de Almeida 2015.

Study characteristics
Methods	Design: RCT, parallel 2‐arm, single‐site, superiority trial Setting: ICU, tertiary oncology university teaching hospital, São Paulo, Brazil Recruitment: January 2012 to July 2012 Maximum follow‐up: 30 days
Participants	Adult participants who underwent a major surgical procedure for abdominal cancer and required postoperative care in the ICU Liberal: n = 97; mean age (SD) = 64 (14) years Restrictive: n = 101; mean age (SD) = 64 (12) years
Interventions	While in the ICU: Liberal group: received transfusion when Hb was < 9 g/dL Restrictive group: received transfusion when Hb was < 7 g/dL
Outcomes	Primary outcome: a composite of all‐cause mortality or severe clinical complications within 30 days. Severe clinical complications included major cardiovascular complications, septic shock, acute kidney injury requiring renal replacement therapy, ARDS, and reoperation Secondary measures: length of stay in ICU, length of stay in hospital, health‐related QOL, hospital costs
Notes	Trial name: Transfusion Requirements in Surgical Oncologic Patient (TRISOP) Trial registration: (prospective) NCT01502215 Trial funding: "Support was provided solely from institutional and/or departmental sources" (de Almeida 2015 p 37) COI statement by investigators: "the authors declare no competing interests" (de Almeida 2015 p 37)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"... patients were randomized in a 1:1 ratio to one of two erythrocyte strategies ... After consent, the medical staff contacted the study randomization center to register the patient and to be told which group the patient was allocated to ..." (de Almeida 2015 p 30)   "Allocation numbers were derived from a random number table prepared by the chief statistician and were placed in opaque envelopes and opened sequentially to determine the treatment group of the patient" (de Almeida 2015 p 30)
Allocation concealment (selection bias)	Low risk	"To avoid loss of concealment, the group to which the patient was allocated could only be accessed after registration in the study randomization center. Allocation numbers were derived from a random number table prepared by the chief statistician and were placed in opaque envelopes and opened sequentially to determine the treatment group of the patient" (de Almeida 2015 p 30)
Blinding of participants and personnel (performance bias) Objective outcomes	Low risk	Blinding of personnel for this intervention was 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'
Blinding of outcome assessment (detection bias) Objective measures	Low risk	Blinding of mortality (the primary outcome used within this review) was not relevant, and we graded risk of bias as 'low' "The participants and the study investigators who classified outcomes and those who conducted the follow‐up telephone assessments were blinded to the study‐group assignments and had no access to transfusion data" (de Almeida 2015 p 30)
Blinding of outcome assessment (detection bias) Subjective measures	Low risk	"The participants and the study investigators who classified outcomes and those who conducted the follow‐up telephone assessments [including function] were blinded to the study‐group assignments and had no access to transfusion data" (de Almeida 2015 p 30)
Incomplete outcome data (attrition bias) All outcomes	Low risk	There were no withdrawals or exclusions. There were 20 protocol deviations (13 in the liberal group, 7 in the restrictive group). Analysis was done by intention to treat. "If the treating clinicians transfused an additional erythrocyte unit outside the protocol, it was recorded as a protocol deviation. After ICU discharge, the decision to transfuse was left to the discretion of the physician in charge of the patient clinical care. During the 30‐day follow‐up period, if a patient returned to the ICU for any reason, the allocated transfusion strategy was maintained. An intention‐to‐treat analysis was performed and considered the patients in their originally assigned groups" (de Almeida 2015 p 30)
Selective reporting (reporting bias)	Low risk	Trial registration: (prospective) NCT01502215  No reporting bias was apparent
Other bias	Low risk	No other biases identified