Journal of Critical Care (2011) 26, 106.e1–106.e6

Red blood cell transfusions—are we narrowing theevidence-practice gap? An observational study in 5 Israeliintensive care units☆

Jonathan Cohen MDa,b,⁎, Ilya Kagan MDa,b, Remos Hershcovici MD c,d,Sylvianne Bursztein-De Myttenaere MDc,d, Nicola Makhoul MDd,e,Alexander Samkohvalov MDd,e, Moshe Hersch MD f,g, Sharon Einav MD f,g,Vadim Berezovsky MDh,i, Daniel Jorge Jakobson MDh,i, Pierre Singer MDa,b

aGeneral Intensive Care, Rabin Medical Center, Petah Tikva 49100, IsraelbSackler School of Medicine, University of Tel Aviv, 69978, IsraelcGeneral Intensive Care, Lady Davis Medical Center, Haifa 34362, IsraeldRuth and Bruce Rappaport Faculty of Medicine, Technion Institute of Technology, Haifa 32000, IsraeleGeneral Intensive Care, Western Galilee Hospital-Nahariya, 22100, IsraelfGeneral Intensive Care, Shaare Zedek Medical Center, 91031, JerusalemgFaculty of Medicine, Hebrew University of Jerusalem, 91120, IsraelhGeneral Intensive Care, Barzilai Medical Center, Ashkelon 78278, IsraeliFaculty of Health Sciences, Ben-Gurion University of the Negev, Beer-Sheva 84105, Israel

0d

Keywords:Anemia;Critically ill;Red blood celltransfusions

AbstractPurpose: The aim of the study was to document transfusion practices in a cross section of generalintensive care units (ICUs) in Israel and to determine whether current guidelines are being applied.Materials and Methods: This prospective study was performed in 5 general ICUs in Israel over a 3-month period. Red cell transfusion data collected on consecutive patients included the trigger, unitstransfused per transfusion event, and indications, categorized either to treat a specified condition forwhich transfusions may be beneficial (acute hemorrhage, acute myocardial ischemia, or severe sepsis) orto treat a low hemoglobin concentration.Results: Of the 238 patients studied, 50% received at least one red blood cell transfusion. The mainindication for transfusion (43.7%, or 162/368 U transfused) was to treat a low hemoglobinconcentration, in the absence of one of the specified conditions. Total red cell use was 3.0 ± 2.9 Uper admission, and patients received a mean of 1.2 ± 0.4 U per transfusion event. The transfusion triggerfor the whole group was 7.9 ± 1.1 g/dL. This did not differ significantly between the indications apartfrom a significantly higher trigger for patients with acute myocardial ischemia (8.8 ± 0.9 g/dL). Inaddition, patients with a history of heart disease had a higher trigger irrespective of the primary

☆ The authors declare that they have no competing interests.⁎ Corresponding author. General Intensive Care, Rabin Medical Center, Petah Tikva 49100, Israel. Tel.: + 972 3 9376524.E-mail address: jonatanc@clalit.org.il (J. Cohen).

883-9441/$ – see front matter © 2011 Elsevier Inc. All rights reserved.oi:10.1016/j.jcrc.2010.03.010

106.e2 J. Cohen et al.

indication for transfusion and received significantly more units per transfusion event. Patients receivinga transfusion had significantly longer ICU stay and hospital mortality.Conclusions: Our study showed that evidence-practice gaps continue to exist, and it appears thatphysician behavior is mainly driven by the absolute level of hemoglobin. Educational interventionsfocused on these factors are required to limit the widespread and often unnecessary use of this scarceand potentially harmful resource.© 2011 Elsevier Inc. All rights reserved.

