One of the major problems with deciding when to transfuse is the lack of a clearly identified and agreed-upon definition of severe anemia in neonates. Owing to the absence of a universally accepted definition of anemia in neonates necessitating an RBC transfusion and to the wide age-specific reference range of acceptable hematocrit values, treatment thresholds are not standardized. In research settings, various strategies have been employed. Two randomized clinical trials comparing transfusion practices in preterm infants and other prospective studies have used the cutoff hemoglobin level of 8 g/dL or less for defining severe anemia in neonates ^{10, 11} . The definition can also be dependent on the outcomes being assessed. For example, preoperative hematocrit has been examined as a continuous variable and a level lower than 40% was found to be the ideal cutoff point (Youden J index) for predicting overall mortality ¹² .

Although there is more convincing randomized controlled trial (RCT) data from adults supporting the use of restrictive transfusion strategies as compared with liberal transfusion strategies, the evidence in neonates is still highly conflicted. In 2005, an RCT by Bell et al., studying 100 hospitalized preterm infants with birth weights of 500 to 1,300 g stratified to two levels of hematocrit threshold for RBC transfusion, suggested that, although infants in the liberal transfusion group overall received more RBC transfusions (mean ± standard deviation 5.2 ± 4.5 versus 3.3 ± 2.9), in the restrictive transfusion group, the number of donors to whom the infants were exposed was not significantly different. They additionally reported that there was no major difference between restrictive versus liberal groups in the percentage of infants who “did not get transfusions at all” (12% in the liberal group versus 10% in the restrictive group). Notably, infants in the restrictive group were more likely to have major adverse neurologic events, including (a) intraparenchymal brain hemorrhage, (b) periventricular leukomalacia, and (c) more frequent episodes of apnea (both mild and severe episodes), than the liberal transfusion group. This study raised the possibility that the practice of restrictive transfusions may indeed be harmful to preterm infants ¹⁰ .

The study by Bell et al. was followed by the Premature Infants in Need of Transfusion (PINT) study with contrasting results. This RCT included 451 infants with a primary goal to determine whether transfusion of ELBW infants at restrictive versus liberal thresholds had different rates of survival or morbidity at the time of hospital discharge. With a mean birth weight of 770 g and gestational age of 26 weeks, rates of the primary outcome (death before home discharge) were 74.0% in the low-threshold group and 69.7% in the high-threshold group ( p = 0.25, risk difference 2.7%, 95% confidence interval [CI] −3.7% to 9.2%). The study revealed that when a higher hemoglobin threshold in ELBW infants was maintained, a greater number of infants received transfusions, with little evidence of benefit ¹¹ . It is important to note that the trial by Bell et al. and the PINT trial were both limited by not studying neurodevelopment beyond hospital discharge. Thus, the interpretation from these studies is restricted to the duration of hospitalization only and not the long-term effects. The shorter-term outcomes studied to date may be a poor proxy for the long-term neurological outcomes. There is limited knowledge on the effects of limiting oxygen delivery to the brain by withholding transfusions, and this remains an important gap in our understanding on the effects of RBC transfusions on neurodevelopment.

Two ongoing and much-awaited definitive phase III transfusion studies in neonates are the Effects of Transfusion Thresholds on Neurocognitive Outcome of Extremely Low Birth Weight Infants (ETTNO) trial and the Transfusion of Prematures (TOP) trial to determine whether higher hemoglobin thresholds for transfusing ELBW infants resulting in higher hemoglobin levels lead to improvement in (1) survival (primary outcome) and (2) rates of neurodevelopmental impairment at 22–26 months of age.

