Original Article

Transfus Med Hemother 2016;43:297–301 Received: October 22, 2014

DOI: 10.1159/000446253 Accepted: December 10, 2015
Published online: May 23, 2016

Hemoglobin Threshold for Blood Transfusion in a

Pediatric Intensive Care Unit
Madhuradhar Chegondi a Jun Sasaki a André Raszynski a,b Balagangadhar R. Totapally a,b
a
Division of Critical Care Medicine and Nicklaus Children’s Hospital (Formerly Miami Children’s Hospital), Miami, FL, USA;
b Herberth Wertheim College of Medicine, Florida International University, Miami, FL, USA

Keywords old for stable and unstable patients among all groups
Hemoglobin · Threshold · Packed red blood cell · was 7.3 ± 1.3 and 7.9 ± 1.3 (p < 0.0001), respectively. The
Transfusion · Children · PICU observed mortality rate was higher among children who
received transfusion compared to other children admit-
Summary ted to PICU. Conclusion: The hemoglobin threshold for
Objective: To evaluate the hemoglobin threshold for transfusion varied according to clinical conditions. Over-
red cell transfusion in children admitted to a pediatric in- all, the hemoglobin threshold for transfusion was 7.3 ±
tensive care unit (PICU). Methods: Retrospective chart 1.20 g/dl.
review study. Tertiary care PICU. Critically ill pediatric © 2016 S. Karger GmbH, Freiburg
patients requiring blood transfusion. No intervention.
Results: We analyzed the charts of all children between
1 month and 21 years of age who received packed red
blood cell (PRBC) transfusions during a 2-year period.
The target patients were identified from our blood bank Introduction
database. For analysis, the patients were subdivided into
four groups: acute blood loss (postsurgically, trauma, or Transfusion of blood products, especially of packed red blood
acute gastrointestinal bleeding from other causes), he- cells (PRBCs), is an important aspect of care in critically ill children.
matologic (hematologic malignancies, bone marrow Clinical indications for blood transfusions in children and adults
suppression, hemolytic anemia, or sickle cell disease), are similar, but hemoglobin threshold, transfusion volume in rela-
unstable (FiO2 > 0.6 and/or on inotropic support), and tion to body weight, and infusion rates are not well delineated.
stable groups. We also compared the pre–transfusion Children admitted to a pediatric intensive care unit (PICU)
hemoglobin threshold in all unstable patients with that have an increased risk of developing anemia due to the underlying
of all stable patients. A total of 571 transfusion episodes etiology and illness severity, iatrogenic blood loss, hemodilution,
in 284 patients were analyzed. 28% (n = 160) of transfu- poor nutrition, and blunted bone marrow response [1–4]. A large,
sions were administered to patients in the acute blood prospective, multicenter, observational study [5] reported an inci-
loss group, 36% (n = 206) to hematologic patients, 17% dence rate of 74% of anemia in PICU patients; 33% of patients had
(n = 99) to unstable patients, and 18% (n = 106) to stable anemia on admission. Studies from Kenya have shown a signifi-
patients. The mean pre-transfusion hemoglobin (± SD) in cantly increased mortality in children with a hemoglobin level of
all children as well as in the acute blood loss, hemato- less than 5 g/dl and a decreased mortality after transfusion [6–8].
logic, unstable and stable groups was 7.3 ± 1.20, 7.83 ± Transfusion of PRBCs is common in the PICU. The overall inci-
1.32, 6.97 ± 1.31, 7.96 ± 1.37, 7.31 ± 1.09 g/dl, respec- dence of transfusion has been reported as 17% of all PICU patients
tively. The transfusion threshold for acute blood loss and and as up to 50% in children with a stay longer than 48 hours [9].
unstable groups was higher compared to hematologic However, PRBC transfusions may result in multiple adverse effects
and stable groups (p < 0.001; ANOVA with multiple com- including increased mortality, transfusion reactions, volume over-
parisons). The mean pre-transfusion hemoglobin thresh- load, infections, and immunosuppression [10].

