Lacroix 2007.

Study characteristics
Methods	Design: RCT, parallel 2‐arm, multicentre trial Setting: 19 tertiary‐care paediatric ICUs in 4 countries (Canada, Belgium, USA, UK)  Recruitment: November 2001 to August 2005 Maximum follow‐up: 30 days
Participants	648 (637 following withdrawals) stable critically ill children with Hb < 9.5 g/dL within 7 days after admission to an ICU were randomly allocated to 1 of 2 groups: Liberal group: n = 317; M/F = 191/126; mean (SD) age = 39.6 (51.9) months Restrictive group: n = 320; M/F = 190/130; mean (SD) age = 35.8 (46.2) months
Interventions	Liberal group: transfused when Hb fell to < 9.5 g/dL, with target range of 11.0 g/dL to 12.0 g/dL Restrictive group: transfused if Hb fell to < 7.0 g/dL, with target range of 8.5 g/dL to 9.5 g/dL
Outcomes	28‐day mortality, sepsis, transfusion reactions, infection, length of stay
Notes	Trial name: Transfusion Requirements in the Pediatric Intensive Care Unit (TRIPICU) study Trial registration (not prospective): ISRCTN37246456 Trial funding: supported by grants (84300 and 130770) from the Canadian Institutes of Health Research and by grants (3348 and 3568) from the Fonds de la Recherche en Santé du Québec COI statement by investigators: "Drs. Lacroix and Hébert report receiving consulting fees and grant support from Johnson & Johnson; Dr. Hébert also reports receiving consulting fees and unrestricted funds from Novo Nordisk and Amgen serving as a Career Scientist of the Ontario Ministry of Health (1994–2004), and receiving unrestricted training funds from Canadian Blood Services; Dr. Hume reports being employed by the Canadian Blood Services; and Dr. Peters reports receiving consulting fees from Baxter, Xoma, and Eli Lilly. No other potential conflict of interest relevant to this article was reported" (Lacroix 2007 p 1609)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Randomization was centralized, with assignment data posted on the Internet. Patients were assigned to the study groups in blocks of 2 or 4 that were randomly distributed and stratified according to center and three age groups (≤28 days, 29 to 364 days, and >364 days)" (Lacroix 2007 p 1610)
Allocation concealment (selection bias)	Low risk	Allocation was Internet based and central
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' "Clinical staff and parents of the participants were aware of the assignments to study groups, but physicians, nurses, and research staff were unaware of the block‐randomization strategy" (Lacroix 2007 p 1610)
Blinding of outcome assessment (detection bias) Objective measures	Low risk	Mortality was the primary outcome. Blinding of mortality (the primary outcome used within this review) is not relevant, and we graded risk of bias as '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	Full data for 626/648; withdrawals were more common in the restrictive than the liberal arm (7 vs 4); protocol violations were more common in the liberal than the restrictive arm (10 vs 1). Both ITT and per‐protocol analyses were conducted
Selective reporting (reporting bias)	Unclear risk	No reporting bias was apparent; however, trial was retrospectively registered (June 2004; when recruitment began in November 2001). ISRCTN37246456
Other bias	Low risk	No other biases were apparent