Age of transfused blood and outcomes in children. In children and neonates, as in adults, the age of stored RBCs and adverse outcomes has been a highly debated question. The Age of Red Blood Cells in Premature Infants (ARIPI) RCT showed no difference in outcomes in premature very-low-birth-weight (VLBW) infants requiring a transfusion when fresh RBCs (mean age of 5.1 days) versus standard blood bank practices (mean age of 14.6 days) were followed ¹³ . However, there were concerns raised about the generalizability of the ARIPI study. The primary critique was that the mean duration of RBC storage in the standard-of-care group in the ARIPI study was not reflective of the mean storage duration in the US, in which the average storage age was reported to be higher (about 18 days). Furthermore, the hemoglobin thresholds for transfusion were not pre-specified or standardized, allowing for higher hemoglobin levels for each infant, and the storage solution and manufacturing of units in different anticoagulant-preservative solutions were not addressed in the study ¹⁴ . Currently, the Age of Blood in Children in Pediatric Intensive Care Units (ABC PICU) study, examining outcomes comparing transfused RBCs stored for not more than 7 days or standard-issue RBCs (expected mean RBC storage duration of 17–21 days), is awaiting completion. Evidence in older children from the Tissue Oxygenation by Transfusion in Severe Anemia With Lactic Acidosis (TOTAL) randomized clinical trial (290 children 6–60 months old) showed no difference in post-transfusion correction of mean lactate levels between young blood (median age of 8 days) versus older blood (median age of 32 days) ¹⁵ .

Adverse outcomes from red blood cell transfusions

RBC transfusions are based on the assumption that the transfusion leads to a direct and immediately effective increase in oxygen delivery to tissues. However, although the correlation with transfused red cell mass and improved oxygenation in neonates has not been proven, RBC transfusions do come with inherent risks for adverse complications in the recipient. Although many of these inherent risks are similar to ones in older children and adults, neonates have some specific adverse associations and outcomes potentially related to RBC transfusions, including the development of necrotizing enterocolitis (NEC), intraventricular hemorrhage (IVH) ^{16– 18} , retinopathy of prematurity (ROP) ¹⁹ , and chronic lung disease (CLD) ^{20, 21} as well as mortality. Interpretation of UK analysis of the Serious Hazards of Transfusion (SHOT) scheme data estimated the rates of an adverse outcome to be 18:100,000 red cells issued for children younger than 18 years and 37:100,000 for infants younger than 12 months compared with 13:100,000 for adults ⁷ . Thus, infants and neonates are disproportionately more predisposed to adverse effects from RBC transfusions than are other populations.

Meta-analyses/systematic reviews studying these relationships give contrasting results as more and more studies addressing the topics surface. As an example, in 2012, a meta-analysis of 11 retrospective case-control studies and one cohort study showed that recent exposure to transfusion was associated with NEC in neonates. Neonates who developed NEC were at overall higher risk of mortality ²² . However, a more updated meta-analysis published recently which addresses the same question from 17 observational studies showed no temporal relationship between RBC transfusions and NEC ²³ . This meta-analysis highlighted how the effect size for this relationship differed between matched case-control studies (odds ratio [OR] 1.20, 95% CI 0.58–2.47; p = 0.63) and differed from those reported in cohort studies (OR 0.51, 95% CI 0.34–0.75; p ≤0.01). This example and contrast in results based on the study type and methodology underscores the need for high-quality prospective evidence in the area.

The important question, however, is whether the adverse effects are from anemia itself or from the transfusions, and methodologically sound prospective studies are required to answer this fundamental question. Patel et al., in a prospective study assessing various predictors of adverse outcomes as time-varying co-variates, recently showed that among VLBW infants, severe anemia, instead of RBC transfusions, is associated with an increased risk of NEC, thus proposing the need for further studies to evaluate whether preventing severe anemia is more important than minimizing RBC transfusions ²⁴ . Likewise, Goobie et al. ¹² recently suggested an association between preoperative anemia and postoperative mortality in neonates ²⁵ .

Within the transfused group as well, the outcome could vary depending on the hemoglobin transfusion threshold. Keir and Stanworth et al. published the most comprehensive review and meta-analysis to date on the harmful effects and associations potentially attributable to RBC transfusions ²⁶ . Together inclusive of 61 studies (16 randomized trials and 45 non-randomized studies), this meta-analysis revealed no evidence of difference in rates of mortality between restrictive and liberal strategies for transfusion (eight RCTs: risk ratio 1.24, 95% CI 0.89–1.67, heterogeneity = 0%). Similarly, no differences were seen for other outcomes, including NEC, ROP, CLD, or IVH ²⁶ .

There are no consensus guidelines on how the transfusion practices require specific modification for neonatal use. The standardization of (1) definitions of adverse effects in neonates and (2) associations of RBC transfusion with many of these adverse effects in neonates, through an international consensus of experts, is essential.

