Minimizing blood loss and the need for transfusions in very premature infants

---

The Canadian Paediatric Society gives permission to print single copies of this document from our website. For permission to reprint or reproduce multiple copies, please see our copyright policy.

Iatrogenic anemia, secondary to blood draws, increases the need for transfusions in preterm infants.[1] There are risks associated with drawing blood, such as pain and its related complications, and sepsis from repeated breakage of the skin. Packed red blood cell (PRBC) transfusions may lead to acute lung injury, graft-versus-host disease,[2] and to increased in-hospital mortality of very low birth weight (VLBW) infants.[3] The association between PRBC transfusions and necrotizing enterocolitis remains controversial.[4]-[6]

Variations exist among transfusion practices in neonatal intensive care units (NICUs), suggesting that blood sampling methods, the frequency of routine testing, the noninvasive monitoring of infants and indications or thresholds for transfusions may differ.[7]-[9] Reports of reduced transfusion rates and amounts, following implementation of quality improvement initiatives, further suggest that strategies to minimize blood loss and reduce transfusions can be effective.[10][11] The purpose of the present statement is to summarize evidence-based strategies to reduce blood sampling and blood transfusions in very preterm infants.

Methods

A comprehensive search of the literature was performed by an expert librarian using MEDLINE, including in process and other nonindexed citations (1946 to August 31, 2014), Embase (1947 to August 31, 2014) and the Cochrane Central Register of Controlled Trials (CENTRAL, Issue 2, June 2014). The population of interest included VLBW infants (<1500 g at birth) during their hospital admission. A total of 1491 references were retrieved, of which 43 full-text articles and six Cochrane reviews were reviewed. A search of the grey literature was also performed.

The hierarchy of evidence from the Centre for Evidence-Based Medicine[12] was applied to the publications identified (Table 1). Recommendations are based on the format by Shekelle et al [13](Table 2).

Does the timing of umbilical cord clamping at delivery influence the need for and frequency of PRBC transfusions?

In a systematic review, Rabe et al[14](15 trials involving 738 infants <36 weeks gestational age [GA]) reported that delayed cord clamping (30 s to 180 s) is associated with fewer infants requiring a PRBC transfusion during their hospital stay (n=7 trials; RR 0.61 [95% CI 0.46 to 0.81]; number needed to treat [NNT]=8) and fewer transfusions per infant (mean difference [MD] −1.26 [95% CI −1.87 to −0.64]). Delayed cord clamping should be performed in all preterm infants who are not in immediate need of resuscitation (level of evidence: 1a).

TABLE 1
Levels of evidence for therapies
Level	Nature of evidence
1a	Systematic review (with homogeneity) of RCTs
1b	Individual RCT (with narrow CI)
2a	Systematic review R (with homogeneity) of cohort studies
2b	Individual cohort study (or low-quality RCT, eg, <80% follow-up)
3a	Systematic review (with homogeneity) of case-control studies
3b	Individual case-control study
4	Case-series (and poor quality cohort and case-control studies)
5	Expert opinion without explicit critical appraisal, or based on physiology, bench research or first principles
RCT Randomized controlled trial

TABLE 2
Grades of recommendation
A	Consistent level 1 studies
B	Consistent level 2 or 3 studies
C	Level 4 studies
D	Level 5 evidence or troublingly inconsistent or inconclusive studies or any level

In a second systematic review, Ghavam et al[15] reported that delayed cord clamping or cord milking (ie, squeezing or stripping the cord toward the newborn immediately after birth to decrease the time required to transfer placental blood) significantly reduced the number of blood transfusions in infants <30 weeks GA or weighing <1000 g, as compared with early cord clamping (weighted mean difference [WMD] −2.22 [95% CI −2.52 to −1.92]). Only one small study in this review evaluated cord milking, which reduced the need for a PRBC transfusion during hospital stay (65% versus 30%).[16] March et al[17] observed a lower rate of blood transfusions in 75 infants <29 weeks GA managed with cord milking rather than early cord clamping: RR 0.86 (95% CI 0.7 to 1.0). Patel et al[18] reported that cord milking was associated with a decrease in the need for blood transfusion in infants born <30 weeks GA (57% versus 79%), as compared with historical controls. No benefit to cord milking was observed in a study involving 44 infants by Allan et al,[19] or in a second study involving 58 infants by Rabe et al.[20] None of these studies reported harm or higher rates of intraventricular hemorrhages in the cord milking group. There is insufficient evidence at present to recommend umbilical cord milking as routine practice (level of evidence: 2b).