© 2016 S. Karger GmbH, Freiburg Madhuradhar Chegondi, MD

1660–3796/16/0434–0297$39.50/0 Division of Critical Care Medicine
Fax +49 761 4 52 07 14 Nemours Children’s Hospital
[email protected] Accessible online at: 13535 Nemours Parkway, Orlando, FL 32827, USA
www.karger.com www.karger.com/tmh [email protected]
 In order to decrease the risk of transfusion complications, the Each child’s pre- and post-transfusion clinical variables, diagnosis, and de-
American Association of Blood Banks recommends a restrictive mographics (including age, sex, weight, length of stay, and mortality) were
documented. Transfusion variables collected included pre-transfusion and
transfusion threshold of 7–8 g/dl in hospitalized and stable patients
post-transfusion hemoglobin concentrations and volume of PRBC infused.
and suggests adhering to a restrictive strategy in hospitalized pa- Each transfusion episode was counted in the total number of transfusions if the
tients with preexisting cardiovascular disease. A hemoglobin hemoglobin level was measured before the transfusion. Multiple transfusions
threshold of 8 g/dl or less is recommended for transfusion if pa- with no pre-transfusion hemoglobin values were considered as a single transfu-
tients are symptomatic [11]. sion episode.

There is no agreed hemoglobin level for PRBC transfusions in

Statistical Analysis
children admitted to a PICU. The threshold for transfusion may Descriptive data are presented as mean ± SD for parametric continuous
vary with underlying diagnosis and physiologic stability. In a retro- data, median ± interquartile range (IQR) for nonparametric data. Nonparamet-
spective survey of five PICUs in the USA [12], nearly 55% of all ric ANOVA with Bonferroni multiple comparison tests was used to compare
PICU patients with pre-transfusion hemoglobin levels of less than the means of all variables in four groups. Mann-Whitney U test was used to
compare pre-transfusion hemoglobin levels in all stable versus all unstable
9 g/dl received one or more PRBC transfusions. Despite the publi-
groups. Chi-square or Fisher’s exact test was used to analyze binary data. Paired
cation of the Transfusion Requirements in Critical Care in 1999 t-test was used to compare pre- and post-transfusion hemoglobin values. A p
[13], there remains great variability in the pediatric critical care value < 0.05 was considered significant.
specialists practice with respect to the threshold hemoglobin level
for PRBC transfusion [14]. An overly restrictive transfusion strat-
egy may increase the risk of tissue hypoxia and a too liberal trans- Results
fusion strategy may lead to an increased incidence of adverse ef-
fects. A survey of pediatric critical care specialists from Canada and A total of 571 transfusion episodes in 284 patients were ana-
Europe regarding PRBC transfusion practices revealed a wide lyzed. Nearly 9% of all patients admitted to our PICU received a
range of hemoglobin transfusion thresholds (7–13 gm/dl) and of transfusion. 28% (n = 160) of transfusions were administered to
PRBC transfusion volumes [14, 15]. In addition, the threshold patients in the acute blood loss group, 36% (n = 206) to hemato-
hemoglobin levels for transfusions varied with the underlying logic patients, 17% (n = 99) to unstable patients, and 18% (n = 106)
medical or surgical conditions [15]. Our study is the first to com- to stable patients.
pare the PRBC transfusion thresholds among four groups of PICU The mean age of our patient population was 6.98 ± 6.78 years.
patients with different clinical conditions. The mean age in the acute blood loss and hematological groups
The aim of our study was to evaluate the hemoglobin threshold was significantly different from that in the unstable and stable
for red cell transfusion among four clinical groups in children ad- groups (table  1). The median PICU length of stay of all clinical
mitted to a PICU. groups was 22 days (IQR 10–45 days). The median PICU length of
stay of the acute blood loss and hematological groups was signifi-
cantly different compared to that of the unstable and stable groups
Material and Methods (p < 0.001) (table 1).

