Kirpalani 2019.

Methods	International, multicenter, prospective, unblinded, RCT in 17 hospitals recruiting between 2014 to 2017 Infants were randomised using 1:1 allocation, variable block sizes and stratification by site and GA. A sealed opaque envelope was opened after delivery. A sample size of 592 was sufficient to detect a reduction in the rate of BPD/death from 65% to 52.5% with 80% power, adjusting for interim analyses and multiple births. Stopping rules for efficacy, and signals for harm (including death and early death at < 48 hours of life), were both prespecified. An independent DSMB including a neonatal ethicist reviewed all data
Participants	Infants from 23 to 26 weeks' gestational age were eligible if they required resuscitation for inadequate respiratory effort or heart rate < 100 bpm Exclusion criteria: resuscitation not provided; refusal of informed consent; clinically suspected pulmonary hypoplasia Consent was sought, (approved by local IRB boards) either antenatally (all sites) or by a deferred process (6 sites) A total of 460 infants were recruited; 34 families refused post‐waiver consent, and 1 infant had a missing primary outcome. Thus 425 were analysed
Interventions	Treatment with SI (Up to 2 SLI; first at 20 cmH₂O for 15 seconds, followed if needed by a second SI of 25 cmH₂O for 15 seconds) ‐ as compared to standard care
Outcomes	Primary outcome of BPD or death at 36 weeks' postmenstrual age The DSMB upon review (the 2nd for efficacy and the 4th for safety), halted the trial for harm. Demographics of infants did not significantly differ by group (Table 1). Rates of the primary outcome, or its 2 components, were not statistically different (Table 2). (RR 1.10, 95% CI 0.9 to 1.3). Rates of pneumothorax and IVH were similar. An excess of early deaths (< 48 hours of age) was seen in the SI arm (7.5% vs 1.4%, P = 0.002). Furthermore, in a blinded adjudication, 12/19 early deaths were considered as possibly attributable to resuscitation (SI N = 11 vs NRP 1), but no cause of death predominated
Notes	Study was registered at ClinicalTrials.gov (Identifier: NCT02139800). The DSMB halted the trial for harm.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	The trial used computer‐generated permuted block randomisation, with variable block sizes of 2, 4, or 6, stratified by site and gestational age
Allocation concealment (selection bias)	Low risk	Sealed opaque envelopes were used, colour coded by gestational age strata, and opened on confirming eligibility
Blinding (performance bias and detection bias) All outcomes	High risk	Assigned intervention could not be blinded to the resuscitation team
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	The primary outcome was analysed blinded to allocation. Unclear for the secondary outcomes
Incomplete outcome data (attrition bias) All outcomes	Low risk	All infants accounted for
Selective reporting (reporting bias)	Low risk	All outcomes specified in the protocol were reported in the manuscript
Other bias	Unclear risk	Trial lacks power because only 426 infants were enrolled (instead of 600)