Steer 2003.

Study characteristics
Methods	Study design: double‐blind RCT Location: Australia Setting: one neonatal intensive care nursery of a large maternity teaching hospital (Mater Mothers’ Hospital, Brisbane) Duration: September 1993 to November 1995
Participants	Inclusion criteria: 127 infants < 31 weeks' gestational age who had received or who were anticipated to receive at least 48 hours of mechanical ventilation Exclusion criteria: major congenital abnormality, infection, major neurological condition, severe intraventricular hemorrhage, MV for a period greater than 28 days, previous exposure to methylxanthine therapy.
Interventions	This study contained three study arms with different doses. According to our definition outlined in the protocol, two of them were high doses (very high and moderately high) and one was a low dose. Very high dose: loading dose 60 mg/kg/day and maintenance dose 30 mg/kg/day Moderately high dose: loading dose 30 mg/kg/day and maintenance dose 15 mg/kg/day Low dose: loading dose 6 mg/kg/day and maintenance dose 3 mg/kg/day
Outcomes	Outcomes reported that are considered for this review: severe IVH ≥ grade 3; lesions indicative of brain injury (detected by US and MRI): severe IVH ≥ grade 3; duration of MV; retinopathy of prematurity ≥ grade 3; extubation failure (defined as either an inability to extubate from mechanical ventilation within 48 hours of caffeine loading for a planned extubation or the use of reintubation or doxapram within seven days of commencing caffeine therapy; any tachycardia; tachycardia leading to suspension of study intervention; necrotizing enterocolitis (NEC); and somatic growth.
Notes	Caffeine initiation time: postnatal mean (SD) age of 4.4 (2.5) days in the very high‐dose group, 3.8 (2.1) days in the moderately high‐dose group, and 3.5 (1.5) days in the low‐dose group Baseline imbalances: none The trial was terminated after an interim analysis after enrolling infants corresponding to approximately 50 % of the a priori calculated sample size No trial protocol available
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Enrolled infants were allocated to one of three group using a computer‐generated list of random numbers by a hospital pharmacist. The pharmacist was not associated in any other way with the study or with the clinical management of the infants.
Allocation concealment (selection bias)	Unclear risk	Not described.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	The identity of the treatments was not disclosed to the medical or nursing staff.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	The identity of the treatments was not disclosed to the investigators.
Incomplete outcome data (attrition bias) All outcomes	Low risk	8 out of 127 randomized infants failed to receive caffeine therapy after randomization; 9 out of 127 did not complete the planned seven day course of caffeine. Outcomes for these 17 infants are included in the ITT analysis.
Selective reporting (reporting bias)	Unclear risk	Trial protocol not available.
Other bias	Low risk	None known.