Thrombocytopenia in neonates is traditionally defined as a platelet count of less than 150 × 10 ⁹/L and is classified as mild (100–150 × 10 ⁹/L), moderate (50–99 × 10 ⁹/L), or severe (<50 × 10 ⁹/L) ³³ . However, the question is whether there is a standard definition of thrombocytopenia which can be applied to neonates of all gestational ages. The incidence of thrombocytopenia in neonates is extremely variable depending on the gestational age of the studied population. Whereas the overall incidence of thrombocytopenia in newborns at the time of birth is relatively low (less than 1%) ³⁴ , among NICU patients, the incidence of thrombocytopenia has been reported to be inversely correlated to gestational age. Specifically, among neonates with a birth weight less than 1,000 g, it is estimated to be as high as more than 50% ^{35, 36} . In 2009, Wiedmeier et al., in a large observational study on neonatal platelet counts, including data on about 47,000 infants (between 22 and 42 weeks of gestational age), suggested a predictable increment in platelet counts since birth with an increase in gestational age, with a mean increase in platelet count of ~2 × 10 ⁹/L per week increase in gestational age ³⁷ . These findings suggest that there is not one single threshold to define thrombocytopenia in neonates and that different thresholds need to be used to define thrombocytopenia in preterm infants depending on gestational age ³⁷ .

A clear relationship between low platelet counts and consequential clinical bleeding or major hemorrhage has not been established in neonates ³² . Several studies have found that a multitude of factors besides the severity of thrombocytopenia predict the bleeding risk in neonates, the degree of prematurity being a predominant factor. Thus, it will be important to develop improved tests to assess primary hemostasis and bleeding risk in neonates. At present, indications, thresholds, and rationale for prophylactic platelet transfusions to correct thrombocytopenia remain controversial and in need of a strong evidence base to guide safe platelet transfusion practices in neonates, especially the most premature infants.

There continues to be widespread heterogeneity in the pretransfusion thresholds for platelet count and overall platelet transfusion practices between various hospitals and countries. In general, moderate thrombocytopenia (50–150 × 10 ⁹/L) is not supposed to be detrimental. In an earlier study of prophylactic platelet transfusion thresholds, over 150 thrombocytopenic neonates less than 33 weeks’ gestation age (weight of 500 to 1,500 g) were randomly assigned to receive platelet transfusions to maintain the platelet count at either greater than 150 × 10 ⁹/L or greater than 50 × 10 ⁹/L in the first week of life. This study suggested that liberal platelet transfusion threshold (150 × 10 ⁹/L) did not protect against IVH ³⁸ . Another retrospective observational cohort study examined infants with platelet counts of less than 50 × 10 ⁹/L in the NICU, and preterm infants with platelet counts of less than 50 × 10 ⁹/L in whom platelet transfusions were withheld did not suffer from serious hemorrhage, suggesting that lower platelet count thresholds may be safe for otherwise stable infants ³⁹ . In a multicenter, retrospective cohort study of 972 VLBW infants from six US NICUs, the pretransfusion count was at least 50 × 10 ⁹/L for 653 (65.4%) of 998 transfusions. While platelet transfusion decisions were taken with higher severity of illness, there was neither an association between the severity of thrombocytopenia and the subsequent risk for IVH nor a lower risk of IVH due to platelet transfusions (hazard ratio 0.92, 95% CI 0.49–1.73; p = 0.80) ⁴⁰ . There is a need for RCT evidence to guide the optimal platelet transfusion decisions in premature neonates.

Studies have also assessed the feasibility of transfusing platelets on the basis of platelet mass (platelet count times mean platelet volume) instead of the platelet count as a transfusion trigger. The use of platelet mass-based NICU transfusion guidelines has been shown in few studies to be feasible as well as safe, as they were associated with fewer platelet transfusions overall and no recognized increase in hemorrhagic problems ^{41, 42} .

The major clinical outcome of concern for neonates with severe thrombocytopenia is risk of major bleeding, including intracranial hemorrhage and specifically IVH and periventricular hemorrhage (PVH). Recently, a neonatal bleeding assessment tool (NeoBAT) was proposed aiming for standardized bleeding assessments in this high-risk population ⁴³ . The NeoBAT score is being applied in prospective studies and trials of neonates which are mentioned subsequently. There are published algorithms which are helpful to direct the diagnosis and specific investigations of different etiologies of thrombocytopenia. Specific indications for platelet transfusions like neonatal alloimmune thrombocytopenia (NAIT) need a high index of suspicion for early diagnosis and prompt attention with specific requirements like antigen-negative platelets.

