Primary outcomes

1. Mortality: before discharge from initial hospital or before a defined period of follow‐up (28 days, 12 months, or 18 months postnatal age, or a combination).

2. ROP, grade 3 or more prior to discharge home (ICCRP 2005).

3. Severe adverse findings at ultrasound (grades 3 to 4 IVH (Papile 1983), hydrocephalus, cortical atrophy, or periventricular leukomalacia) during first hospitalisation (Pinto‐Martin 1995).

4. CLD requiring additional oxygen at 36 weeks' postmenstrual age or prior to discharge home (Shennan 1988).

5. NEC, stage 2 or greater (Bell 1978).

6. Cerebral palsy by physician assessment.

7. Developmental delay (developmental quotient more than two standard deviations below the mean on validated assessment tool of cognitive function (e.g. Bayley Score of Infant Development).

8. Blindness (visual acuity less than 20/200 in best eye).

9. Deafness (hearing loss requiring amplification or cochlear implantation).

Secondary outcomes

1. Composite outcome of death or severe adverse outcomes:

   1. mortality or severe morbidity (or its complement, survival without severe morbidity) at initial hospital discharge, where significant morbidity is defined as:

      1. ROP, grade 3 or more prior to discharge home (ICCRP 2005);

      2. severe adverse findings at ultrasound (grades 3 to 4 IVH (Papile 1983), hydrocephalus, cortical atrophy, or periventricular leukomalacia) during first hospitalisation (Pinto‐Martin 1995);

      3. CLD requiring additional oxygen at 36 weeks' postmenstrual age or prior to discharge home (Shennan 1988); or

      4. NEC, stage 2 or greater (Bell 1978).

2. Composite outcome of mortality or severe adverse neurosensory outcome (or its complement, survival without serious adverse neurosensory outcome) at a defined period of follow‐up at age 18 to 24 months' adjusted gestational age or older, where adverse neurosensory outcome is defined as:

   1. cerebral palsy by physician assessment;

   2. developmental quotient (more than two standard deviations below the mean on validated assessment tool of cognitive function (e.g. Bayley Score of Infant Development);

   3. blindness (visual acuity less than 20/200 in best eye); or

   4. deafness (hearing loss requiring amplification or cochlear implantation).

3. Other outcomes:

   1. late‐onset sepsis (sepsis diagnosed more than 72 hours after birth)

   2. length of mechanical ventilation (days)

   3. donor exposure

   4. numbers of RBC transfusions

   5. length of initial neonatal intensive care unit stay (days)

   6. markers of inflammation or oxidative stress (if available), or both, including tumour necrosis factor (TNF)‐α, monocyte chemoattractant protein‐1, interleukin (IL)‐1, IL‐6, IL‐8, and total oxidant load

4. Transfusion reactions as defined by the Serious Hazards of Transfusion (SHOT) scheme (Stainsby 2008):

   1. acute transfusion reaction;

   2. delayed transfusion reaction;

   3. transfusion‐related acute lung injury;

   4. transfusion‐associated graft‐versus‐host disease;

   5. post‐transfusion purpura; or

   6. transfusion‐transmitted infection.

Selection bias (random sequence generation and allocation concealment)

Performance bias

For each included study, we categorised the methods used to blind study personnel from the knowledge of which intervention a participant received as:

• low risk ‐ adequate for personnel (a placebo that could not be distinguished from the active drug was used in the control group);

• high risk ‐ inadequate (personnel aware of group assignment);

• unclear risk ‐ no or unclear information provided.

Detection bias (blinding)

For each included study, we categorised the methods used to blind outcome assessors from knowledge of which intervention a participant received. We categorised the methods used for detection bias as:

• low risk ‐ adequate (follow‐up was performed with assessors blinded to group assignment);

• high risk ‐ inadequate (assessors at follow‐up were aware of group assignment);

• unclear risk ‐ no or unclear information provided.

Attrition bias (outcome data)

For each included study and for each outcome, we described the completeness of data including attrition and exclusions from the analysis. We noted whether attrition and exclusions were reported, the numbers included in the analysis at each stage (compared with the total randomised participants), reasons for attrition or exclusion where reported, and whether missing data were balanced across groups or were related to outcomes. Where sufficient information was reported or supplied by the trial authors, we included the missing data in the analyses. We categorised the methods with respect to the risk attrition bias as:

• low risk ‐ adequate;

• high risk ‐ inadequate;

• unclear risk ‐ no or unclear information provided.

Reporting bias (selective outcome reporting)

For each included study, we described how we investigated the risk of selective outcome reporting bias and what we found. We assessed the methods as:

• low risk ‐ adequate (where it was clear that all of the study's prespecified outcomes and all expected outcomes of interest to the review were reported);

• high risk ‐ inadequate (where not all of the study's prespecified outcomes were reported; one or more reported primary outcomes were not prespecified; outcomes of interest were reported incompletely and so could not be used; study did not include results of a key outcome that would have been expected to have been reported);

• unclear risk ‐ no or unclear information provided (the study protocol was not available).

Other bias

For each included study, we described any important concerns that we had about other possible sources of bias. We assessed whether each study was free of other issues that could put it at risk of bias as:

• low risk ‐ no concerns of other bias raised;

• high risk ‐ concerns raised after multiple looks at the data with the results made known to the investigators; difference in number of participants enrolled in abstract and final publications of the paper;

• unclear ‐ concerns raised about potential sources of bias that could not be verified by contacting the authors.

Two review authors independently made judgements about risk of bias. We resolved discrepancies through consensus.

Dichotomous data

For dichotomous data, we presented results as risk ratio, risk difference, and mean difference where appropriate using Review Manager 5 software (RevMan 2014). We calculated 95% confidence intervals.

Continuous data

For continuous data, we used the mean difference if outcomes were measured in the same way between trials. We used the standardised mean difference to combine trials that measured the same outcome but used different methods.