Our tertiary care PICU with 28 beds available, has on average 1,600 admis- Pre-Transfusion Hemoglobin
sions per year and cares for all critically ill children, except for those after car- The mean pre-transfusion hemoglobin value of our study popu-
diac surgery. After obtaining institutional review board approval, the charts of
all children who were admitted to our PICU during a 2-year-period and who
lation was 7.3 ± 1.20 g/dl. The mean pre-transfusion hemoglobin for
received PRBC transfusions were retrieved. The target patient population data acute blood loss patients was 7.83 ± 1.32 g/dl, for hematologic pa-
were identified and retrieved from our blood bank database. Any blood transfu- tients 6.97 ± 1.31 g/dl, for unstable patients 7.96 ± 1.37 g/dl, and for
sions outside of the PICU (e.g. operating room, emergency room, medical and stable patients 7.31 ± 1.09 g/dl. The mean pre-transfusion hemo-
surgical floor) were not included. We excluded patients on ECMO and hemodi- globin values in patients with acute blood loss and unstable patients
alysis and patients who received an exchange transfusion.
Patients’ personal identifiers (including unique patient characteristics) were
were significantly higher than those of hematological and stable pa-
removed from the data prior to analysis. For analysis, patients were subdivided tients (p < 0.05) (table  1). The mean pre-transfusion hemoglobin
into four groups: acute blood loss, hematologic, unstable, and stable groups. thresholds for all stable and all unstable patients were 7.3 ± 1.3 and
The acute blood loss group included patients who received transfusions after 7.9 ± 1.3, respectively (p < 0.0001). The numbers of transfusion epi-
surgery or for overt internal or external bleeding. The hematologic group in- sodes below 7 g/dl and above 8 g/dl were 37 (37/160 = 23.1%) and 62
cluded patients with Hemoglobinopathy (mainly sickle cell disease), leukemia,
lymphoma, solid tumors, bone marrow suppression, post bone marrow trans-
(62/160 = 38.7%) in the acute blood loss group, 75 (75/206 = 36.4%)
plantation, and patients who were receiving chemotherapy. The unstable group and 29 (29/206 = 14%) in the hematologic group, for patients 17
included children with FiO2 requirement over 0.6 and patients who were re- (17/99 = 17.2%) and 41(41/99 = 41.4%) in the unstable group, and
ceiving inotropic support and were not assignable to acute blood loss or hema- 35 (35/106 = 33%) and 16 (16/106 = 15.1%) in the stable group.
tologic groups. The stable group included patients who did not meet the criteria
of the other groups. In a separate analysis, all unstable patients (any child with
FiO2 requirement above 0.6 and/or reception of inotropic support irrespective
Volume of PRBC
of the underlying condition) were compared with all stable patients irrespective The mean PRBC volume transfused for all patients was 11.52 ±
of the cause of anemia. 9.94 ml/kg. In acute blood loss patients, the mean PRBC volume

298 Transfus Med Hemother 2016;43:297–301 Chegondi/Sasaki/Raszynski/Totapally

Table 1. Demographic, hematologic, and transfusion variables in all children and children in various groups

All children Acute blood loss Hematologic Unstable Stable P value

Number of children 284 110 72 46 56

Number of transfusions 571 160 206 99 106

Male, % 56.69 40.99 22.98 13.66 17.39

Mean age, years 6.98 ± 6.78 8.85 ± 6.76* 9 ± 6.82* 3.35 ± 5.05 3.62 ± 5.34 <0.001

Median length of stay, days (range) 22 (10–45) 12 (6–30)* 17.5 (9.75–33.2)* 43 (23.5–95) 38.5 (20–64) <0.001

Pre-transfusion Hb, g/dl <0.001

Mean ± SD 7.3 ± 1.20 7.83 ± 1.32** 6.97 ± 1.31 7.96 ± 1.37** 7.31 ± 1.09
Min./max. 2.4/12.4 3.4/11.8 2.4/10.3 3.7/12.4 4.6/10.2

Mean PRBC transfusion volume, ml/kg 11.52 ± 9.94 13.08 ± 11.34*** 9.79 ± 9.23*** 12.05 ± 7.05 12.10 ± 11.13 <0.05

Post-transfusion Hb, g/dl <0.001

Mean ± SD 9.83 ± 1.97 10.31 ± 1.59 9.04 ± 1.73**** 10.39 ± 1.80 10.11 ± 1.67
Min./max. 4.4/7.1 7.1/14.8 4.4/14.3 6.7/15.6 6.1/13.6

*Acute blood loss and hematologic groups different from unstable and stable groups.
**Acute blood loss and unstable groups different from hematologic and stable groups.
***Acute blood loss group different from hematologic group.
****Hematologic group different from other groups.