1. Introduction

Anemia is very common in critically ill patients beingtreated in the intensive care unit (ICU), and by the third ICUday, about 95% of patients have a hemoglobin (Hb)concentration below normal [1]. A significant number ofthese patients will receive red blood cell (RBC) transfusionswith the aim of improving clinical outcomes. However, anumber of findings have questioned the efficacy and safety ofthis therapeutic intervention. Firstly, the demonstratedincrease in oxygen delivery has not been consistently shownto translate into an increase in oxygen use [2,3]. Secondly, theTransfusion Requirements in Critical Care (TRICC) studyrevealed that patients assigned to a restrictive transfusionstrategy, that is, Hb level between 7 and 9 g/dL with atransfusion trigger of 7 g/dL, demonstrated a trend towardbetter clinical outcomes when compared to patients assignedto a liberal transfusion strategy, that is, Hb level between 10and 12 g/dL and a transfusion trigger of 10 g/dL [4]. Foryounger patients (b55 years of age) and thosewhowere less ill(Acute Physiology and Chronic Health Evaluation[APACHE] II score b 20), the mortality rate was significantlylower in the restrictive group. Thirdly, recent studies havestrengthened the association between transfusions and adverseevents. These include an independent association with thedevelopment of the acute respiratory distress syndrome [5], anincrease in the incidence of nosocomial infections [6], as wellas increased length of ICU stay and mortality [7,8]. Finally,RBCs are a limited and expensive resource.

In view of these findings, it has been suggested thatlimiting RBC transfusions may reduce morbidity andmortality. Although a transfusion trigger of 7 g/dL is nowconsidered adequate for most stable critically ill patients, withthe goal of maintaining the Hb level between 7 and 9 g/dL[9,10], certain clinical conditions may benefit from RBCtransfusions so that increasing the Hb level specifically inthese circumstances may be considered appropriate [11].However, evidence may take many years to diffuse intopractice and studies have revealed that compliance withcurrent recommendations remains low [11,12]. Defining thereasons for this would be an important step in improving andpromoting accepted transfusion practices.

We undertook the present prospective, observational,multicenter study to document transfusion practices in across section of general ICUs in Israel and to determinewhether current guidelines are being applied.

2. Materials and methods

2.1. Study population

The study was conducted in 5 general ICUs from 5different Israeli institutions: the Lady Davis Medical Center,Haifa, a 12-bed closed, teaching ICU in a 928-bed hospital;the Rabin Medical Center, Tel Aviv, a 12-bed closed,teaching ICU in a 1300-bed hospital; the Western GalileeHospital, Nahariya, a 10-bed, closed, teaching ICU in a 500-bed hospital; the Shaare Zedek Medical Center, Jerusalem, a5-bed, closed, teaching ICU in a 340-bed hospital; and theBarzilai Medical Center, Ashkelon, a 5-bed, closed, teachingICU in a 500-bed hospital. The study was approved by therelevant local ethics committees who waived consent as thestudy was noninterventional, did not require deviation fromroutine medical practice, and confidentiality was respected.The ICUs in the study have a dedicated team of specialistintensivists solely responsible for patient care. None of theICUs has a written protocol specifically relating totransfusion practices and the decision regarding RBCtransfusion was left to the judgment of the attendingphysician. Apart from a study coordinator at each site, theICU staff was not made aware of the study. The studyprotocol was registered at ClinicalTrials.gov with thenumber NCT00413816.

2.2. Study protocol

The study was a prospective, observational, noninterven-tional survey. All consecutive patients admitted to theparticipating ICUs over a 3-month period during 2007 wereenrolled. Data regarding blood transfusions as detailed belowwere collected by a study coordinator at each participatingICU. The data sources were bedside records, patient charts,and in particular, discussion with the ICU staff regarding theindication for transfusion. All data were collected on a dailybasis for each patient throughout the ICU stay, entered intothe study database, and subsequently analyzed.

2.3. Data collection in ICU

The following data were collected for each patient onadmission to the ICU: age, sex, admission type (surgical,medical, trauma), Hb level, APACHE II score [13], and

Table 1 Demographics and baseline characteristics

Variables n = 238

Age (y) 59.1 ± 18.5Sex (% male) 131 (55)APACHE II score 21.9 ± 7.1Admission categorySurgical (%) 107 (45)Medical (%) 105 (44.2)Trauma (%) 26 (10.8)Medical historyIschemic heart disease (%) 65 (27.5)Congestive heart failure (%) 33 (13.7)Admission Hb level (g/dL) 10.5 ± 1.9

