Transfusion thresholds for preterm infants: How low should we go?
Article Outline
Abbreviations: RBC, Red blood cell , VLBW, Very low birth weight
Neonatologists have struggled for decades with the decision of when to transfuse preterm infants with red blood cells (RBCs). As a result of phlebotomy blood losses and inadequate erythropoiesis, very low birth weight (VLBW) infants become progressively anemic. In an effort to limit the risks associated with RBC transfusion, many neonatal units have adopted more restrictive guidelines for transfusing preterm infants.1, 2, 3, 4 Unfortunately, these changes in practice have not been accompanied by systematic examination of the risks and benefits of restricting transfusions.
See related article, p 301
There is a dearth of evidence on which to base transfusion decisions for VLBW infants. Until recently, the only information available was derived from several small clinical trials that provided no clear guidance for transfusion decisions for VLBW infants.5, 6, 7, 8 Several of these studies suggested possible benefits from more liberal transfusion guidelines, in particular, less frequent severe apnea6 and faster weight gain.7 Two additional, more recent studies9, 10 provide new information about the relative risks and benefits of using restrictive (transfusion at lower hemoglobin) rather than more liberal criteria for transfusing VLBW infants; these are the Iowa Trial9 and the PINT (Premature Infants in Need of Transfusion) Trial, an elegant study by Kirpalani, Whyte, and colleagues,10 which appears in this issue of The Journal of Pediatrics. Both trials compared the outcomes of two groups of infants who were randomly assigned to restrictive or liberal transfusion criteria, based on hematocrit or hemoglobin thresholds for transfusion. Despite differences in experimental design and results, these new trials together strengthen the base of evidence available to guide decisions about providing RBC transfusions for VLBW infants. Of course, key questions remain unanswered and will spur further research in this important area.
The Iowa Trial9 was a single-center randomized clinical trial with sample size calculated to test whether using lower hematocrit thresholds for RBC transfusion would reduce the number of transfusions received by infants with birth weight of 500 to 1300 g. The transfusion thresholds varied with the level of respiratory support needed and, indirectly, with postnatal age.
The PINT Trial10 was a larger multicenter randomized clinical trial designed to examine the impact of transfusion strategy on the incidence of a composite outcome—death, retinopathy of prematurity, bronchopulmonary dysplasia, or abnormal brain ultrasound—in infants with birth weight below 1000 g. The restrictive and liberal transfusion groups were defined by the hemoglobin thresholds for RBC transfusion; the thresholds varied with age and with the level of respiratory support needed.
The transfusion thresholds used for the restrictive transfusion groups were similar in the two trials (Table). However, the transfusion thresholds for the liberal transfusion group were higher in the Iowa Trial. As a result, the separation achieved in mean hemoglobin between the two study groups at 4 to 6 weeks of age was 2.5 times larger in the Iowa Trial, 2.7 g/dL compared with 1.1 g/dL (Table). In the PINT Trial, by the time of discharge from the hospital, the mean hemoglobin values for the two transfusion groups were no longer different, 10.6 g/dL versus 10.8 g/dL.
Table. Comparison of major clinical trials of restrictive versus liberal criteria for red blood cell transfusion in preterm infants
| Iowa Trial9 | PINT Trial10 | |||
|---|---|---|---|---|
| Restrictive | Liberal | Restrictive | Liberal | |
| Participating centers | 1 | 10 | ||
| No. of subjects | 100 | 451 | ||
| Treatment allocation | Random assignment | Random assignment | ||
| Prognostic stratification criteria | Birth weight | Birth weight, center | ||
| Mean birth weight (g) | 954 | 958 | 771 | 769 |
| Mean gestational age (wk) | 28 | 28 | 26 | 26 |
| Transfusion thresholds (hemoglobin, g/dL)⁎ | ||||
 Highest transfusion threshold | 11.3 | 15.3 | 11.5 | 13.5 |
| Lowest transfusion threshold | 7.3 | 10.0 | 7.5 | 8.5 |
| Mean hemoglobin† (g/dL), all subjects | 8.3 | 11.0 | 10.1 | 11.2 |
| Mean hemoglobin difference (g/dL) | 2.7 | 1.1 | ||
| Mean number of transfusions | 3.3 | 5.2‡ | 4.9 | 5.7 |
| Mean number of RBC donor exposures | 2.2 | 2.8 | 2.1 | 2.6‡ |
| Infants never transfused | 10% | 12% | 11% | 5%‡ |
| Died | 4% | 2% | 22% | 18% |
| Brain injury§ (% of survivors) | 12% | 0%‡ | 13% | 16% |
| Severe ROP∥ (% of survivors) | 4% | 4% | 19% | 18% |
| Death or brain injury | 16% | 2%‡ | 31% | 31% |
⁎ For Iowa Trial, threshold hemoglobin values computed from hematocrit values by dividing by 3. For PINT Trial, threshold hemoglobin values for capillary samples used for this comparison. Transfusion thresholds in both studies varied with respiratory status and postnatal age. |
† Mean hemoglobin levels and differences before transfusion at age 6 weeks (Iowa Trial) or at age 4 weeks (PINT Trial). |
‡ Statistically significant difference, restricted versus liberal, P < .05. |
§ As defined in each study. |
∥ Retinopathy of prematurity, stage 3 or higher. |