Does a low hemoglobin concentration threshold for blood transfusion safely reduce the need for transfusions?

Whyte et al[21] performed a systematic review in which a low hemoglobin (Hb) transfusion threshold, based on whether the baby is on respiratory support and on postnatal age, is defined as the following: a Hb concentration of 115 g/L (respiratory support) or 100 g/L (no support) in the first week, a Hb concentration of 100 g/L (respiratory support) or 85 g/L (no support) in the second week, and a Hb concentration of 85 g/L (respiratory support) or 75 g/L (no support) in the third week. These thresholds closely match those used in the largest trial (Kirpalani et al[22]). The review included four trials (n=614 infants <32 weeks GA and weighing <1500 g) and the authors reported that lower Hb concentration thresholds were associated with a modest reduction in the need for ≥1 blood transfusion(s) (RR 0.95 [95% CI 0.91 to 1.0]), the number of transfusions per infant (WMD −1.12 [95% CI −1.75 to −0.49]), as well as in donor exposure (WMD −0.54 [95% CI −0.93 to −0.15]). No significant effect was observed for risk of death or serious morbidity at discharge. A 2014 review concurs with the above results (level of evidence: 1a).[23]

Kirpalani et al[22] (n=451) reported outcomes at 18 to 21 months corrected GA and found no increased risk for death or neurodevelopmental impairment (RR 1.45 [95% CI 0.94 to 2.21]). A post hoc secondary analysis, using a less severe definition of cognitive delay (mental developmental index <85), reported rates of 45% versus 34%, favouring the high threshold group: OR 1.81 (95% CI 1.12 to 2.93) (P=0.016).[24] These results should be considered as hypothesis generating rather than practice changing (level of evidence: 1b).

In conclusion, and based on the findings of the Premature Infants in Need of Transfusion (PINT) study[22] and the Cochrane review, in which no difference was observed in the main primary outcome (death or disability), using the lower threshold of Hb concentration when transfusing is recommended, always recognizing the need for individualized targets based on the patient’s health status.

Does the volume of blood transfused reduce the need for further PRBC transfusions?

A randomized controlled trial (RCT) involving 20 infants weighing <1500 g by Wong et al[25] found that receiving 20 mL/kg versus 15 mL/kg of PRBC did not effect the need for further transfusion. However, methodological limitations prevent drawing any firm conclusions regarding efficacy. Paul et al[26] observed that receiving 20 mL/kg of PRBC versus 10 mL/kg led to a higher hematocrit without significant physiological detriment in a RCT involving 13 infants weighing <1500 g, although differences in transfusion volumes were not assessed.

Transfusing a higher volume of blood may decrease the risk for exposure to >1 donor (level of evidence: 2b).

Does treatment with recombinant human erythropoietin safely reduce the need for PRBC transfusions?

In a systematic review (27 trials involving 2209 infants), Ohlsson et al[27] reported that erythropoietin administered in the first week of life (intravenous [IV] or subcutaneous [SC], 70 IU to 2100 IU per week, three times a week) for two to nine weeks or until discharge, reduces the need for ≥1 transfusion(s) postrandomization (RR 0.79 [95% CI 0.73 to 0.85]; NNT=7). There was no increased risk for retinopathy of prematurity (RoP) (all stages); however, early erythropoietin was associated with a greater risk for severe RoP (stage three or higher): RR 1.4 (95% CI 1.02 to 2.13); number needed to harm =33. The same authors (30 trials involving 1591 infants)[28] found that the later use of erythropoietin (at eight to 28 days of age), administered by SC, IV or by mouth (PO) (150 IU/kg/week to 2100 IU/kg/week SC, 200 IU/kg/week to 300 IU/kg/week IV or 7000 IU/kg/week PO) reduces the need for ≥1 transfusion (RR 0.71 [95% CI 0.64 to 0.79]; NNT=6). Although a trend was noted, no significantly higher risk for RoP was identified. A significant reduction in the number of transfusions per infant (MD −0.22 [95% CI −0.38 to −0.06]) was observed; however, donor exposure is probably not reduced because most patients, whether they were treated early or late, received a transfusion before randomization. The routine use of erythropoietin is, therefore, not recommended (level of evidence: 1a).