106.e3Evidence-practice gap in RBC transfussions

medical history of ischemic heart disease (if the patient had adocumented history of myocardial infarction or angina) and/or congestive heart failure (as determined from the caserecords). Data collected during the ICU stay included the Hblevel recorded closest to the RBC transfusion, defined as thetransfusion trigger; the main indication for transfusion; thenumber of RBC transfusions, defined as transfusion events;the number of RBCs transfused per transfusion event; thetotal number of RBC transfusions during the ICU stay; anycomplication judged to be directly related to the transfusion;and the Hb level at ICU discharge. A pretransfusion Hbconcentration value was available for all patients. Indicationswere categorized as either to treat a condition for whichtransfusions may be beneficial [11], namely, (i) the presenceof acute hemorrhage, defined as an estimated blood loss ofmore than 300 mL in a 24-hour period; (ii) severe sepsis; and(iii) acute myocardial ischemia; or to treat a low Hbconcentration for other reasons. In addition, ICU durationof stay, as well as ICU and hospital mortality, was noted. Thestudy coordinators were instructed to record any events thatmight have been considered complications of the transfusionas defined in the literature [14] such as (i) transfusionreactions, including hemolytic (defined by the presence offever, chills, rigors, chest/back/abdominal pain associatedwith laboratory evidence of hemolysis), febrile (defined by a1°C increase in temperature during or soon after atransfusion, which may be associated with chills, rigors,and discomfort), septic, and urticarial (defined by thepresence of hives with or without edema, pruritis, andangioedema) reactions; (ii) transfusion-related acute lunginjury, defined as PaO2/fraction of inspired oxygen of lessthan 300 mm Hg with bilateral pulmonary changes in theabsence of both cardiogenic pulmonary edema and other riskfactors for ALI; and (iii) transfusion-associated circulatoryoverload, defined as the presence of dyspnea, cough,tachycardia, hypertension, and widened pulse pressure.

2.4. Statistical analyses

Results are expressed as mean ± SD. Statisticalcomparisons were performed by the Fisher exact test,Student t test, and Mann-Whitney U test as appropriate.Analysis of variance was used to detect differences betweenmore than 2 continuous variables. A P value of less than .05was considered significant. Statistical calculations wereperformed using SPSS software (version 12.0; SPSS Inc,Chicago, Ill).

3. Results

During the study period, 240 patients were admitted to theparticipating ICUs. Two patients were excluded from thestudy due to inadequate data collection regarding admissionHb level and transfusion trigger, so that data from 238

patients was available for analysis. Patient demographics andbaseline characteristics are shown in Table 1.

Table 2 shows the transfusion data during the ICU stay.Half of the patients received RBC transfusion/s, with mostpatients receiving 2 or more. There was no significantdifference between the overall mean transfusion trigger forthe whole group and for those without bleeding (7.9 ± 1.1g/dL vs 7.9 ± 0.9 g/dL, respectively; P = .82). Thecommonest indication for transfusion was to treat a low Hblevel (43.7%), followed by acute hemorrhage, severe sepsis,and acute myocardial ischemia. Apart from a significantlyhigher transfusion trigger when the indication was acutemyocardial ischemia, the trigger did not differ significantlybetween the other indications. Compared to the wholegroup, the transfusion trigger was also significantly higherfor patients with comorbidities of ischemic heart disease orcongestive heart failure, irrespective of the primaryindication for transfusion; in addition, these patientsreceived significantly more units per transfusion event.Complications of RBC transfusions were reported in 3.2%of transfusion events. These included chills in 1.6% ofpatients, fever in 0.8%, and rash in 0.8%.

Demographic and outcome differences between patientswho received transfusions and nontransfused patients areshown in Table 3. Transfused patients were significantlyolder and had a higher APACHE II score. In addition,patients admitted for trauma were transfused most frequentlyfollowed by those admitted for surgery and least frequentlyfor medical reasons. Transfused patients had a significantlylonger ICU length of stay and hospital mortality.

4. Discussion

Following the publication of the TRICC study [4], there isevidence that many critical care practitioners have adopted amore restrictive approach to RBC transfusions. Changes inattitude were documented in a Canadian scenario-basednational survey showing significant modification of transfu-sion practices over time from 1993 to 2003 [15]. Changes in

Table 2 Transfusion data during ICU stay

Variable Transfusion trigger (Hb level, g/dL) No. of units transfused

No. of patients transfused (%) 119 (50%)Units transfused1 (%) 45 (37.8)N2 (%) 74 (62.2)Mean transfusion events/patient 3.0 ± 2.9Units transfused/event 1.6 ± 1.2Total group 7.9 ± 1.1Indication for transfusionBleeding N 300 mL (25.2%) 7.9 ± 1.7 ⁎ 2.6 ± 1.8 ⁎⁎

