Lotze 1998.

Methods	Multicentre (44 centres) study in the US. Study period September 1 1992 to October 23 1995. Blinding of randomisation: Yes (central randomisation) Blinding of treatment: Yes (dosing investigators) Complete follow‐up: Yes Blinding of outcome measurement: Yes Stratification: primary diagnosis disease severity (oxygenation index)
Participants	Infants > 2000 gm, gestational age > 36 weeks, age < 120 hrs with MAS, PPHN or sepsis and severe respiratory failure but without any major congenital anomalies or IVH > Grade I.
Interventions	The treatment group (n = 87) received modified bovine surfactant extract (Survanta, 100 mg/kg) or air placebo (up to 4 doses prior to ECMO and 4 additional doses if ECMO was required). The control group (n = 81) received air placebo.
Outcomes	PRIMARY: Need for ECMO, severe complications, mortality.
Notes	Only infants with MAS are included in this review. Data for this group were obtained by the authors of the first version of this review
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Central randomisation. Computer‐generated random numbers Stratification: primary diagnosis disease severity (oxygenation index).
Allocation concealment (selection bias)	Low risk	The treatment assignments were made by having the pharmacist or dosing investigator at each site report the primary diagnosis and mean entry oxygen index for each participant to a central randomisation centre, Bio‐Pharm Clinical Services, Inc., which issued a participant number and treatment assignment on the basis of a computer‐generated random number.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Study treatments were administered by dedicated dosing investigators at each site. The dosing investigators shielded the infant with drapes or a screen during treatment, and all other personnel left the immediate bedside area during the dosing procedure. The dosing investigator took the same amount of time to prepare and administer either treatment, and when treatment was complete, all supplies were stored in a locked area. Dosing investigators were prohibited from participating in any other aspect of the infants care and from revealing the treatment assignment.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	With the exception of dosing, the infant's clinical care during the 28 days of the study was provided by clinical investigators who were unaware of the treatment assignment.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All 330 randomised infants accounted for (168 of these infants were enrolled on the basis of MAS, and the remainder on the basis of PPHN or sepsis) . Two infants were later withdrawn from the study when consent was withdrawn. Their limited data were subsequently excluded from analysis. The diagnosis on which their enrolment was based was not stated.
Selective reporting (reporting bias)	Unclear risk	The protocol for the study was not available to us so we cannot ascertain if there were any deviations from the protocol