7, a rescue indomethacin treatment was given. Otherwise, If there was still moderate to severe shunting through the PDA on day 7 in infants receiving respiratory support, indomethacin treatment was initiated.
Outcomes	Relevant outcomes for this study included: mortality; surgical PDA ligation; receipt of any pharmacotherapy for hs‐PDA; receipt of repeat courses of pharmacotherapy for a persistent hs‐PDA; receipt of rescue medical treatment for hs‐PDA; CLD, IVH, PVL, NEC; spontaneous intestinal perforation; sepsis; oliguria; duration of ventilatory support; duration of supplementary oxygen.
Identification	Sponsorship source: none Country: Belgium Setting: 4 Neonatal Intensive Care Units (1) University Hospital Antwerpen (2) Queen Paola Children’s Hospital, Antwerpe (3) Neonatal Intensive Care Unit, St‐Jan Ziekenhuis (4) Division of Neonatology, University Hospital Gent Authors name: Dr Bart Van Overmeire (MD, PhD) Institution: Department of Pediatrics, Division of Neonatology, University Hospital, Antwerpen Address: Department of Pediatrics, Division of Neonatology, University Hospital, Antwerpen, Wilrijkstraat 10, B‐2650 Edegem, Belgium
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Method of sequence generation not described
Allocation concealment (selection bias)	Low risk	Use of sealed envelopes
Blinding of participants and personnel (performance bias) All outcomes	High risk	The treatment group, likely known to healthcare providers and caregivers as the early treatment group, received a treatment dose by IV on day 3; the late treatment group received nothing.
Blinding of outcome assessment (detection bias) All outcomes	High risk	As above
Incomplete outcome data (attrition bias) All outcomes	Low risk	Outcomes reported for all randomized infants
Selective reporting (reporting bias)	Unclear risk	Study protocol unavailable
Other bias	Low risk	Appeared free of other bias

ASD: Atrial septal defect
BPD: Bronchopulmonary dysplasia
bpm: beats per minute
BUN: blood urea nitrogen
BW: birth weight
CLD: chronic lung disease
CTRI: Clinical Trials Registry of India
DIC: disseminated intravascular coagulation
GA: gestational age
GI: gastrointestinal
hs‐PDA:hemodynamically significant patent ductus arteriosus
IMV: intermittent mandatory ventilation
IRDS: idiopathic respiratory distress syndrome
IV: intravenous
IVH: intraventricular hemorrhage
LA:AO: Left atrium to aortic root ratio
LV: left ventricle
NaCI: Sodium Chloride
NEC: necrotizing enterocolitis
NICU: neonatal intensive care unit
NSAIDs: non‐steroidal anti‐inflammatory drugs
PDA: patent ductus arteriosus
PFO: patent foramen ovale
PH: pulmonary hypertension
PIVH: peri/intraventricular hemorrhage
PVL: periventricular leukomalacia
RDS:respiratory distress syndrome
ROP: retinopathy of prematurity
SD: standard deviation
us: ultrasound
VSD: ventricular septal defect

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Aranda 2009	Wrong indication
Babaei 2018	Late start of treatment
Betkerur 1981	Duplicate
Clyman 2019	Late start of treatment
Couser 1996	Wrong patient population
Harkin 2016	Wrong intervention
Jannatdoust 2014	Wrong patient population
Juujrvi 2019	Duplicate
Kluckow 2012	Duplicate
Mahony 1982	Wrong patient population
Mahony 1985	Wrong patient population
Ment 1985	Wrong patient population
Ment 1988	Wrong patient population
Ment 1994a	Wrong patient population
Ment 1994b	Wrong patient population
Monset Couchard 1983	Late start of treatment
Mullett 1982	Wrong patient population
Nestrud 1980	Late start of treatment
Neu 1981	Late start of treatment
Osborn 2003	Wrong study design
Peckham 1984	No data specifically on early treated infants
Rennie 1986	Wrong patient population
Rudd 1983	Late start of treatment
Salama 2008	Wrong comparator
Sangtawesin 2006	Wrong indication
Sangtawesin 2008	Wrong patient population
Weesner 1987	Wrong indication
Yanagi 1981	Late start of treatment
Yeh 1981	Late start of treatment
Yeh 1982	Late start of treatment

Characteristics of studies awaiting classification [ordered by study ID]

ACTRN12616001517460.