Severe sepsis (24.4%) 7.9 ± 0.7 1.4 ± 1.0Myocardial ischemia (6.7%) 8.8 ± 0.9 1.1 ± 0.4Increase Hb concentration (43.7%) 7.7 ± 0.9 1.1 ± 0.3ComorbiditiesIschemic heart disease 8.4 ± 1.1 ⁎⁎⁎ 1.84 ± 0.2Congestive heart failure 8.5 ± 0.9 1.54 ± 0.4

⁎ P = .07 for difference between transfusion trigger and indication.⁎⁎ P = .01 for difference between number of units transfused per indication.⁎⁎⁎ P = .03 for difference between transfusion trigger in presence of comorbidities and whole group transfusion trigger.

106.e4 J. Cohen et al.

actual practice are evident from the results of studiesexamining transfusion practices in critically ill patients.Thus, for example, although the transfusion rate in the CRIT(Anemia and blood transfusion in the critically ill – currentclinical practice in the United States) study was 44% [16],more recent studies have revealed a rate of 23% in ananalysis of a database of 139 hospitals in the United States[17] and 33% in the European Sepsis Occurrence in Acutelyill Patients study [18]. The transfusion rate in our study wasmuch higher, namely 50%, which represented 368 U ofRBCs. This did not appear to be related to our specific ICUpopulation because patient-related factors associated withRBC administration in our study, such as older age, higherseverity of illness score, and admission diagnosis of traumaor surgery, were similar to those reported in otherepidemiological studies [16-19].

To better understand the reasons for this, we examinedpossible factors that may have resulted in this high rate,

Table 3 Demographic and outcome differences betweenpatients who received transfusions and nontransfused patients

Variable Transfused Nontransfused P

Age, y 63.0 ± 16.5 55.0 ± 19.5 b.001Sex, male/female 55:45 45:55 .35APACHE II 23.2 ± 7.1 20.2 ± 6.9 .003Admission category (%)Trauma 65.4 34.5 .01Surgical 57.5 42.5Medical 40.6 59.4ICU length of stay, d 13.9 ± 16.2 4.7 ± 6.2 .001ICU mortality, % 24.8 18.8 .28In-hospital mortality, % 40 25.2 .02

specifically indications for transfusion, the transfusiontrigger, and the amount of RBC units administered pertransfusion event. Epidemiological studies have typicallydefined the main indications for transfusion as acutebleeding, a low Hb level, or altered or reduced physiologicreserve [19-21]. However, the latter indication was notfurther defined in these studies, and in addition, no singlephysiologic variable has been shown to establish the need forRBC transfusions [22]. Recently, it has been recommendedthat RBC transfusions may be beneficial (and thus indicated)in certain clinical settings [9,11]. Because it was a purpose ofthe present study to assess compliance with current guide-lines, we proposed 3 indications for transfusion based onthese recommendations. Firstly, RBC transfusions areclearly indicated in the presence of acute bleeding. Secondly,concern has been expressed that critically ill patients withischemic heart disease may require higher levels of Hb.Although anemia has been associated with increasedmorbidity and mortality in patients with underlying ischemicheart disease [23], a recent review concluded that onlypatients with ST elevation myocardial infarction appeared tobenefit from transfusion [24]. We thus included acutemyocardial ischemia as a potential indication. Thirdly,RBC transfusions are among the interventions shown toimprove outcome in the early, pre-ICU management ofpatients with septic shock [25]. Current recommendations forthese patients in the ICU, however, do not advocate routineRBC transfusion but rather in the presence of coexistentconditions, such as myocardial ischemia, acute hemorrhage,or lactic acidosis [26]. To capture these circumstances, weincluded severe sepsis as the third potential indication.Finally, we included a fourth indication, namely to treat alow Hb concentration, intended to reflect transfusionstriggered primarily by the level of Hb itself.

106.e5Evidence-practice gap in RBC transfussions

We believe that this categorization of indications allowedus to make further assumptions regarding transfusionpractice. We found that the main indication for RBCtransfusion, 44% of all those administered, or 162 of 368U, was to treat a low Hb concentration, a finding that hasbeen reported by others [20,21]. However, the fact that thetransfusion trigger did not differ significantly for indicationsof bleeding, severe sepsis, or to treat a low Hb concentrationsuggests that in effect the critical element in decision makingwas not the indication but rather the actual level of Hb. Thetransfusion trigger for our whole patient group was 7.9 g/dL,which is compatible with that found in 2 recent studies, viz,8.2 g/dL and 8.0 g/dL, respectively [7,27]. It thus appearsthat for many physicians an Hb level of less than 8 g/dLremains the lower “acceptable” level, irrespective of cause.More studies documenting the safety of Hb levels in thisrange may be useful in instilling confidence to changetransfusion practices.