The recent guidelines from the British Society of Hematology (BSH) recommend that for stable children younger than 4 months, prophylactic platelet transfusion should be given at a platelet count of less than 30 × 10 ⁹/L and for stable children older than 4 months, prophylactic platelet transfusions should be given at a platelet count of less than 10 × 10 ⁹/L. The guidelines admit that, given the lack of studies in pediatrics, the platelet transfusion indication for critically ill children or those with hematologic and oncologic malignancies who develop severe thrombocytopenia is based on adult study data ⁴⁴ .

Adverse outcomes from platelet transfusions. Platelet transfusions come with risks inherent to any blood component transfusion, including clinical errors in administration and incorrect blood component transfused. They also have significant risks of bacterial and septic infections due to storage at room temperature and risks of allergic reaction as well as TACO and TRALI due to the innate plasma component of platelet products.

Although the number of platelet transfusions might be a surrogate marker for level of illness and thus predictive of mortality, Baer et al., in a study of 1,600 thrombocytopenic NICU patients, suggested that multiple platelet transfusions may be contributing directly to a small fraction of mortality in neonates ⁴⁵ . These relationships remain associations at best and need to be validated in prospectively designed studies.

Data from the UK’s SHOT national hemovigilance scheme against a population-based epidemiological study of transfused patients have indicated that transfusion-associated adverse events are disproportionate in the number of occurrences in the neonatal population as compared with older children and adults. Transfusion reactions are likely highly under-recognized and under-reported in neonates, especially the most premature infants ⁷ .

A pathogen-reduced platelet product (INTERCEPT) was approved for use in the US by the US Food and Drug Administration in 2014 and has been used in a small number of neonates worldwide ⁴⁶ . Pathogen-reduced platelet products have been used in some European countries for over a decade. The published RCTs comparing the use of pathogen-inactivated platelets with conventional platelets did not specifically enroll neonates. There are few prospective observational studies which have found no evidence of major adverse events in infants (<1 year old). In neonatal patients treated with phototherapy devices with a peak energy wavelength of less than 425 nm, use of the INTERCEPT blood system for the preparation of platelet components is contraindicated owing to a potential for erythema from interaction between ultraviolet light and amotosalen (interceptbloodsystem.com) ⁴⁷ . Other pathogen reduction products in the pipeline, such as Mirasol, a riboflavin (vitamin B ₂)-based system, are still in clinical trials. A recent Cochrane review of RCTs comparing the transfusion of pathogen-reduced platelets with standard platelets suggested no evidence of a difference in the incidence of clinically significant bleeding complication or all-cause mortality. The review also suggested that pathogen-reduced platelet transfusions increase the risk of platelet refractoriness and platelet transfusion requirements. Furthermore, evidence on subgroup differences in multiple transfusion trials between the two pathogen-reduced platelet technologies (Intercept and Mirasol) revealed that for all-cause mortality and the interval between platelet transfusions, Intercept was more favorable; however, these studies are not directly applicable to children/neonates ⁴⁸ . Longitudinal studies are needed for both products and their use in neonates as well as older children.

Study of platelet transfusion strategies to prevent or mitigate bleeding in neonatal and adult patients has been long outlined as clinical trial opportunities in transfusion medicine by the National Heart, Lung, and Blood Institute thinktank of experts ⁴⁹ . Platelet trigger trials assessing clinically relevant outcomes are now under way in preterm neonates with thrombocytopenia. PlaNeT1 was an observational study of current practice and outcomes of platelet transfusions in neonates. PlaNeT2 ( http://www.planet-2.com) is an ongoing randomized prophylactic threshold trial randomly assigning neonates to a 25 versus 50 × 10 ⁹/L transfusion threshold. The NeoBAT bleeding assessment score is being used by the PlaNeT2 study. Thus far, treatments or research studies with thrombopoietic growth factors have not been undertaken in neonates. Besides studies of the platelet transfusion threshold, studies are needed to characterize the overall platelet function and the global hemostatic profile of neonates, especially preterm infants, as well as the use of thromboelastography (TEG) or thromboelastometry (ROTEM) to monitor hemostatic treatment to better predict their bleeding risk and thus direct prophylactic or therapeutic measures accordingly for IVH/PVH ⁵⁰ .

