Fergusson D, Hébert PC, Lee SK, Walker CR, Barrington KJ, Joseph L, Blajchman MA, Shapiro S. Clinical Outcomes Following Institution of Universal Leukoreduction of Blood Transfusions for Premature Infants. JAMA. 2003;289(15):1950-1956. doi:10.1001/jama.289.15.1950
Author Affiliations: Departments of Epidemiology and Biostatistics (Mr Fergusson and Drs Joseph and Shapiro) and Pediatrics (Dr Barrington), McGill University, Montreal, Quebec; University of Ottawa Centre for Transfusion Research, Ottawa Hospital (Mr Fergusson and Dr Hèbert), Clinical Epidemiology Unit, Ottawa Health Research Institute (Mr Fergusson and Dr Hébert), and Department of Pediatrics, University of Ottawa (Dr Walker), Ottawa, Ontario; Department of Pediatrics, University of British Columbia, Vancouver (Dr Lee); and Department of Pathology, McMaster University Medical Center, and Canadian Blood Services, Hamilton, Ontario (Dr Blajchman).
Context Leukocytes present in stored blood products can have a variety of biological
effects, including depression of immune function, thereby increasing nosocomial
infections and possibly resulting in organ failure and death. Premature infants,
given their immature immune state, may be uniquely predisposed to the effects
of transfused leukocytes.
Objective To evaluate the clinical outcomes following implementation of a universal
prestorage red blood cell (RBC) leukoreduction program in premature infants
admitted to neonatal intensive care units (NICUs).
Design and Setting Retrospective before-and-after study conducted in 3 Canadian tertiary
care NICUs from January 1998 to December 2000.
Patients A total of 515 premature infants weighing less than 1250 g who were
admitted to the NICU, received at least 1 RBC transfusion, and survived at
least 48 hours were enrolled. The intervention group consisted of infants
admitted in the 18-month period following the introduction of universal leukoreduction
(n = 247) and the control group consisted of infants admitted during the 18
months prior to the introduction of universal leukoreduction (n = 268).
Main Outcome Measures Primary outcomes were nosocomial bacteremia and NICU mortality, compared
before and after implementation of universal leukoreduction using multivariate
regression. Secondary outcomes included bronchopulmonary dysplasia, retinopathy
of prematurity, necrotizing enterocolitis, and intraventricular hemorrhage.
Results The proportion of infants who acquired bacteremia after an RBC transfusion
was 79/267 (29.6%) in the nonleukoreduction period and 63/246 (25.6%) in the
leukoreduction period. For NICU mortality, there were 45 deaths (16.8%) in
the nonleukoreduction period and 44 deaths (17.8%) in the leukoreduction period.
The adjusted odds ratio (OR) for bacteremia was 0.59 (95% confidence interval
[CI], 0.34-1.01) and for mortality was 1.22 (95% CI, 0.59-2.50). The adjusted
ORs for bronchopulmonary dysplasia and retinopathy of prematurity were 0.42
(95% CI, 0.25-0.70) and 0.56 (95% CI, 0.33-0.93), respectively. The adjusted
ORs for necrotizing enterocolitis and grade 3 or 4 intraventricular hemorrhage
were 0.39 (95% CI, 0.17-0.90) and 0.65 (95% CI, 0.35-1.19), respectively.
The adjusted OR for a composite measure of any major neonatal morbidity was
0.31 (95% CI, 0.17-0.56). Crude and adjusted rates for all secondary outcomes
suggest that leukoreduction was associated with improved outcomes.
Conclusion Implementation of universal prestorage leukoreduction was not associated
with significant reductions in NICU mortality or bacteremia but was associated
with improvement in several clinical outcomes in premature infants requiring
More than 50% of infants weighing less than 1250 g at birth and who
are admitted to neonatal intensive care units (NICUs) require red blood cell
(RBC) transfusions.1 Despite recent trends
in decreasing transfusion thresholds and the development of technologies designed
to avoid exposure to blood, such as erythropoietin, transfusions remain an
important lifesaving measure in the care of premature infants.