Our study also clearly revealed that the other majordeterminant of transfusion decisions was the presence ofheart disease. Thus, the trigger for patients with acutemyocardial ischemia was significantly higher than for allother indications. This is in line with current recommenda-tions suggesting that a transfusion trigger between 8 and 10g/dL is reasonable for these patients [9]. However, we noted,as have others [28], that the transfusion trigger was alsosignificantly higher for patients with a history of ischemicheart disease compared to those without. In addition, thesepatients received more units per transfusion event. In thisregard, in a recent review, Gerber concluded that in theabsence of acute myocardial ischemia, transfusions are oflimited clinical use in patients with ischemic heart disease,while carrying the potential for serious adverse conse-quences [24]. Indeed, recent guidelines suggest that in thepresence of stable cardiac disease, transfusion should beconsidered only when the Hb level is less than 7 g/dL [29].Application of this guideline may result in significantlydecreased RBC use. Regarding the number of unitsadministered per transfusion, this was mainly influencedby the presence of active bleeding (2.6 ± 1.8 U/event) and ahistory of ischemic heart disease, as referred to above (1.8 ±1.6 U/event). However, in their absence, patients received amean of 1.2 ± 0.4 U/event, which is in keeping with theguidance given by the TRICC study [4]. Our findings,together with those the Canadian survey [14] that showed asignificant increase in the use of single-unit transfusions overtime, from 10% in 1993 to 56% in 2003, suggest thatphysicians are increasingly avoiding overtransfusion.

There are potential limitations to our study. Being anobservational study, no uniform definitions of sepsis andacute myocardial ischemia were used among the participat-ing ICUs. In addition, not all the factors that may haveinfluenced transfusion-related decisions where the indicationwas to increase the Hb level for other reasons may have beencollected. Although the study was conducted in 5 of the 15general ICUs in Israel, they represent a cross section of ICUs

in Israel, so that our results may reflect general transfusionpractices in Israel. In support of this, the transfusion practiceswe describe are similar to those noted in recent studies inother parts of the world.

In conclusion, evidence-practice gaps continue to existand resulted in a higher transfusion rate in participating IsraeliICUs compared to recent experience in other countries. Itappears from our study design that transfusion decisions weremainly initiated by the absolute Hb level, in the absence ofconditions for which transfusions may be beneficial (such asactive hemorrhage, acute myocardial ischemia, or severesepsis). Whether this is related to a lack of knowledgeregarding indications for transfusion or a reluctance to acceptthe safety of a low Hb level in stable critically ill patientsrequires further investigation. We believe that individualunits involved in the study will benefit from the informationgathered and enable focused educational interventions tolimit the widespread and often unnecessary use of this scarceand potentially harmful resource.

References

[1] Corwin HL, Parsonnet KC, Gettinger A. RBC transfusion in the ICU.Is there a reason. Chest 1995;108:767-71.

[2] Fernandes Jr CJ, Akamine N, De Marco FV, et al. Red blood celltransfusion does not increase oxygen consumption in critically illseptic patients. Crit Care 2001;5:362-7.

[3] Lorente JA, Landín L, De Pablo R, et al. Effects of blood transfusionon oxygen transport variables in severe sepsis. Crit Care Med 1993;21:1312-8.

[4] Hébert PC, Wells G, Blajchman MA, et al. A multicenter, randomized,controlled clinical trial of transfusion requirements in critical care.Transfusion Requirements in Critical Care Investigators, CanadianCritical Care Trials Group. N Engl J Med 1999;340:409-17.

[5] Zilberberg MD, Carter C, Lefebvre P, et al. Red blood cell transfusionsand the risk of acute respiratory distress syndrome among the criticallyill: a cohort study. Crit Care 2007;11:R63.

[6] Taylor RW, O'Brien J, Trottier SJ, et al. Red blood cell transfusionsand nosocomial infections in critically ill patients. Crit Care Med2006;34:2302-8.

