Villanueva 2013.

Study characteristics
Methods	Design: RCT, parallel 2‐arm, single‐site, phase 4 trial Setting: large general hospital, Barcelona, Spain Recruitment: June 2003 to December 2009 Maximum follow‐up: 45 days
Participants	Participants > 18 years of age who had haematemesis or melena, or both (due to upper GI bleeding) 921 randomised Liberal group: n = 460; mean age (SD) = 64 (16) years. After 15 withdrawals, 445 were included in analysis Restrictive group: n = 461; mean age (SD) = 66 (15) years. After 17 withdrawals, 444 were included in analysis
Interventions	Liberal group: transfused when Hb < 9 g/dL Restrictive group: transfused when Hb as < 7 g/dL In both groups, 1 unit of RBCs was transfused initially
Outcomes	Primary outcome: death at 45 days Secondary outcomes included: rate of further bleeding, rate of in‐hospital complications
Notes	Trial registration (December 2006 to not prospective): NCT00414713 Trial funding: Fundació Investigació Sant Pau COI statement by investigators: "Dr. Guarner reports receiving consulting fees from Sequana Medical. No other potential conflict of interest relevant to this article was reported" (Villanueva 2013 p 20)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random sequence generation was computer generated
Allocation concealment (selection bias)	Low risk	Trial used sealed consecutively numbered, opaque envelopes
Blinding of participants and personnel (performance bias) Objective outcomes	Low risk	Blinding of personnel for this intervention is not feasible, but in our view, for objective outcomes such as mortality (the primary outcome used within this review), we graded risk of bias as 'low' Clinicians and participants were not blinded
Blinding of outcome assessment (detection bias) Objective measures	Low risk	Blinding of mortality (the primary outcome used within this review) is not relevant, and we graded risk of bias as 'low' Mortality was the primary trial outcome. Assessors of other outcomes were not documented to be blinded
Blinding of outcome assessment (detection bias) Subjective measures	Low risk	No data from subjective outcomes (e.g. function)
Incomplete outcome data (attrition bias) All outcomes	Low risk	There were 15 withdrawals from the liberal arm (of 460 participants) and 17 withdrawals from the restrictive arm (of 461 participants)
Selective reporting (reporting bias)	Unclear risk	Trial was registered post hoc (3 years after recruitment began). The primary outcome changed from death at 30 days (December 2006) to death at 45 days (November 2007) (all data were obtained from primary author of trial report by JC)
Other bias	Low risk	No other biases were apparent