transfused was 13.08 ± 11.34 ml/kg, while it was 9.79 ± 9.23 ml/kg [17]. There is limited data from clinical trials comparing the out-
in hematological patients (p < 0.05) (table 1). comes with restrictive and liberal transfusion protocols. Data from
clinical trials on adults have shown improved patient outcomes
Post-Transfusion Hemoglobin with a restrictive transfusion strategy. A meta-analysis review of
After transfusion, the mean hemoglobin concentration in- adult patients [18] was able to identify 10 randomized clinical trials
creased from 7.3 ± 1.20 to 9.83 ± 1.97 g/dl in all patients (p < 0.001; where different PRBC transfusion triggers were analyzed with ap-
paired t-test). The post-transfusion hemoglobin was lower in the propriate methodology. The transfusion thresholds evaluated in
hematologic group compared to other groups (p < 0.05) (table 1). these trials were between 7 and 10 g/dl, and the mortality was one
The mean post-transfusion hemoglobin thresholds of stable and fifth lower (RR 0.80; 95% CI 0.63–1.02) with restrictive transfusion
unstable patients among all groups were 9.7 ± 1.8 and 10.3 ± 1.7 regimens.
(p < 0.0001), respectively. A multicenter randomized clinical trial [19] studied liberal ver-
sus restrictive transfusion strategies in 100 preterm infants and
Mortality Rate showed that the restrictive red cell transfusion group had a higher
The mortality rate in the 284 transfused patients was 6.69% risk of intraparenchymal brain hemorrhage, periventricular leu-
(n = 19), with 4 in the acute blood loss, 5 in the hematologic, and komalacia, and apnea. Another clinical trial [20] in 451 premature
10 in the unstable group. As expected, the mortality was different infants showed that the mortality or severe morbidity rate was
among the four groups, with the highest mortality among unstable 2.6% higher in the restrictive group compared to the liberal trans-
patients (21.73%) and no mortality among stable patients (p < 0.001; fusion strategy group, but the difference was not significant. The
chi-square test with Yates correction) (table  1). The mortality Transfusion Stategies for Patients in Pediatric Intensive Care Units
among those who did not receive any transfusion was 1.4% (p < 0.001; (TRIPICU) study [16] found that in 637 stable patients in PICUs a
OR 5.6 (2.6–12.4); Fisher’s exact test). transfusion threshold of 7 g/dl rather than a liberal threshold of 9.5
g/dl reduced transfusion requirements by 44% (0.9 ± 2.6 vs. 1.7 ±
2.2 units/patient; p < 0.001). Although there were no differences in
Discussion mortality or multiorgan dysfunction rate between the two groups,
the TRIPICU study [16] recommended a restrictive transfusion
The mean hemoglobin threshold for transfusion in our unit was strategy in clinically stable patients in PICUs. This study does not
7.3 g/dl and differed according to the underlying clinical condition address the transfusion threshold specifically for unstable children
group. It was higher for acute blood loss and unstable patients and children with hemolytic anemia and active bleeding. In a re-
compared to hematologic and stable patients. Our threshold hemo- cent study on transfusion practice changes in PICUs over a decade
globin values for transfusion were similar to those of published and a significant decline in overall transfusions from 10.5% to 6.8%
data for stable patients [16] and post-surgical critically ill children [21] was found.

