Kola 2020.

Study characteristics
Methods	Design: RCT, single‐centre, prospective, open‐label, parallel‐arm, non‐inferiority trial Setting: tertiary care, university teaching hospital (Pondicherry, India) Recruitment: June 2015 to May 2017 Maximum follow‐up: 45 days
Participants	Patients 18 years of age or older who presented to the emergency surgical unit with a diagnosis of upper GI bleeding   Liberal group: n = 112; M/F not reported; mean (SD) age = 49.8 (14.9) Restrictive group: n =112; M/F not reported; mean (SD) age = 47.9 (14.8)
Interventions	Liberal group: transfusion if Hb was < 8 g/dL with target Hb > 10 g/dL Restrictive group: transfusion if Hb was < 7 g/dL with target Hb > 9 g/dL
Outcomes	Primary outcome: mortality. The time period was not designated, but deaths were reported as in‐hospital, during follow‐up, and overall mortality within follow‐up of 45 days   Secondary outcomes: number of days from admission to death, rebleeding episodes (in‐hospital bleeding and rebleeding during 45‐day follow‐up), Hb value before death, number of sessions of endoscopic treatment, requirement for banding/sclerosant treatment, requirement of SB tube placement and duration, incidence of transfusion reaction (major/minor), dose, duration of octreotide infusion, length of hospital stay
Notes	Trial registration: CTRI/2017/09/009682, but not prospective. Trial appears to have been registered only in May 2017, when, it is reported, recruitment had closed. Maximum follow‐up was 45 days. Study was submitted for publication in April 2020 and was accepted in June Trial funding: "nil"(Kola 2020 p 19) COI statement by investigators: "there are no conflicts of interest" (Kola 2020 p 19)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Stratified permuted block randomization was done using a computer program with randomly selected unequal block sizes of 4 and 6. Stratification was done based on variceal bleeding vs. non‑variceal bleeding" (Kola 2020 p 14)
Allocation concealment (selection bias)	Low risk	"Allocation concealment was performed using a serially numbered opaque sealed envelope (SNOSE) technique. The envelopes were opened by the residents on duty and allocation was carried out at the time of admission of the patient" (Kola 2020 p 14)
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'
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'. Primary outcome is mortality and risk of bias is low
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	All outcomes reported. No loss to follow‐up
Selective reporting (reporting bias)	Unclear risk	Trial first recruited in 2015; finished in May 2017; registration also in May 2017
Other bias	Low risk	Investigators reported some baseline imbalances (e.g. cirrhosis), but they were not statistically significant