In another systematic review (two trials, n=286), Aher et al[29] found no significant reduction in the need for ≥1 transfusions (RR 0.91 [95% CI 0.78 to 1.06]) or in the number of transfusions per infant (typical MD −0.32 [95% CI −0.92 to 0.29]) when erythropoietin was administered early (in the first eight days of life) or later (≥8 days of life). A higher risk for severe RoP was observed in the early group (RR 1.40 [95% CI 1.05 to 1.86]; NNH=6 [95% CI 3 to 33]) (level of evidence: 1a).

In an RCT conducted by Gumy-Pause et al,[30] a progressive increase in the dose of erythropoietin based on the absolute reticulocyte count did not prove more effective than a fixed dose. Also, a small RCT[31] with methodological limitations did not demonstrate a difference in the proportion of infants requiring a transfusion when comparing oral with SC erythropoietin. A noninferiority trial found IV erythropoietin to be noninferior to SC erythropoietin.[32] (level of evidence: 2b.)

Ohls et al[33] randomly assigned 102 infants <29 weeks GA to receive SC darbepoietin (10 mg/kg weekly), SC erythropoietin (400 U three times per week) or placebo in the first week of life. Infants also received supplemental iron, folate and vitamin E. While this study was not an equivalence trial, both darbepoietin and erythropoietin were found to be effective in reducing the need for transfusion, compared with placebo (level of evidence: 2b).

Overall, although erythropoietin use has been associated with a reduced number of transfusions, the number of donor exposures may not be reduced and concern about a higher risk for RoP remains. Therefore, erythropoietin use should be individualized for patients whose parents have refused other blood products.

Does enteral iron supplementation reduce the need for blood transfusions?

One Cochrane systematic review (26 trials involving 2726 infants) by Mills et al[34] from 2012, reported no hematological benefit from administering an iron supplement within the first 8.5 weeks of life. Transfusion needs were not assessed. Another systematic review by Long et al[35] that same year (15 studies) reported a similar outcome: only three studies reported on transfusions and two of three found no effect from iron therapy. One study[36] randomly assigned 133 infants weighing <1301 g to receive supplemental iron (2 mg/kg per day), either when feeds reached 100 mL/kg/day or at 61 days of age. Fewer infants in the early iron group were transfused after 14 days. However, many had been transfused by the time they were seven days of age.

Taylor et al[37] randomly assigned 153 infants weighing <1500 g who received supplemental iron (2 mg/kg/ day) above and beyond iron-fortified formula or expressed breast milk when their feeds had reached 120 mL/kg/day and until they turned 36 weeks corrected GA or were discharged from hospital. Joy et al[38] randomly assigned 93 infants weighing <1500 g who were to receive 2 mg/kg/day of elemental iron starting at two or six weeks of age. There was no significant difference in the need for transfusion or the number of transfusions per infant in either study. In another small RCT by Fujiu et al,[39] no benefit was observed for infants receiving 4 mg/kg/day of iron starting at a mean of 20 days of age, when compared with placebo. Both groups received erythropoietin (400 U/kg/week).

Overall, there is no evidence of benefit for using supplemental iron therapy to prevent or reduce the need for blood transfusions in the neonatal period. However, administering supplemental iron in physiological doses (2 mg/kg/day to 3 mg/kg/day, or 4 mg/kg/day to 6 mg/kg/day if an infant is believed to be iron deficient), once full oral feeds have been achieved, has been demonstrated to improve hemoglobin and ferritin levels after two months of treatment and to reduce the risk for iron-deficiency anemia (along with an associated risk for lower cognitive function) in the first year of life.[34][40] Human milk fortifier and enriched preterm formulas provide at least 2 mg/kg/day of iron to infants who are not fluid restricted (level of evidence: 2b).

Does the use of noninvasive monitoring in preterm infants reduce the need for transfusions?