The neonatal coagulation system matures slowly after birth, and, depending on the levels of various factors at birth, the “normal” reference ranges in neonates and infants are quite different from those in adults. Even within the neonates, the reference ranges vary depending on the gestational age. However, owing to the lack of awareness of the different ranges, transfusion decisions in neonates are unfortunately made on the basis of the adult ranges or “cutoffs” used to define coagulopathy.

A national audit of transfusion practice from the UK indicated that almost 50% of plasma transfusions are given prophylactically to neonates with “abnormal coagulation values” with no evidence of active bleeding, in order to prevent IVH, although there is very low-grade evidence to support the effectiveness of this practice. Likewise, there is widespread use of plasma reported in hospitalized children in the absence of demonstrated efficacy or effectiveness ⁵² .

Karam et al. and PlasmaTV investigators reported, in an international point prevalence study of plasma transfusions in critically ill children, that about one-third of the plasma transfusions given in a PICU setting are without any evidence of bleeding or an invasive procedure or surgery warranting a plasma transfusion and that only 22% were administered in case of critical bleeding ⁵³ .

In a follow-up of an a priori secondary analysis of a prospective observational study, investigators of the PLASMA TV study reported lower PICU mortality in the solvent detergent versus fresh frozen plasma (FFP)/frozen plasma (OR 0.40, 95% CI 0.16–0.99; p = 0.05). Based on these results, the authors suggested that solvent detergent plasma use in critically ill children may be associated with improved survival ⁵⁴ . There are no dedicated studies about effectiveness/adverse effects of pathogen-reduced products in term and preterm neonates.

Current expert opinion supports the “therapeutic” use of plasma either (1) with active bleeding or (2) concurrent to invasive procedures in patients at high risk of bleeding and those recognized to have an abnormal coagulation profile. An abnormal coagulation profile would be identified as the conventional coagulation parameters prothrombin time (PT) or activated partial thromboplastin time (aPTT) being significantly above the normal age and gestation adjusted reference ranges.

The BSH recommends that prophylactic FFP “not” be administered to “non-bleeding” children with minor prolongation of PT/aPTT, including prior to surgery, but it may be considered for surgery to critical sites ⁴⁴ . BSH guidelines also advise that plasma not be used as a means for volume replacement or for prevention of bleeding (for example, IVH) in neonates ⁴⁴ .

Plasma use in neonates comes with multiple risks: it could range from a simple allergic reaction secondary to plasma proteins or a febrile non-hemolytic reaction to severe life-threatening reactions, including TRALI, TACO, febrile reactions, and hemolysis, all likely under-recognized and under-reported in neonates.

Although plasma is usually transfused for its hemostatic properties, there have been reports of an association between plasma transfusion and venous thrombotic outcomes in neonatal as well as pediatric patients, with the coagulation factors and platelet-derived microparticles in plasma being considered responsible for the thrombotic effects ⁵⁵ . The neonatal study reported a dose-dependent effect with more than five times the odds of thrombosis in neonates who received greater than 50 mL/kg of plasma by day 5 of age. Raffini et al., in a study of plasma transfusions in US hospitals, found that the rate of venous thrombosis in children (including neonates) receiving plasma was 10 times greater than the rate for hospital admissions without plasma transfusion ⁵² .

The first necessary step in furthering research and evidence-based data to guide the indications for plasma use in neonates is the establishment of a well-validated and age-specific reference range for coagulation tests in preterm and term neonates and its variability with severity of illness. The ranges must be established for the classic tests, including PT/aPTT/international normalized ratio, as well as for the various global hemostatic assays, including TEG and ROTEM. The next step is having a validated bleeding prediction and assessment battery for neonates to standardize the actual bleeding. Correlation of the prolonged values with actual risk for clinically significant bleeding is another key step, and no correlation has been established so far. Finally, there is a need for RCTs to explore the prophylactic use of plasma in neonates deemed at high risk for bleeding or with abnormal coagulation tests. Hemostatic dysfunction is also prevalent among neonates with moderate-to-severe hypoxic-ischemic encephalopathy and is associated with an increased risk of bleeding and high transfusion burden ⁵⁶ .