[7] Zilberberg MD, Stern LS, Wiederkehr DP, et al. Anemia, transfusionsand hospital outcomes among critically ill patients on prolonged acutemechanical ventilation: a retrospective cohort study. Crit Care2008;12:R60.

[8] Malone DL, Dunne J, Tracy JK, et al. Blood transfusion, independentof shock severity, is associated with worse outcome in trauma. JTrauma 2003;54:898-905.

[9] Corwin H. Anemia and red blood cell transfusion in the critically ill.Sem Dialysis 2006;19:513-6.

[10] Marik PE, Corwin HL. Efficacy of red blood cell transfusion in thecritically ill: a systematic review of the literature. Crit Care Med2008;36:2667-74.

[11] Corwin HL, Shorr AF. Red blood cell transfusion in the critically ill:when is it time to say enough? Crit Care Med 2009;37:2114-6.

[12] Levy MM, Abraham E, Zilberberg M, et al. A descriptive evaluation oftransfusion practices in patients receiving mechanical ventilation.Chest 2005;127:928-35.

[13] Knaus WA, Draper EA, Wagner DP, et al. APACHE II: a severity ofdisease classification system. Crit Care Med 1985;13:818-29.

[14] Hendrickson JE, Hillyer CD. Noninfectious serious hazards oftransfusion. Anesth Analg 2009;108:759-69.

106.e6 J. Cohen et al.

[15] Hébert PC, Fergusson DA, Stather D, et al. Canadian Critical CareTrials Group: revisiting transfusion practices in critically ill patients.Crit Care Med 2005;33:7-12.

[16] Corwin HL, Gettinger A, Pearl RG, et al. The CRIT study: anemia andblood transfusions in the critically ill–current clinical practice in theUnited States. Crit Care Med 2004;32:39-52.

[17] Dasta J, Mody SH, McLaughlin T, et al. Current management ofanemia in critically ill patients: analysis of a database of 139 hospitals.Am J Ther 2008;15:423-30.

[18] Vincent JL, Sakr Y, Sprung C, et al. Sepsis Occurrence in Acutely IllPatients (SOAP) Investigators: Are blood transfusions associated withgreater mortality rates? Results of the Sepsis Occurrence in Acutely IllPatients study. Anesthesiology 2008;108:31-9.

[19] Vincent JL, Baron J-F, Reinhart K, et al. Anemia and bloodtransfusions in critically ill patients. JAMA 2002;288:1499-507.

[20] Chohan SS, NcArdle F, McClelland SJ, et al. Red cell transfusionpractice following the transfusion requirement in critical care (TRICC)study: prospective observational cohort study in a large UK intensivecare unit. Vox Sang 2003;84:211-8.

[21] Rao MP, Boralessa H, Morgan C, et al. North Thames Blood InterestGroup: Blood component use in critically ill patients. Anaesthesia2002;57:530-4.

[22] McLellan SA, McLelland DB, Walsh TS. Anemia and red blood celltransfusions in the critically ill patient. Blood Rev 2003;17:195-208.

[23] Rawn J. The silent risks of blood transfusion. Curr Opin Anaesth2008;21:664-8.

[24] Gerber DR. Transfusion of packed red blood cells in patients withischemic heart disease. Crit Care Med 2008;36:1068-74.

[25] Rivers E, Nguyen B, Havstad S, et al. Early Goal-Directed TherapyCollaborative Group. Early goal-directed therapy in the treatment ofsevere sepsis and septic shock. N Engl J Med 2001;19:1368-77.

[26] Dellinger RP, Levy MM, Carlet JM, et al. Surviving Sepsis Campaign:international guidelines for management of severe sepsis and septicshock: 2008. Crit Care Med 2008;36:296-327.

[27] Corwin HL, Gettinger A, Fabian T, et al. Efficacy and safety oferythropoietin alpha in the critically ill. N Engl J Med 2007;357:965-76.

[28] Walsh TS, McClelland DB, Lee RJ, et al. Prevalence of ischemic heartdisease at admission to intensive care and its influence on red celltransfusion thresholds: multicentre Scottish study. Br J Anaesth2005;94:445-52.

[29] Napolitano LM, Kurek S, Luchette FA, et al. Clinical practiceguideline: red blood cell transfusion in adult trauma and critical care.Crit Care Med 2009;7:3124-57.