Rocha 2002.

Methods	Randomised controlled trial, stratified by weight (500 to 999 g, 1000 to 1499 g, 1500 to 1699 g). Study duration ‐ 18 months, August 1998 to February 2000
Participants	Single centre, Neonatal Intensive Care Unit, University Hospital, Brazil Inclusion criteria: gestational age at birth 32 to 34 weeks (experimental: mean 32.7 weeks, SD 1.8, range not reported; control: mean 32.5 weeks, SD 2, range not reported) and birth weight < 1700g (experimental: mean 1276 g, SD 283 g; control: mean 1262 g, SD 270 g), mothers wished to breast feed, clinically stable, not initially on parenteral nutrition Sample size: 83 randomised (46 experimental/cup, 37 control/bottle); 78 included in analysis (44 experimental/cup, 34 control/bottle)
Interventions	Infants in both groups fed by orogastric tube until 1600 g. Experimental: supplements or complements given by cup according to the recommendations of Kuehl 1997 and Lang 1994a. Dummy not offered. Control: supplements or complements given by bottle
Outcomes	Breast feeding prevalence on discharge, at first follow‐up visit and at 3 months post discharge. Weight gain (calculated as the difference between weight at the beginning of the intervention and weight at the end of 1 week during feeding observation, reported in g/kg/d). Length of feeding time (1 week after beginning oral feeds). Oxygen saturation Breast feeding was defined as an infant exclusively or partially breast fed directly at the breast.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: ". controlled experimental study with stratified randomisation"; "Within each stratum, the infants were randomly assigned to 1 of 2 feeding groups by drawing lots".
Allocation concealment (selection bias)	Unclear risk	Quote: "Infants were randomly assigned to 1 of 2 feeding groups by drawing lots". Comment: Mechanism for drawing of lots not reported, therefore unclear whether allocation was concealed
Blinding (performance bias and detection bias) All outcomes	High risk	Blinding of intervention not possible. Blinding of outcome assessment not reported
Incomplete outcome data (attrition bias) On discharge home	Low risk	Missing outcome data (n = 5, 6%) (experimental 2, control 3): Control 3: gastro‐oesophageal reflux, bronchopulmonary dysplasia, maternal cocaine use Experimental 2: protocol violation, bronchopulmonary dysplasia Comment: low risk of bias due to incomplete outcome data. Small difference in proportions of missing data across groups, although protocol violations only in experimental group (4% experimental, 8% control). Overall small proportion of missing data (6%)
Incomplete outcome data (attrition bias) 3 months post discharge	Low risk	Missing outcome data (n = 5, 6%) (experimental 2, control 3): Control 3: gastro‐oesophageal reflux, bronchopulmonary dysplasia, maternal cocaine use Experimental 2: protocol violation, bronchopulmonary dysplasia Comment: low risk of bias due to incomplete outcome data. Small difference in proportions of missing data across groups, although protocol violations only in experimental group (4% experimental, 8% control). Overall small proportion of missing data (6%)
Selective reporting (reporting bias)	Unclear risk	Before clinical trial registration requirements, all expected outcomes reported
Other bias	Low risk