SIT 2014.

Study characteristics
Methods	Multicentre randomised trial in 29 clinical centres in the USA, Canada, France and the UK. Recruitment: December 2004 to May 2010. The last participant enrolled completed the exit visit on July 29, 2013.
Participants	Inclusion criteria: children aged 5 to 15 years, confirmed diagnosis of haemoglobin SS or haemoglobin Sβ0 thalassaemia, and at least one infarct‐like lesion on the screening MRI scan defined as an MRI signal abnormality that was at least 3 mm in one dimension and that was visible in two planes on fluid‐attenuated inversion recovery (FLAIR) T2‐weighted images, as determined by agreement of two of the three trial neuroradiologists. Exclusion criteria: history of focal neurologic deficit associated with an infarct on brain MRI, a seizure disorder, treatment with hydroxyurea in the previous 3 months, a history of regular transfusion therapy, or imaging or non‐imaging TCD measurement that was above the trial‐defined thresholds. Participant flow: 1210 registered for screening; 1074 had screening MRI evaluated by neuroradiology committee; 675 had normal, 20 indeterminate MRI; 379 had infarct‐like lesions on screening MRI; 291 had infarct‐like lesions adjudicated by neurology committee; 220 had pre‐randomisation MRIs adjudicated by neuroradiology committee; 196 underwent randomisation. Transfusion arm: N = 99 (15 crossed over to observation) Sex: male: 59 (60%); F: 40 (40%) Age: 5 to 7: 26 (26%): 8 to 10: 35 (35%); 11 to 13: 32 (32%); 14 to 15: 6 (6%) TCD velocity: median (IQR) cm/sec: 147 (123‐168) (N = 98) Lesions on initial MRI: 99 (100%) Parental report of recurring headaches: yes: 37 (37%); No: 62 (63%) Steady state haemoglobin: median (IQR): g/L: 77 (72 to 84) Phenotypes: not stated (Included only HbSS or HbSβº) Hb F% median (IQR): 9.0 (4.0 to 14.0) Alpha thalassaemia: not reported Observation arm: N = 97 (6 crossed over to transfusion) Sex: male: 52 (54%); F: 45 (46%) Age: 5 to 7: 28 (29%): 8 to 10: 32 (33%); 11 to 13: 29 (30%); 14 to 15: 8 (8%) TCD velocity: median (IQR) cm/sec: 143 (131 to 163) Lesions on initial MRI: 97 (100%) Parental report of recurring headaches: yes: 43 (44%); No: 54 (56%) Steady state haemoglobin: median (IQR): g/L: 79 (74 to 89) Phenotypes: not stated trial included only HbSS or HbSβº Hb F% median (IQR): 10.0 (5.0 to 15.0) Alpha thalassaemia: not reported
Interventions	Transfusion: transfusion arm received a transfusion approximately monthly to maintain a target haemoglobin concentration greater than 90 g/L and a target haemoglobin S concentration of 30% or less. Red cell component: leucocyte‐depleted, negative for haemoglobin S. Red cell matching: ABO, Rh and Kell antigens. Iron chelation: ferritin levels were monitored before each transfusion. Site investigators were advised to initiate chelation therapy for participants who had ferritin levels greater than 1500 ng per millilitre for 2 or more consecutive months. Observation: observation arm received standard care with no treatment for silent infarcts and no hydroxyurea therapy and were evaluated quarterly
Outcomes	Primary outcome: the recurrence of infarct or haemorrhage as determined by neuroimaging, clinical evidence of permanent neurologic injury, or both. A new infarct had to meet the criteria for a SCI; an enlarged SCI was defined as a previously identified silent cerebral infarct that increased by at least 3 mm along any linear dimension in any plane on MRI; TIA, included in secondary analyses of neurologic outcomes, defined as an event that resulted in focal neurologic deficits that lasted less than 24 hours, did not result in abnormalities on T2‐weighted or FLAIR images that were indicative of an acute infarct, and had no other reasonable medical explanation. Secondary outcomes: changes in cognition, assessed by measurement of IQ scores with the Wechsler Abbreviated Scale of Intelligence12 or the Wechsler Preschool and Primary Scale of Intelligence III; also assessed scores on the Behavior Rating Inventory of Executive Function (BRIEF).