The Iowa Trial found a reduction in the number of transfusions but not the number of donor exposures in the restrictive transfusion group. The lack of difference in the number of donors presumably resulted from the use of a single-donor transfusion program.11 The PINT Trial found a reduction in the number of transfusions given according to study criteria in the restrictive transfusion group, but this reduction was offset by an increase in the number of transfusions given for clinical indication; consequently, the trend toward fewer total (study and nonstudy) transfusions in the restrictive transfusion group was not statistically significant. The PINT Trial found no reduction in donor exposures with restrictive transfusion guidelines when all transfusions were considered, including platelet and plasma transfusions. However, when only RBC transfusions were compared, the number of donors was lower in the restrictive transfusion group (Table).
Perhaps the best measure of success in limiting transfusions is the number of infants who require no transfusions. The PINT Trial found that significantly more infants avoided RBC transfusion altogether when restrictive criteria were used, 11% versus 5% (Table). The Iowa Trial found no such difference, with 10% to 12% of infants in both groups avoiding all transfusions.
In the Iowa Trial, the infants in the restrictive transfusion group were more likely to develop parenchymal brain hemorrhage or periventricular leukomalacia; they also had more frequent episodes of apnea, both total and severe apnea. No such differences were found in the PINT Trial.
How are we to interpret these important differences in the results of the two trials? The lack of significant reduction in number of RBC transfusions with the use of restrictive guidelines in the PINT Trial may have resulted from the smaller separation in hemoglobin between the two study groups, but it was also affected by the increased number of transfusions that were outside of study protocol but clinically indicated in the restrictive transfusion group. The Iowa Trial showed that a greater difference in transfusion thresholds between the restrictive and liberal transfusion groups can reduce total transfusions but not donor exposures when a single-donor transfusion program is used. Paradoxically, use of restrictive transfusion criteria did not increase the number of infants who required no transfusions at all in the Iowa Trial, but it did in the PINT Trial.
The failure of the PINT Trial to replicate the increase in apnea and brain injury seen in the Iowa Trial with restrictive transfusion practices could have several explanations. The first possible explanation is that these findings in the Iowa Trial resulted from bias or random statistical error. Apnea as recorded by the bedside nurse is somewhat subjective, and the nurses were not blinded to the infants’ transfusion assignments. Moreover, the composite outcome of parenchymal brain hemorrhage or periventricular hemorrhage was not planned initially as part of the data analysis but was added after the study was completed when the clustering of both single events (parenchymal brain hemorrhage and periventricular leukomalacia) in the restrictive transfusion group was noted. The second possible explanation, one that should not be dismissed lightly, is that the higher hemoglobin concentrations maintained in the Iowa liberal transfusion group compared with the corresponding group in the PINT Trial may have conferred protection against apnea and brain injury.
In the 21st century, the cumulative wisdom generated by multiple clinical trials on the same subject is often addressed using systematic review and meta-analysis. In a meta-analysis of restrictive versus liberal RBC transfusion guidelines for VLBW infants, the PINT Trial will outweigh the Iowa Trial and other smaller trials based simply on the large number of subjects in the PINT Trial; this will occur even though the “treatment effect” is smaller in the PINT Trial than in the Iowa Trial, with a hemoglobin separation of only 1 g/dL between the restrictive and liberal transfusion groups.
If the results of both trials are assumed to be valid, clinicians are left weighing the potential benefits of restrictive transfusion guidelines—more infants avoided transfusion altogether with restrictive transfusions in the PINT Trial (but not the Iowa Trial)—against the potential benefits of liberal transfusion guidelines, especially protection against major brain injury found in the Iowa Trial (but not confirmed by the PINT Trial). If we believe the finding of the PINT Trial that restrictive transfusion guidelines protect more infants from receiving one or more transfusions, the number needed to treat would be 17; in other words, 17 infants would have to be transfused according to the more restrictive criteria to allow one infant to avoid transfusions altogether. If we believe the finding of the Iowa Trial that liberal transfusion guidelines reduce the risk of brain injury, the number needed to treat to achieve this benefit would be 8; in other words, 8 infants would have to be more liberally transfused to prevent one case of parenchymal brain hemorrhage or periventricular leukomalacia. The number needed to treat to prevent one infant from dying or having major brain injury would be seven. These calculations based on the Iowa Trial should be viewed with caution because the study was relatively small.