Methods	Study title: Early PARacetamol (EPAR) to promote early closure of the ductus arteriosus in preterm infants Randomized controlled trial Study period: 2016 to 2020
Participants	Preterm infants < 6 hours old; born at < 29 weeks' gestation with a hemodynamically significant PDA defined as: PDA size > 1 mm & < 30% right to left shunt
Interventions	Very early treatment: intravenous paracetamol at a dose of 15 mg/kg (1.5 ml/kg) as a single bolus, followed by 7.5 mg/kg (0.75 mL/kg) every 6 hours for 5 days Expectant management: 5% dextrose as placebo at 1.5 mL/kg as a single bolus, followed by 0.75 mL/kg every 6 hours for 5 days
Outcomes	Primary outcome: any intervention for management of PDA up to 5 days
Notes	Preliminary results presented at the PAS Neonatology Summer Webinar series 2020; details of the trial sought from primary author

NCT01630278.

Methods	Study title: Early ibuprofen treatment of patent ductus arteriosus (PDA) in premature infants (TRIOCAPI) Multicenter randomized controlled trial Study period: 2012 to 2019
Participants	Very preterm infants (born < 28 weeks' gestational age) with a large PDA, selected by an early echocardiogram within 12 hours after birth
Interventions	Very early treatment group received intravenous ibuprofen and placebo group received intravenous normal saline within 12 hours after birth
Outcomes	Primary outcome: 2‐year survival without cerebral palsy
Notes	Preliminary results presented at the PAS Neonatology Summer Webinar series 2020; details of the trial sought from primary author

PAS: Pediatric Academic Societies
PDA: patent ductus arteriosus

Characteristics of ongoing studies [ordered by study ID]

ISRCTN84264977.

Study name	Outcome after Selective early Closure of patent ductus ARteriosus (PDA) in extreme preterm infants: a randomised controlled trial (Baby‐OSCAR)
Methods	Multicenter randomized controlled trial
Participants	Preterm infants born between 23 + 0 to 28 + 6 weeks of gestation; less than 72 hours old; confirmed by echocardiography to have a large PDA which is at least 1.5 mm in diameter (determined by gain optimized color Doppler), has unrestrictive pulsatile (left to right) flow (ratio of flow velocity in PDA maximum (vmax) to minimum (vmin) > 2:1)) or, growing flow pattern (< 30% right to left), and no clinical concerns of pulmonary hypertension
Interventions	Very early treatment: intravenous ibuprofen started within 72 hours of age Control group: placebo (clear sterile preservative‐free solution for intravenous injection)
Outcomes	Primary outcome: composite outcome of incidence of death by 36 weeks' of postmenstrual age, or moderate or severe BPD at 36 weeks' of postmenstrual age
Starting date	1 July 2014
Contact information	Professor Samir Gupta; University Hospital of North Tees, Hardwick Road, Stockton‐on‐Tees TS19 8PE; Tel: 01642 624 232
Notes	EudraCT No.: 2013‐005336‐23

NCT02884219.

Study name	Early treatment versus expectative management of patent ductus arteriosus in preterm infants: a multicentre, randomised, non‐Inferiority trial in Europe (BeNeDuctus Trial)
Methods	Multicenter, randomized, non‐inferiority trial
Participants	Preterm infants (born at a GA less than 28 weeks) at postnatal age 24 to 72 hours with a hemodynamically significant PDA defined as: PDA diameter  > 1.5 mm and predominantly left‐to‐right transductal shunt (≥ 66% of the cardiac cycle)
Interventions	Very early treatment arm: intravenous ibuprofen started within 72 hours of age Expectative management arm: no cyclo‐oxygenase inhibitor drugs; no other conservative management strategies specified
Outcomes	The primary endpoint is the composite of mortality, and/or NEC (Bell stage ≥ IIa), and/or BPD at a PMA of 36 weeks
Starting date	The 1st patient was included in the study in December 2016.
Contact information	Tim Hundscheid, Department of Paediatrics, Division of Neonatology, Radboud University Medical Center, Nijmegen, Radboud Institute for Health Sciences, Amalia Children's Hospital, Internal postal code 804, Geert Grooteplein Zuid 10, 6525, GA, Nijmegen, The Netherlands Phone: +31 24 361 44 30, Email: tim.hundscheid@radboudumc.nl
Notes	ClinicalTrials.gov identifier: NCT02884219

BPD: Bronchopulmonary dysplasia
GA: gestational age
NEC: necrotizing enterocolitis
PDA: patent ductus arteriosus
PMA: postmenstrual age

Differences between protocol and review

2020

We made the following changes to the published protocol (Mitra 2019).