Hemoglobin Threshold for Blood Transfusion in Transfus Med Hemother 2016;43:297–301 299
a Pediatric Intensive Care Unit
 In a further analysis [17], PRBC transfusion thresholds in post- were children after craniofacial or spinal surgery. Thresholds for
surgical PICU patients were studied by examining a subgroup of transfusion in these patients are in general set by the surgical teams
124 postoperative patients of the original TRIPICU study patients. in our hospital. As discussed in the methodology, the present study
The study found similar outcomes compared to the overall TRI- did not include children after cardiovascular surgery.
PICU study population, and a transfusion threshold of 7 g/dl for In summary, the hemoglobin threshold for transfusion varied
stable post-surgical patients in PICUs was recommended. The according to the clinical condition in our PICU. Overall, the hemo-
threshold hemoglobin value for the acute blood loss group in our globin threshold for transfusion in our PICU was 7.3 ± 1.2 g/dl.
study was 7.8 g/dl. The acute blood loss group in our study mainly The threshold hemoglobin level for transfusion was higher for chil-
included patients who had experienced active bleeding either after dren with acute blood loss and unstable children. The mortality
surgery or from gastrointestinal tract. We included all actively was higher among children who required blood transfusion.
bleeding and post-surgical patients irrespective of hemodynamic
stability. The British Society of Hematology [22] also suggested a
postoperative hemoglobin level of 7 g/dl as transfusion threshold Study Limitations
in both children and adults with stable postoperative cardiac
function. This study is a retrospective single-center study. The findings
Another subgroup analysis of septic patients enrolled in the from one center may not apply to other centers with different pa-
TRIPICU study [23] found no evidence of an increased risk of ad- tient populations. The mortality data obtained allows us only to
verse outcome when comparing restrictive and liberal transfusion state an association between the severity of disease, transfusion
strategies. The data suggested that a hemoglobin level of 7 g/dl may probability and mortality, but no causal relation that blood trans-
be safe for stable septic patients in PICUs. A recent literature re- fusions are an independent risk factor of mortality can be
view on PRBC transfusion in PICU patients [24] showed that a assumed.
threshold hemoglobin level above 7 g/dl does not yield improved
outcomes; and several smaller studies suggested an increased risk
of morbidity and mortality in PICU patients exposed to blood Acknowledgements
transfusions.
We thank Dr. S. Melnick, director of the blood bank at Nicklaus Children’s
In our study we allocated all children who received transfusion
Hospital (formerly Miami Children’s Hospital) for providing the data.
into four groups. As expected, unstable patients, defined as patients
requiring 60% of O2 or more and/or were on vasoactive medica-
tions, received transfusions at a higher threshold level (7.96 ± 1.37
Disclosure Statement
g/dl). In addition, patients in the acute blood loss group also re-
ceived blood transfusion at a higher threshold value (7.83 ± 1.32 None.
g/dl). Most of the post-surgical children requiring transfusions

References
1 Shander A: Anemia in the critically ill. Crit Care Clin 8 Lackritz EM, Hightower AW, Zucker JR, Ruebush TK 13 Hébert PC, Wells G, Blajchman MA, Marshall J, Mar-
2004; 20: 159–178. 2nd, Onudi CO, Steketee RW, Were JB, Patrick E, tin C, Pagliarello G, Tweeddale M, Schweitzer I, Yetisir
2 Scharte M, Fink MP: Red blood cell physiology in criti- Campbell CC: Longitudinal evaluation of severely ane- E; Transfusion Requirements in Critical Care Investi-
cal illness. Crit Care Clin 2003; 31(12 suppl):S651–657. mic children in Kenya: the effect of transfusion on gators, Canadian Critical Care Trials Group: A multi-
3 Krafte-Jacobs B, Levetown ML, Bray GL, Ruttimann mortality and hematologic recovery. Aids 1997; 11: center, randomized, controlled clinical trial of transfu-
UE, Pollack MM: Erythropoietin response to critical 1487–1497. sion requirements in critical care. N Engl J Med 1999;
illness. Crit Care Clin 1994; 22: 821–826. 9 Armano R, Gauvin F, Ducruet T, Lacroix J: Determi- 340: 409–417.
4 Rogiers P, Zhang H, Leeman M, Nagler J, Neels H, nants of red blood cell transfusions in a pediatric criti- 14 Laverdière C, Gauvin F, Hébert PC, Infante-Rivard C,
Mélot C, Vincent JL: Erythropoietin response is cal care unit: a prospective, descriptive epidemiological Hume H, Toledano BJ, Guertin MC, Lacroix J; Cana-
blunted in critically ill patients. Intensive Care Med study. Crit Care Med 2005; 33: 2637–2644. dian Critical Care Trials Group: Survey on transfusion
1997; 23: 159–162. 10 Dellinger EP, Anaya DA: Infectious and immunologic practices of pediatric intensivists. Pediatr Crit Care
5 Bateman ST, Lacroix J, Boven K, Forbes P, Barton R, consequences of blood transfusion. Crit Care 2004; Med 2002; 3: 335–340.
Thomas NJ, Jacobs B, Markovitz B, Goldstein B, Han- 8(suppl 2):S18–23. 15 Nahum E, Ben-Ari J, Schonfeld T: Blood transfusion
son JH, Li HA, Randolph AG; Pediatric Acute Lung 11 Carson JL, Grossman BJ, Kleinman S, Tinmouth AT, policy among European pediatric intensive care physi-
Injury and Sepsis Investigators Network: Anemia, Marques MB, Fung MK, Holcomb JB, Illoh O, Kaplan cians. J Intensive Care Med 2004; 19: 38–43.
blood loss, and blood transfusions in North American LJ, Katz LM, Rao SV, Roback JD, Shander A, Tobian 16 Lacroix J, Hébert PC, Hutchison JS, Hume HA, Tucci
children in the intensive care unit. Am J Respir Crit AA, Weinstein R, Swinton McLaughlin LG, Djulbego- M, Ducruet T, Gauvin F, Collet JP, Toledano BJ, Robil-
Care Med 2008; 178: 26–33. vic B; Clinical Transfusion Medicine Committee of the lard P, Joffe A, Biarent D, Meert K, Peters MJ; TRI-
6 English M, Ahmed M, Ngando C, Berkley J, Ross A: AABB: Red blood cell transfusion: a clinical practice PICU Investigators; Canadian Critical Care Trials
Blood transfusion for severe anaemia in children in a guideline from the AABB. Ann Intern Med 2012; 157: Group; Pediatric Acute Lung Injury and Sepsis Investi-
Kenyan hospital. Lancet 2002; 359: 494–495. 49–58. gators Network: Transfusion strategies for patients in
7 Lackritz EM, Campbell CC, Ruebush TK 2nd, High- 12 Goodman AM, Pollack MM, Patel KM, Luban NL: pediatric intensive care units. N Engl J Med 2007; 356:
tower AW, Wakube W, Steketee RW, Were JB: Effect Pediatric red blood cell transfusions increase resource 1609–1619.
of blood transfusion on survival among children in a use. J Pediatr 2003; 142: 123–127.
Kenyan hospital. Lancet 1992; 340: 524–528.

