Low versus high haemoglobin concentration threshold for blood transfusion for preventing morbidity and mortality in...

Infants of very low birthweight frequently receive red blood cell transfusions during their primary hospital stay. Generally, this is guided by predetermined haemoglobin or haematocrit thresholds according to a protocol or as prompted by clinical situations, including critical illness or surgery. Recommendations advocate maintaining higher thresholds in the early weeks when the risk of major morbidity is highest, while permitting lower thresholds after this time. In truth, clinicians worry about the potential effect of chronic anaemia on neurodevelopmental outcomes, as well as the risk of transfusion-related complications in the immature host. Clinical trials have reflected this practice by comparing haemoglobin levels adjusted for critical illness, comparing transfusion algorithms that use fixed differences between haemoglobin thresholds, with both thresholds progressively lowered across postnatal age. This is an update of a review first published in 2011.

To evaluate the effect of lower (restrictive) compared with higher (liberal) haemoglobin thresholds for transfusion, with or without adjustment for age and critical illness with either fixed or variable transfusion volume, on mortality or later neurodevelopmental outcomes assessed in later infancy at approximately two years postmenstrual age, or the number of transfusions in very low birthweight infants.

Searches were conducted in January 2024 in CENTRAL, MEDLINE, Embase, CINAHL, Epistemonikos, and trial registries. We searched the reference lists of related systematic reviews and trials.

We selected randomised controlled trials (RCTs) of lower or restrictive haemoglobin/haematocrit thresholds compared with liberal or higher haemoglobin/haematocrit thresholds for transfusion in low birthweight infants within three days of birth.

We used standard Cochrane methods. Our main outcomes were a combined outcome of death or neurodevelopmental impairment, all-cause mortality, and the number of transfusions per infant. We expressed our results using mean difference (MD), standardised mean difference (SMD), risk ratio (RR), and risk difference (RD) with 95% confidence intervals (CIs). We used GRADE to assess the certainty of evidence.

Six trials, enrolling 3451 infants, compared transfusion strategies utilising a lower (restrictive) haemoglobin threshold compared to a higher (liberal) haemoglobin threshold. The transfusion thresholds used in these trials reflected prevailing clinical practice at the time of study design. For comparative purposes, they have been labelled as 'restrictive' and 'liberal'. The trials were similar in design, although each used slightly different transfusion algorithms and intervention thresholds. The three larger trials also conducted later neurosensory assessments. The number of infants included in outcome calculations varies across in-hospital versus post-discharge outcomes, assessment methods, and exclusion criteria, and our analysis uses the denominators reported in the original publications. Overall, utilising a lower compared to a higher haemoglobin transfusion threshold results in little or no difference in the combined outcome of death or neurodevelopmental impairment at 18 to 26 months postmenstrual age (RR 1.02, 95% CI 0.95 to 1.09; I² = 55%; RD 0.01, 95% CI -0.03 to 0.04; 3 studies, 3041 infants; high-certainty evidence). Mortality at 18 to 26 months was also not different (RR 0.99, 95% CI 0.83 to 1.17; I² = 0%; RD -0.00, 95% CI -0.03 to 0.02; I² = 0%; 3 studies, 3186 infants; high-certainty evidence). Infants allocated to the restrictive threshold may receive fewer transfusions during the primary hospital stay (mean difference in transfusion number per infant -1.05, 95% CI -1.26 to -0.84; I² = 84%; 6 studies, 3451 infants; low-certainty evidence).

The trials comparing lower or restrictive versus higher or liberal haemoglobin thresholds for transfusion show little to no difference in important outcomes at hospital discharge and at the time of later neurodevelopmental follow-up. The use of restrictive, as compared to liberal, haemoglobin or haematocrit transfusion thresholds in these trials in infants of very low birthweight results in modest reductions in transfusion exposure and haemoglobin levels. The safety of haemoglobin levels below these lower limits has not been evaluated and should only be considered in the context of randomised controlled trials.