Six studies (involving 200 infants weighing <1500 g and 145 infants <1000 g) examining simultaneous blood CO₂ and noninvasive CO₂ measures were identified. Four (n=723) assessed end-tidal CO₂, one assessed transcutaneous CO₂ (n=46) and one assessed both (n=37).[41]-[46] The correlation between blood CO₂ and noninvasive monitoring was acceptable (correlation coefficient 0.55 to 0.82). In infants with significant pulmonary disease, the end-tidal CO₂ partial pressure of CO₂ may be quite discrepant due to large dead space volumes. No evidence of poorer accuracy in the monitoring devices was found for VLBW infants (level of evidence: 3b).

In one systematic review (22 trials, 1628 infants),[47] the correlation was excellent between transcutaneous and serum bilirubin in the entire population, including infants <32 weeks of age and <29 weeks of age (two studies). Estimates of bias and precision for both devices being tested were comparable with those reported for term populations. This finding suggests that transcutaneous bilirubin devices can be used as a screening tool and to help reduce blood sampling (level of evidence: 2a). No study has examined the impact of transcutaneous bilirubin measurement on the need for blood transfusions in preterm newborns.

Two studies compared Hb values obtained via a pulse co-oximeter (Masimo Corporation, USA) with laboratory values (83 infants in total; 213 paired samples). The reported correlation was moderate (0.75 and 0.758) with bias of 0.1±1.56 g/dL and 0.86±3.4 g/dL.[48][49] A third study examined the correlation between spectroscopic Hb measurements (using a transcutaneous device: Mediscan 2000 [Mediscan Systems, India]) in 80 neonates (with a mean age of 29.5 weeks). The correlation coefficient was high (0.96 to 0.99) among 313 observations.[50] (Level of evidence: 4.)

While there is no direct evidence that noninvasive monitoring reduces the need for a transfusion, the timely use of such devices may reduce the need for blood testing and thus, indirectly, the need for transfusion.

Is there evidence to support the use of point-of-care testing techniques for reducing PRBC transfusions?

Point-of-care testing typically involves bedside assessments of at-risk infants and usually requires a very small amount of blood to yield immediate results. Mahieu et al[8] (n=1397) found significant decreases in the number of transfusions per infant (2.53 versus 1.57), in the proportion of infants who required a transfusion and in phlebotomy losses, after a point-of-care analyzer (electrolytes, gas, hemoglobin, bilirubin) had been introduced in the NICU, despite steady numbers for the tests being performed. Madan et al[11] (n=80) also found a decrease in the number of transfusions and phlebotomy losses per infant. Point-of-care testing should be considered in preterm infants. This strategy requires support from the institution’s biochemistry laboratory and compliance with accreditation standards (level of evidence: 4).

Is there evidence to support the use of in-line or other continuous monitoring techniques for reducing PRBC transfusions?

Widness et al[51] tested an in-line blood gas and chemistry monitor in a randomized fashion, and compared it with results using a benchtop analyzer for a population of very preterm infants (n=93). A reduction in blood transfusions in the first week and a decrease in phlebotomy losses over the first two weeks of the study were observed. Moya et al[52] investigated readings from a different in-line blood gas analyzer (Sensicath [Optical Sensors Inc, USA] system) for 44 preterm infants. Over six days of use, the in-line catheter reduced phlebotomy losses; however, there was no observed decrease in the need for transfusion. The need for an arterial line in both studies limited the use of such techniques. While the evidence for continuous in-line monitoring demonstrates some promise, data is limited and the clinical use of these devices is restricted by the need for an indwelling arterial catheter (levels of evidence: 2b and 3b).

Beardsall et al[53] investigated the accuracy of a SC continuous glucose monitoring system in VLBW infants (n=188). Although the system was well tolerated, it failed to meet the American Diabetes Association standard of <10% error for each sample, particularly when blood glucose levels were low. A study by Baumeister et al[54] (n=10) used a SC microdialysis catheter and found good correlation between serum and microdialysis technique, along with sufficient diagnostic sensitivity to detect low serum glucose levels (level of evidence: 4).

Is there evidence to support a particular blood sampling technique (arterial, venous or capillary) to reduce total phlebotomy losses and/or PRBC transfusions?

Blood sampling is technically challenging in preterm infants. Venous and arterial sampling requires skill and experience. Capillary sampling requires less skill. However, no literature was available to determine whether one technique was superior to another in terms of minimizing blood loss or the need to repeat testing because of sampling difficulties.