Notes	Funding: supported by grants from the National Institute of Neurological Disorders and Stroke (5U01NS042804, 3U01NS042804 [American Recovery Reinvestment ACT supplementary grant] to Dr DeBaun); the Institute of Clinical and Translational Sciences, National Center for Research Resources, and the National Center for Advancing Translational Sciences, Clinical and Translational Research; NIH Roadmap for Medical Research (UL1TR000448, to Washington University; UL1TR001079, to Johns Hopkins University; and UL1TR000003, to the Children’s Hospital of Philadelphia); and Research and Development in the National Health Service, UK. Declarations of interest: Dr McKinstry reports receiving honoraria and lecture fees from Siemens Healthcare and consulting fees from Guerbet; Dr Woods, receiving fees for serving on a data and safety monitoring board from Mast Therapeutics and grant support from ClinDatrix and Novartis; Dr Kwiatkowski, receiving fees for serving on an advisory board from Shire Pharmaceuticals, consulting fees from Shire Pharmaceuticals and Sideris Pharmaceuticals, and grant support from Resonance Health; Dr. Heiny, receiving lecture fees from Novartis; Dr. Redding‐Lallinger, receiving grant support from Eli Lilly and Mast Therapeutics; and Dr. Casella, receiving honoraria, travel support, and consulting fees through his institution from Mast Therapeutics and being an inventor and a named party on a patent and licensing agreement for an assay panel of brain biomarkers for the detection of brain injury (PCT US2011/056338), licensed to ImmunArray with pending royalties only. No other potential conflict of interest relevant to this article was reported. Trial registration: NCT00072761 & ISRCTN52713285. Mean length of follow‐up: children were followed for a median of 3 years. Power calculation: a sample size of 204 participants (102 in each group) would give the trial 85% power to detect a decrease of at least 86% in the prevalence of the primary end point, assuming a 10% dropout rate and a crossover rate of 16% from transfusion to observation and 3% from observation to transfusion, at a two‐tailed nominal alpha level of 0.05.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation assignments were provided by the statistical data coordinating centre with the use of a permuted block design, with stratification according to site, age, and sex. Participants were assigned in a 1:1 ratio to the observation group or the transfusion group and were followed until the occurrence of a trial end‐point event or until exit from the trial.
Allocation concealment (selection bias)	Low risk	Assignments were provided by the statistical data coordinating centre.
Blinding of participants and personnel (performance bias) All outcomes	High risk	By the nature of the trial treatments used (blood transfusions vs observation), it is impractical to make SIT trial blinded (masked).
Blinding of outcome assessment (detection bias) Stroke and TIA	Low risk	Members of neuroradiology and neurology committees, who were unaware of the trial‐group assignments, adjudicated neurologic and MRI findings.
Blinding of outcome assessment (detection bias) All outcomes apart from stroke or all‐cause mortality	High risk	Unblinded trial.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All outcomes reported. All participants accounted for, conducted an intention‐to‐treat and per protocol analysis for the primary outcomes The primary end point was ascertained for 185 of the 196 participants (94%). Of the 99 participants randomly assigned to the transfusion group, 90 started receiving transfusions within 4 weeks after assignment. The cross‐over rate from transfusion to observation was 15% (15 of 99 participants); 9 participants declined blood transfusion, and 6 crossed over to observation at a median time of 34 days.
Selective reporting (reporting bias)	Low risk	Protocol available and all planned outcomes reported.
Other bias	Unclear risk	Among participants in the observation group, 32% received transfusions (a median of 3 transfusions each), including 6 participants who crossed over to regular monthly transfusions at a median of 1.7 years. During the course of the trial, hydroxyurea was started in 14 of 97 participants (14%) in the observation group and in 3 of 99 (3%) in the transfusion group because of disease severity. Exclusion criteria included treatment with hydroxyurea. Not clear how long or when treatment began ‐ possible contamination and unknown effect on outcomes. 6 also crossed over to regular transfusion giving 20% cross‐over rate to either hydroxyurea or transfusion.