Leukocytes present in RBC transfusions may depress immune function,
thereby increasing the presence of nosocomial infections and possibly resulting
in organ failure and death.2,3 However,
randomized trials, conducted exclusively in adults in a variety of surgical
settings, have not all observed increased rates of postoperative nosocomial
infections.4,5 In addition, there
is a paucity of studies examining the possible risks and benefits of leukoreduction
of RBC products in premature infants.6 Given
their immature immune systems, neonates may be uniquely predisposed to the
effects of transfused leukocytes. Leukocytes from transfusions may depress
the immune response, generate alloantibodies, and, perhaps, produce widespread
microvascular injury through the enhanced generation of free radicals in susceptible
tissue beds such as the lungs and retina.2 Furthermore,
it is conceivable that the leukoreduction process increases RBC hemolysis
or results in unexpected clinical consequences.7 Based
on the available research, the benefits of leukoreduction remain unclear in
this vulnerable population.
We therefore evaluated clinical outcomes following the institution of
a national universal prestorage leukoreduction program among premature infants
weighing less than 1250 g who were admitted to a NICU.
The implementation of a nationwide Canadian universal prestorage leukoreduction
program for RBC products in 1999 enabled us to conduct a before-and-after
study using 3 study sites from the Canadian Neonatal Network. The study was
conducted at 3 tertiary care NICUs: Children and Women's Health Centre in
Vancouver, British Columbia; Royal University Hospital in Saskatoon, Saskatchewan;
and Mt Sinai Hospital in Toronto, Ontario.
The intervention group consisted of patients admitted to the NICU in
the 18-month period following the introduction of leukoreduction. The control
group, the nonleukoreduction cohort, consisted of all eligible infants admitted
to the NICU in the 18-month period preceding the introduction of universal
All premature infants from the 3 NICUs who weighed less than 1250 g,
received at least 1 allogeneic RBC transfusion, and survived at least 48 hours
were included. Infants surviving less than 48 hours were excluded to remove
those with an extremely poor prognosis, such as overwhelming infections acquired
from their mothers. Infants were also excluded if they were previously admitted
to the NICU or received both leukoreduced and nonleukoreduced RBC transfusions.
The research ethics committees of all participating institutions approved
the study protocol.
Information gathered on each admission from the Canadian Neonatal Network
database included the patient's gestational age and sex, illness severity
scores including the SNAP II (Score for Neonatal Acute Physiology) on days
1 and 3, and Apgar scores at 1 and 5 minutes. Important interventions were
recorded, including use of supplemental oxygen, continuous positive airway
pressure (CPAP) and mechanical ventilation, central and peripheral venous
access, RBC and platelet transfusions, use of intravenous immunoglobulin,
cardiopulmonary resuscitation, and number of blood draws. The use of medications
such as vasoactive drugs, antibiotics, surfactants, and corticosteroids was
also recorded. Recorded maternal risk factors included type of delivery (cesarean
vs vaginal) and use of any antenatal steroids. Information on the number,
date, and volume of RBCs transfused was collected from the Canadian Neonatal
Network database for 1 site (Royal University Hospital) and from the respective
blood banks for the 2 other sites.
Data abstractors for the Canadian Neonatal Network database were responsible
for collecting standardized information from every eligible admission to the
NICU. Infants who were moribund at admission were not included. Data were
gathered from medical records during admission to the NICU and transcribed
directly into computerized case report forms. Standardized definitions were
instituted to ensure consistency among sites. Any data items not available
were scored as missing.
Canadian Blood Services introduced universal prestorage leukoreduction
throughout Canada during 1999. Once collected, RBCs are passed through a Leukotrap-RC
leukocyte reduction filtration system (Pall Corp, Port Washington, NY). Leukoreduction
with this technology reduces white blood cell content of a unit of RBCs from
an average 3.0 × 109 per unit to 2.5 × 105 per
unit, a decrease of 4 logarithms. An anticoagulant solution of citrate, phosphate,
and double dextrose (CPD-2) is then added to each unit with 100 mL of additive
solution 3 (AS-3). All blood products were produced by 1 agency that conducts
national quality control measures of the leukofiltration program.8 Prior to transfusion, each unit is divided into aliquots
suitable for the neonatal population.