Is there evidence to support limiting routine blood sampling?

No evidence exists as to what constitutes routine blood sampling for extremely premature infants in the NICU. However, in one review article, Carroll et al[55] reported large inter-unit variability in the amounts of blood drawn in the first two weeks of life (up to 100% difference in one study).[56] Phlebotomy losses in excess of amounts needed for analysis were also reported by Lin et al,[57] with a mean (± SD) volume of blood overdraw of 19.0±1.8% in a single-centre cohort study. Unmarked blood collection tubes, a birth weight <1 kg, a perceived need for single specimen draws and the individual phlebotomist were all factors influencing overdraw (level of evidence: 4).

Recommendations for practice

Based on summaries of the evidence described above, the following guidance is offered for minimizing blood loss and the need for transfusion in extremely preterm infants:

• Delay cord clamping (optimally for 30 s to180 s after birth) for preterm infants who are not in immediate need of resuscitation (grade A recommendation).
• While cord milking may be considered as an alternative to delayed cord clamping, because so few patients have been enrolled in randomized trials to date, the technique cannot be recommended as routine practice at the present time (grade B recommendation).
• For infants in the first and second week of life, minimum hemoglobin levels of 100 g/L and 85 g/L, respectively, are recommended. Infants requiring respiratory support may require transfusions at a higher threshold. In view of the concerns regarding the increased risk for cognitive delay, clinicians should avoid using thresholds lower than those tested in published RCTs (grade D recommendation).
• A higher volume of transfused blood (20 mL/kg) should be considered when transfusing a preterm baby, if the hemodynamic and respiratory status of the patient permits (grade D recommendation).
• Routine use of erythropoietin (SC, IV or PO) is not recommended. Dosage and delivery method should be individualized if parents withhold consent to transfuse blood (grade A recommendation).
• Supplemental iron does not appear to reduce the need for blood transfusion in preterm neonates. For treatment beyond two months of age, however, iron supplements may improve hematological values and help to avoid iron deficiency anemia. Supplementation with physiological doses (2 mg/kg/day to 3 mg/kg/day, or 4 mg/kg/day to 6 mg/kg/day in newborns who are iron deficient) should be considered (grade D recommendation).
• Noninvasive CO₂ monitoring should be considered when caring for ventilated preterm infants. Exercise caution in infants with significant pulmonary disease because of potential for greater inaccuracy (grade D recommendation).
• Using noninvasive bilirubin monitoring devices before phototherapy is recommended (grade D recommendation).
• Point-of-care testing should be considered in preterm infants. This strategy requires support from the institution’s biochemistry laboratory and compliance with accreditation standards (grade C recommendation).
• Caregivers should scrutinize and limit blood sampling to the minimum required for safe clinical care. Using marked tubes to indicate the minimum volume to collect, and clustering blood samples to reduce the number of skin breaks and painful procedures are recommended (grade D recommendations).

Acknowledgements

This position statement has been reviewed by the Community Paediatrics Committee of the Canadian Paediatric Society.

---

CPS FETUS AND NEWBORN COMMITTEE


Members: Ann L Jefferies MD (past Chair), Thierry Lacaze-Masmonteil MD (Chair), Leigh Anne Newhook MD (Board Representative), Leonora Hendson MD, Brigitte Lemyre MD, Michael R Narvey MD, Vibhuti Shah MD, S Todd Sorokan (past member)
Liaisons: Linda Boisvert RN, Canadian Association of Neonatal Nurses; Andrée Gagnon MD, College of Family Physicians of Canada; Robert Gagnon MD, Society of Obstetricians and Gynaecologists of Canada; Juan Andrés León MD, Public Health Agency of Canada; Patricia A O’Flaherty MN MEd, Canadian Perinatal Programs Coalition; Eugene H Ng MD, CPS Neonatal-Perinatal Medicine Section; Kristi Watterberg MD, Committee on Fetus and Newborn, American Academy of Pediatrics

Principal authors: Brigitte Lemyre MD, Megan Sample MD, Thierry Lacaze-Masmonteil MD

---

References

---

Disclaimer: The recommendations in this position statement do not indicate an exclusive course of treatment or procedure to be followed. Variations, taking into account individual circumstances, may be appropriate. Internet addresses are current at time of publication.