What is the bottom line? The documented benefits of restrictive transfusion practice are a slight reduction in the number of transfusions and, in the absence of a single-donor program, exposure to fewer RBC donors; in addition, the PINT Trial showed an increase in the number of infants who required no transfusions at all. When weighed against the potential benefits of liberal transfusion in the Iowa Trial—lower risk of apnea and major brain injury—even though not confirmed in the PINT Trial, the conclusion is that we should stop the movement toward more restrictive transfusion practices for VLBW infants and take comfort in the low risks of modern transfusion medicine. The risks of RBC transfusion have lessened with improving blood banking methods. Efforts to apply more restrictive transfusion guidelines have been driven, in part, by the risks of transfusion, which include transmission of infection from donor to recipient. Careful screening of donors and leukoreduction or radiation of the blood have reduced this risk to a very low level. Moreover, the widespread adoption of single donor transfusion programs for patients in the neonatal intensive care unit has markedly limited exposure of blood recipients to multiple donors. These programs enable assignment of a unit of blood to meet all the transfusion needs of one or two infants for up to 6 weeks. As a result, most VLBW infants are exposed to only one donor, although some require two.12 Single-donor transfusion programs should be the norm for VLBW infants and probably for all patients in the neonatal intensive care unit.
Decisions about transfusing VLBW infants with RBCs should be based on the risks and benefits of applying various transfusion guidelines and further refined based on the clinical status of individual infants. The risks of transfusion per se are very low and should carry less weight in making transfusion decisions.
The question remains of how far we can push the anemic preterm infant before transfusing him. Efforts to eliminate transfusions should be revisited in light of the minimal benefits of restrictive transfusion practice shown in these two trials and the potentially major benefits of liberal transfusion practice shown in the Iowa Trial. Perhaps the drive to eliminate transfusions by tolerating moderate to severe iatrogenic anemia should be halted until more information is available. The advantage of fewer transfusions is small compared with the potential benefit of more liberal transfusions in protecting the brain.
References
- . Changing patterns of red blood cell transfusion in very low birth weight infants . J Pediatr . 1996;129:680–687
- . Impact of transfusion guidelines on neonatal transfusions . J Perinatol . 1998;18:92–97
- . Changing practices of red blood cell transfusions in infants with birth weight less than 1000 g . J Pediatr . 2000;136:220–224
- . Red blood cell transfusions in very and extremely low birthweight infants under restrictive transfusion guidelines (is exogenous erythropoietin necessary?) . Arch Dis Child Fetal Neonatal Ed . 2001;84:F96–F100
- . The role of RBC transfusion in the premature infants . Am J Dis Child . 1984;138:831–833
- A randomized trial to develop criteria for administering erythrocyte transfusions to anemic preterm infants 1 to 3 months of age . J Perinatol . 1989;9:246–253
- . Empiric red cell transfusion in asymptomatic preterm infants . Acta Paediatr . 1993;82:30–34
- . The effect of blood transfusion protocol on retinopathy of prematurity (a prospective, randomized study) . Pediatrics . 1999;104:514–518
- Randomized trial of liberal versus restrictive guidelines for red blood cell transfusion in preterm infants . Pediatrics . 2005;115:1685–1691
- A randomized controlled trial of a restrictive (low) versus liberal (high) transfusion threshold for extremely low birth weight infants (the PINT study) . J Pediatr . 2006;149:301–307
- AS-1 red cells for neonatal transfusions (a randomized trial assessing donor exposure and safety) . Transfusion . 1996;36:873–878
- . Data-driven blood banking practices for neonatal RBC transfusions . Transfusion . 2000;40:1528–1540
The author of this editorial is also the lead author of one of the two major papers discussed in the editorial.
PII: S0022-3476(06)00560-9
doi:10.1016/j.jpeds.2006.06.033
© 2006 Mosby, Inc. All rights reserved.
Refers to article:
- The premature infants in need of transfusion (pint) study: A randomized, controlled trial of a restrictive (LOW) versus liberal (HIGH) transfusion threshold for extremely low birth weight infants