Transfused RBCs were not leukofiltrated in the control period at any
site. One site (Children and Women's Health Centre) washed RBC transfusions
designated for premature infants during the entire nonleukoreduction period
and during the first 4 months of the leukoreduction period. Washing of RBCs
reduces the proportion of leukocytes by 85% compared with a reduction of greater
than 99.9% for prestorage leukoreduction. This site was included because it
(1) had not used prestorage leukoreduction in the nonleukoreduction period;
(2) provided a large sample of patients; (3) made our results more generalizable
to include centers that currently wash RBCs; and (4) would not unfavorably
bias the results of our study. In fact, including this site would only bias
our results toward the null if washing RBCs is beneficially associated with
our outcomes of interest.
The primary outcome measures in this study were nosocomial bacteremia
and NICU mortality. Nosocomial bacteremia was defined as a positive blood
culture for bacteria whether or not it was subsequently considered to be a
contaminant. For a new infection, there had to be a new organism or a second
positive blood culture drawn at least 7 days after the initial positive test.
Survival status during the index hospitalization was ascertained from medical
Major secondary outcomes included bronchopulmonary dysplasia, retinopathy
of prematurity, necrotizing enterocolitis, and intraventricular hemorrhage.
Bronchopulmonary dysplasia was defined as the ongoing need for assisted ventilation
or supplemental oxygen on day 28 of life. The presence of any grade of retinopathy
was recorded as an outcome in this study. Without documentation of an eye
examination, retinopathy was defined as being absent. A diagnosis of necrotizing
enterocolitis was based on a grade of stage 2 or higher using the criteria
of Bell et al.9 A diagnosis of intraventricular
hemorrhage required the presence of intraventricular blood on a routine image
such as an ultrasound, computed tomography, or magnetic resonance image of
the brain. Intraventricular hemorrhage was graded based on standard criteria
developed by Papile.10 For the purposes of
this study, a composite of grades 3 and 4 intraventricular hemorrhage was
included as an outcome. As a means of evaluating the overall effect of leukoreduction
on multiple organs simultaneously, all secondary outcomes were compared both
separately and as a composite measure.
Length of stay in the NICU and both minor and major interventions received
while in the NICU were recorded as tertiary outcomes. Major interventions
included all major surgical procedures, such as laparotomies and thoracotomies.
Minor interventions included cryogenic or laser therapy for retinopathy of
prematurity, tracheostomy, endoscopic procedures such as bronchoscopy, and
all transcutaneous procedures, such as nephrostomy and cardiac catheterizations.
Use of umbilical vein and artery lines, peripheral arterial lines, venous
cutdowns, and needle aspiration of body fluids were excluded from this category.
Information was also recorded that reflected the intensity of care on
day 1 provided to each premature infant. On day 1, use of supplemental oxygen,
conventional and high-frequency mechanical ventilation, and arterial and central
venous access, as well as the use of medications such as vasopressors, glucocorticoids,
antibiotics, and muscle relaxants, were recorded. Supplemental oxygen was
defined as administration of continuous enriched oxygen in concentrations
exceeding 21% via oxygen hood, nasal cannula, nasal catheter, or face mask
or other forms of respiratory support. The use of "blow-by" oxygen alone was
not sufficient to meet the definition, nor was oxygen administered for a hyperoxia
test. Mechanical ventilation was defined as use of conventional mechanical
ventilation regardless of the respiratory rate. We also recorded use of high-frequency
ventilation using a jet ventilator or oscillator.
Peripheral intravenous access was defined as the presence of 1 or more
intravenous catheters, including heparin locks used for drug administration.
An arterial line was defined as the presence of a central line, including
an umbilical venous line, a Broviac line, or a percutaneous catheter placed
centrally. Unsuccessful attempts at line placement were not reported.
The use of vasopressors was defined as the administration of vasoactive
medications administered through intravenous, intramuscular, or aerosol routes.