300 Transfus Med Hemother 2016;43:297–301 Chegondi/Sasaki/Raszynski/Totapally

17 Rouette J, Trottier H, Ducruet T, Beaunoyer M, Lac- 20 Kirpalani H, Whyte RK, Andersen C, Asztalos EV, Hed- 22 Gibson BE, Todd A, Roberts I, Pamphilon D, Rodeck
roix J, Tucci M; Canadian Critical Care Trials Group; dle N, Blajchman MA, Peliowski A, Rios A, LaCorte M, C, Bolton-Maggs P, Burbin G, Duguid J, Boulton F,
PALISI Network: Red blood cell transfusion threshold Connelly R, Barrington K, Roberts RS: The Premature Cohen H, Smith N, McClelland DB, Rowley M, Turner
in postsurgical pediatric intensive care patients: a ran- Infants in Need of Transfusion (PINT) study: a ran- G; British Committee for Standards in Haematology
domized clinical trial. Ann Surg 2010; 251: 421–427. domized, controlled trial of a restrictive (low) versus Transfusion Task Force: Writing group: transfusion
18 Carson JL, Hill S, Carless P, Hébert P, Henry D: Trans- liberal (high) transfusion threshold for extremely low guidelines for neonates and older children. Br J Hema-
fusion triggers: a systematic review of the literature. birth weight infants. J Pediatr 2006; 149: 301–307. tol 2004; 124: 433–453.
Transfus Med Rev 2002; 16: 187–199. 21 Dallman M, Liu X, Harris AD, Hess JR, Edelman BB, 23 Karam O, Tucci M, Ducruet T, Hume HA, Lacroix J,
19 Bell EF, Strauss RG, Widness JA, Mahoney LT, Mock Murphy DJ, Netzer G: Changes in transfusion practice Gauvin F; Canadian Critical Care Trials Group; PALISI
DM, Seward VJ, Cress GA, Johnson KJ, Kromer IJ, over time in the PICU. Pediatr Crit Care Med 2013; 14: Network: Red blood cell transfusion thresholds in pedi-
Zimmerman MB: Randomized trial of liberal versus 843–850. atric patients with sepsis. Crit Care 2011; 12: 512–518.
restrictive guidelines for red blood cell transfusion in 24 Tyrrell CT, Bateman ST: Critically ill children: to
preterm infants. Pediatrics 2005; 115: 1685–1691. transfuse or not to transfuse packed red blood cells,
that is the question. Pediatr Crit Care Med 2012; 13:
204–209.