Premature neonates commonly receive red blood cell (RBC) transfusions. This study systematically identified and appraised randomized controlled trials (RCTs) where the intervention was 'transfusion of red blood cells' from searches of multiple databases. Primary review outcomes were mortality, neurodevelopmental and respiratory endpoints. Two reviewers extracted data and assigned overall quality. Twenty-seven RCTs were identified and grouped into four predefined categories: trials comparing RBC transfusion versus no transfusion/placebo (n = 3); different thresholds for transfusion (n = 6); differing doses or administration schedule (n = 4), or different types or products of RBC (n = 14). In the threshold group of trials, enrolling 679 neonates, no significant differences in mortality (relative risk 1·22, 95% confidence interval 0·84-1·75) or chronic lung disease were found. Only two trials assessed neurodevelopment outcomes, both within the threshold group, but with differing results. The largest subgroup of RCTs by number evaluated different media for storage of red cells (n = 7), enrolling 221 neonates. The methodological quality of many RCTs was poor. The design of future RCTs can be informed by the lessons from this review. Many trials failed to report on outcomes that would be considered of primary importance to clinicians. Consistent reporting of adverse events is required, and endpoints need to include neurodevelopmental outcomes.
© 2012 Blackwell Publishing Ltd.