Glucocorticoid and antibiotic use on day 1 was documented if an intravenous,
oral, or nebulized preparation was used. Use of muscle relaxants was recorded
daily if at least 1 dose of the medication was administered during the interval
Baseline characteristics before and after the introduction of leukoreduction
were evaluated using measures of central tendency and dispersion. Absolute
differences between periods were calculated for each characteristic with appropriate
95% confidence intervals (CIs). All a priori primary outcomes were compared
using crude and adjusted odds ratios (ORs) with 95% CIs. Crude and adjusted
ORs were also calculated independently for secondary outcomes, including bronchopulmonary
dysplasia, retinopathy of prematurity, necrotizing enterocolitis, and intraventricular
As a second step, we incorporated all secondary outcomes associated
with prematurity into a composite measure. Lengths of NICU stay were compared
using multivariate regression analysis. Based on a priori input of experts
in neonatology and transfusion medicine, all multivariate models incorporated
clinically important variables, including gestational age, sex, center, type
of delivery, antenatal use of glucocorticoids, Apgar score at 5 minutes, SNAP
II as a measure of illness severity, number of days receiving CPAP, interventions
on day 1 including use of supplemental oxygen and mechanical ventilation,
and medications used on day 1 including glucocorticoids or surfactants, vasopressors,
muscle relaxants, and antibiotics.
Adjusted ORs were also calculated for infants who received surfactants
on day 1 vs those who did not as well as for infants who were administered
CPAP and mechanical ventilation on day 1. To better understand the relationship
between mortality and secondary complications of prematurity, we compared
baseline characteristics stratified by survival status and leukoreduction.
As a second step, we compared all outcomes by study period in patients who
died. Causes of death also were compared between infants receiving leukoreduced
and nonleukoreduced RBC transfusions.
In reporting our results, an OR less than 1 suggests that fewer infants
in the leukoreduction group were affected by that outcome, while an OR greater
than 1 suggests that fewer infants in the nonleukoreduction group were affected
by that outcome. Measures of effect for multivariate linear regression were
expressed as number of days of NICU stay saved with 95% CIs. Missing data
for 3 variables (birth weight [3 imputations], Apgar score at 5 minutes [14
imputations], and SNAP II on day 1 [48 imputations]) were estimated using
the multivariate normal procedure.11 The outcome
estimates for the models with and without imputed data were comparable, so
we report the results from the imputation model only. Analyses were carried
out using NCSS statistical software, 2000 (Kaysville, Utah).
A total of 516 premature infants weighing less than 1250 g were identified
from the 3 sites: 237 (107 before and 130 after universal leukoreduction)
from Children and Women's Health Centre, 54 (38 before and 16 after universal
leukoreduction) from Royal University Hospital, and 225 (124 before and 101
after universal leukoreduction) from Mt Sinai Hospital. One infant from Mt
Sinai was removed because the patient received both nonleukoreduced and leukoreduced
RBCs. Thus, a total of 515 neonates who underwent transfusion were included
in the analysis; 268 infants received nonleukoreduced RBC transfusions and
247 received leukoreduced RBC transfusions.
All baseline characteristics except for some respiratory interventions
were comparable in infants in the nonleukoreduction and leukoreduction periods
(Table 1). More infants in the
nonleukoreduction group required mechanical ventilation (89.7% vs 81.3%),
while fewer patients in this group received supplemental oxygen (77.4% vs
84.6%), CPAP (16.7% vs 51.3%), high-frequency ventilation (4.8% vs 9.2%),
and use of surfactants (53.6% vs 67.5%) compared with premature infants receiving
leukoreduced RBC transfusions.
Crude and adjusted rates of nosocomial bacteremia and all separately
assessed secondary outcomes except grade 3 or 4 intraventricular hemorrhages
were lower in infants who received leukoreduced RBCs (Table 2). The results were not significantly different for NICU
mortality (adjusted OR, 1.22; 95% CI, 0.59-2.50). When bronchopulmonary dysplasia,
retinopathy of prematurity, necrotizing enterocolitis, and intraventricular
hemorrhage were considered together as a composite outcome, the adjusted OR
was 0.31 (95% CI, 0.17-0.56). The adjusted length of NICU stay was decreased
by 11.62 days (95% CI, 3.61-19.64) in infants who received leukoreduced RBCs.