1. In the echocardiographic definition of hs‐PDA, we added the criterion “or used a scoring system that incorporated one or more of the echocardiographic measures mentioned above” to include any studies that used a PDA severity scoring system based on established echocardiographic measures to define an hs‐PDA. One included study (EL‐Khuffash 2020) met this expanded definition of hs‐PDA.

2. The outcome “proportion of infants receiving rescue medical treatment (COX inhibitor or acetaminophen dosing, or both) for a hemodynamically significant PDA in the expectant management group” was changed to “proportion of infants receiving rescue medical treatment (COX inhibitor or acetaminophen dosing, or both) for a hemodynamically significant PDA”. This outcome was defined as receipt of rescue pharmacotherapy outside the intended treatment protocol but within the early treatment period as defined by the respective study in either arm (not only the expectant management arm).

3. We did not define neurodevelopmental outcome in the protocol, and assessment was defined as one assessed by a standardized and validated assessment tool, a child developmental specialist, or both at any age reported (outcome data grouped at 12, 18, and 24 months if available). Based on the reported neurodevelopmental outcomes in the included studies, we reported neurodevelopmental outcomes as moderate/severe cognitive delay at 18 to 24 months, moderate/severe motor delay at 18 to 24 months and moderate/severe language delay at 18 to 24 months.

4. For assessment of certainty of evidence, we considered rating down the certainty of evidence by one level for risk of bias if more than 50% of the meta‐analytic weight was contributed by unclear or high risk of bias studies. This was not prespecified in the protocol.

As of July 2019, Cochrane Neonatal no longer searches Embase for its reviews. RCTs and controlled clinical trials (CCTs) from Embase are added to the Cochrane Central Register of Controlled Trials (CENTRAL) via a robust process (see How CENTRAL is created). Cochrane Neonatal has validated their searches to ensure that relevant Embase records are found while searching CENTRAL.

Also starting in July 2019, Cochrane Neonatal no longer searches for RCTs and CCTs on the following platforms: ClinicalTrials.gov or from The World Health Organization’s International Clinical Trials Registry Platform (ICTRP), as records from both platforms are added to CENTRAL on a monthly basis (see How CENTRAL is created). Comprehensive search strategies are executed in CENTRAL to retrieve relevant records. The ISRCTN Registry (at www.isrctn.com/, formerly Controlled‐trials.com), is searched separately.

Starting in September 2020, Cochrane Neonatal no longer searches for RCTs and quasi‐RCTs from CINAHL, as records are identified and added to CENTRAL on a monthly basis through Cochrane's Centralised Search Service project (see How CENTRAL is created at https://www.cochranelibrary.com/central/central-creation#CINAHL%20section).

For the 2020 update, we ran searches in the following databases: CENTRAL via CRS Web and MEDLINE via Ovid. The search strategies are available in Appendix 1. The initial search methods are available in Appendix 2.

Contributions of authors

SM conceived the project and drafted the review.

SM, AS, AvK participated in screening and data extraction.

TD reviewed all drafts and approved the final version of the review.

Sources of support

Internal sources

• No sources of support supplied

External sources

• Vermont Oxford Network, USA

  Cochrane Neonatal Reviews are produced with support from Vermont Oxford Network, a worldwide collaboration of health professionals dedicated to providing evidence‐based care of the highest quality for newborn infants and their families.

Declarations of interest

SM is the principal Investigator of a Canadian Institutes of Health Research (CIHR)‐funded prospective study on the relative effectiveness and safety of pharmacotherapeutic agents for treatment of patent ductus arteriosus in preterm infants. The CIHR grant will be paid to SM's institution.

AS has no interests to declare.

AvK has no interests to declare.

TD is an employee of EVERSANA Inc. EVERSANA Inc. consults for various pharmaceutical and device companies.

New