Results of the unadjusted subgroup analyses for all major outcomes also
suggest a consistent beneficial effect of leukoreduction in infants who did
not receive surfactants, mechanical ventilation, or CPAP on day 1, except
for those who did not receive mechanical ventilation and experienced bronchopulmonary
dysplasia, and those who received surfactants and experienced bacteremia or
grade 3 or 4 intraventricular hemorrhage (Table 3). The CIs for the former 3 associations all included 1.
A post hoc analysis was conducted to examine the influence of leukoreduction
on major neonatal morbidities in infants who had died. The mean (SD) NICU
length of stay was 18.5 (5.3) days for nonsurvivors vs 85.9 (3.6) days for
those who survived. Leukoreduction had an apparent protective but non–statistically
significant effect on bacteremia in nonsurvivors (35.6% among nonleukoreduction
vs 20.5% in the leukoreduction group [OR, 0.47; 95% CI, 0.17-1.22]. There
was no apparent association between leukoreduction and bronchopulmonary dysplasia
or grade 3 or 4 intraventricular hemorrhage in infants who died (OR, 1.03;
95% CI, 0.38-2.78 and OR, 1.06; 95% CI, 0.45-2.50, respectively). There were
too few cases of necrotizing enterocolitis (n = 11) and retinopathy of prematurity
(n = 4) to comment on associations.
Since Children and Women's Health Centre washed RBCs prior to the implementation
of universal prestorage leukoreduction, we conducted a post hoc analysis to
examine the influence of that site on our results. When data from this site
were removed from the analyses, the unadjusted OR for bacteremia was 0.47
(95% CI, 0.23-0.99) and for mortality was 1.43 (95% CI, 0.76-2.68), whereas
the ORs for all of the secondary outcomes except necrotizing enterocolitis
(OR, 0.64; 95% CI, 0.29-1.42) shifted farther away from the null.
This study demonstrates that the implementation of universal prestorage
leukoreduction of allogeneic RBC transfusions was not associated with reduced
NICU mortality in premature infants weighing less than 1250 g but was associated
with improvements in all major secondary outcomes. Indeed, if these secondary
outcomes are considered as a composite outcome, for each 10 premature infants
undergoing transfusion with leukoreduced RBCs, leukoreduction was associated
with the prevention of 1 major secondary complication of premature birth.
This clinical benefit was accompanied by an average decrease of 11 days of
To examine secular trends, we analyzed data for infants who did not
receive any RBC transfusions. In infants weighing less than 1250 g who did
not undergo transfusion, unadjusted rates of bacteremia, mortality, bronchopulmonary
dysplasia, retinopathy of prematurity, and intraventricular hemorrhage were
all increased in the leukoreduction period compared with the nonleukoreduction
period (Table 4). However, the
rate of necrotizing enterocolitis and the length of stay in the NICU decreased
in the leukoreduction period compared with the nonleukoreduction period. Given
the increase in rates for 5 of the 7 outcomes, it does not appear that our
observed results are confounded by changes in care.
Our study demonstrates little about the relationship between leukoreduction
and mortality except that it essentially rules out differences greater than
50% in either direction. Post hoc examination revealed that 50 (56%) of the
89 deaths occurred within the first 14 days and 73 (82%) occurred within the
first 29 days. Thus, as with many interventions, the benefit of leukoreduction
in this population is doubtful. For infants who survive longer than 1 month,
we would have required a much greater sample size to detect differences in
mortality because the probability of death is very low.
To date, we are unaware of any other published study that has examined
the association between leukoreduction of allogeneic RBC products and major
clinical outcomes in premature neonates, including bacteremia, mortality,
or other major complications of prematurity, such as bronchopulmonary dysplasia,
retinopathy of prematurity, necrotizing enterocolitis, and intraventricular
hemorrhage.6 A few studies have shown that
RBC transfusions correlate with an increased risk of bronchopulmonary dysplasia
and retinopathy of prematurity.12- 16 However,
the specific role of leukocytes and leukoreduction in the pathogenesis of
these conditions had not been assessed. If transfused leukocytes result in
or increase the severity of such complications, then this may be mediated
by the enhanced generation of free radicals or caused by some other poorly
defined effect either on the immune system or the microvasculature.17- 19 The beneficial effect
of leukoreduction of RBC products on several organ systems simultaneously
suggests that the putative role of leukoreduction is having a widespread effect
throughout the body in many organ systems.
There are potential sources of bias in this study, particularly in the
sampling of patients. The greatest threat to the observed effects is the possibility
of secular trends. There was evidence of important changes in respiratory
management in the 36-month period of study. Use of surfactants increased during
this time, as did use of CPAP as a mode of ventilatory support compared with
other modes of mechanical ventilation. These secular trends were likely to
be most evident in the development of bronchopulmonary dysplasia. However,
multivariate and stratified analyses consistently demonstrated a beneficial
decrease in the odds of bronchopulmonary dysplasia with or without the use
of surfactants (Table 3). Similarly,
the effects of different ventilatory strategies appeared unrelated to any
of the outcomes except for the unadjusted association between mechanical ventilation
and bronchopulmonary dysplasia. The magnitude and consistency of the findings
among all patient groups in all major outcomes strengthen our conclusions.
One of the major advantages of this study is that our cohort of premature
infants included all consecutive admissions, both with and without transfusion,
during both periods. This patient sampling strategy enabled us to add an extra
level of control in comparing the risks and benefits of leukoreduction in
this population. Additionally, both periods were 18 months in duration and,
thus, reasonably short and symmetrical. This duration minimizes seasonal variation
in the patterns of admission or patient care. Also, the use of multivariate
analyses enabled us to control for the confounding influence of a number of
Other design choices may have limited the inferences drawn from the
data. The relatively small sample size did not allow us to detect meaningful
clinically important differences in the rates of mortality and bacteremia,
if truly present. Indeed, the 95% CIs are quite wide, given the sample size
of 515 infants who underwent transfusion. Also, in limiting our choice of
index nosocomial infections solely to the presence of bacteremia, we may have
missed other important immunomodulating effects of leukoreduction. Because
of resource constraints, one of the additional limitations of our study was
our inability to document the dates of diagnoses of major complications such
as retinopathy of prematurity and intraventricular hemorrhage. Therefore,
we remain uncertain about the time and total exposure to RBC transfusions
in these infants prior to their diagnosis. However, we believe that this information
would not significantly change our results because each of the complications
occurs after prolonged stays in a NICU. More than 50% of infants received
an RBC transfusion within the first 5 days of NICU admission and greater than
75% did within the first 15 days. The issue of timing was not a concern for
other major outcomes, including mortality, bacteremia, and bronchopulmonary
dysplasia, for which accurate estimates of dates were available.
Despite these limitations, the present study has a number of strengths.
First is the methodological quality of data abstraction and entry by the Canadian
Neonatal Network investigators, who used trained data abstractors, a set of
clear definitions, and rigorous data checks and follow-up.20- 22 Second,
the study included a consecutive census of premature infants weighing less
than 1250 g admitted to 3 NICUs representing 3 different geographic regions
In conclusion, implementation of universal prestorage leukoreduction
is associated with improved clinical outcomes in premature infants requiring
allogeneic RBC transfusions as part of their care in the NICU, even though
there were no significant reductions in NICU mortality or bacteremia. Until
there is evidence of harm, we would recommend the adoption of universal leukoreduction
in the care of all infants requiring RBC transfusions. While we believe these
data are persuasive, we would appeal for and endorse the conduct of a large
randomized controlled trial to definitively determine the effectiveness of
prestorage leukoreduction in the neonatal population as well as laboratory
and clinical studies to elucidate the mechanisms of action of such effects.