Red blood cell storage duration and mortality inpatients undergoing percutaneouscoronary interventionSimon D. Robinson, MBChB, MD,a Christian Janssen, PhD,a,b Eric B. Fretz, MD,a Brian Berry, MD,a

Alex J. Chase, MBChB, PhD,c Anthony Della Siega, MD,a Ronald G. Carere, MD,d Anthony Fung, MBBS,e

Gerald Simkus, MD,f W. Peter Klinke, MD,a and J. David Hilton, MDa Victoria BC, Vancouver BC,New Westminster BC, and Alberta, Canada; and Swansea, Wales, UK

Background Blood transfusion has been associated with an increased mortality in patients undergoing percutaneouscoronary intervention (PCI). Although the reasons for this remain unclear, it may be related to the structural and functionalchanges occurring within red blood cells (RBCs) during storage. We investigated whether RBC storage duration was associatedwith mortality in patients requiring transfusion after PCI.

Methods We collected data on all RBC transfusions occurring within 10 days of PCI (excluding those related to cardiacsurgery) using the British Columbia Cardiac Registry and Central Transfusion Registry. Transfusion details were analyzedaccording to 30-day survival.

Results From a total of 32,580 patients undergoing PCI, 909 (2.8%) patients received RBCs with a mean storage durationof 25 ± 10 days. In these 909 patients, mean transfusion volumes were lower in survivors (2.8 ± 2.1 vs 3.8 ± 2.9 U, P = .002)than those who died within 30 days. In a multivariate analysis to adjust for baseline risk, mean RBC storage age (HR 1.02 [95%CI 1.01-1.04], P = .002) and transfusion volume (HR 1.26 [95% CI 1.18-1.34], P b .001) both predicted 30-day mortality.Transfused patients who received only older blood (RBC min age N28 days) appeared to be at greater risk of death (HR 2.49[95% CI 1.45-4.25], P = .001).

Conclusion Red blood cell transfusion is associated with increased 30-day mortality in patients undergoing PCI. Althoughcurrent transfusion practice permits RBC storage for up to 42 days, the use of older red cells may pose an additional hazard tothis patient group. (Am Heart J 2010;159:876-81.)

BackgroundPercutaneous coronary intervention (PCI) along with

anti-thrombotic and anti-platelet pharmacotherapies im-prove outcomes in patients with coronary artery diseaseespecially those presenting with acute coronary syn-dromes.1,2 However, each of these strategies confers anincreased risk of periprocedural bleeding3 with majorbleeding being associated with a significant increase in therisk of death within 30 days.4,5

From the aVictoria Heart Institute Foundation, Victoria BC, Canada, bUniversity of Alberta,Alberta, Canada, cMorriston Cardiac Centre, Swansea, Wales, UK, dSt Paul's Hospital,Vancouver BC, Canada, eVancouver General Hospital, Vancouver BC, Canada, and fRoyalColumbian Hospital, New Westminster BC, Canada.Grant Support Victoria Foundation (Canadian Charity No. 130650898RR0001).Submitted November 9, 2009; accepted February 22, 2010.Reprint requests: Simon Robinson, MBChB, MD, Victoria Heart Institute Foundation 200-1900 Richmond Avenue, Victoria, BC, Canada V8R 4R2.E-mail: sdrobinson@vhif.org0002-8703/$ - see front matter© 2010, Mosby, Inc. All rights reserved.doi:10.1016/j.ahj.2010.02.018

Despite being widely used to treat blood loss, bloodtransfusion appears harmful in patients with coronaryartery disease,4 including those undergoing PCI5 andcardiac surgery.6 Although the reasons for this may bemulti-factorial,7 it has been suggested that this may in partbe due to structural and functional changes occurringwithin red blood cells (RBCs) during storage.8 These exvivo changes are incompletely understood but have beenrelated to decreased myocardial oxygen delivery9 andnitric oxide bioavailability10 following transfusion ofolder blood. There are conflicting data on the magnitudeand time course for RBC degradation with well charac-terized changes occurring both early and late duringstorage. Previous work using an animal model foundreduced survival following transfusion with red cellsstored for N28 days,11 and although not directlycomparable to humans, patients receiving older bloodmay be at the greatest risk from deleterious storagerelated changes within RBCs.Longer RBC storage durations have previously been

linked to increased morbidity and mortality in critically ill

Robinson et al 877American Heart JournalVolume 159, Number 5

patients and those undergoing cardiac surgery.12,13

However, little is known on whether RBC storageduration might influence outcome in patients undergoingPCI. We therefore investigated whether longer RBCstorage was associated with an increased risk of death inpatients requiring transfusion after PCI.

MethodsWe identified subjects receiving a blood transfusion follow-

ing PCI using data from the British Columbia Cardiac (BCCR)and British Columbia Central Transfusion Registries (CTR). TheBCCR has maintained data on all PCI procedures since 1994undertaken at the four hospitals providing coronary revascu-larization for the province of British Columbia (BC), Canada.For this study we included all patients undergoing PCIbetween January 1, 1999, and December 31, 2005. TheBCCR data were entered at the time of procedure by catheterlaboratory staff and held centrally by a dedicated datamanagement team at the Provincial Health Services Authority.The CTR has collected details of all transfusions involvingblood products since 1999 within the province. The use of anelectronic linkage between the British Columbia cardiac andtransfusion registries to identify subjects transfused followingPCI has previously been described.14,15

We collected data on transfusions occurring within 10 days ofPCI. Patients undergoing coronary artery bypass grafting (CABG)within 10 days of PCI were excluded as transfusion events wereassumed to be due to surgical bleeding. All RBC units wereleucoreduced at source. RBC storage age was recorded for 2231of the 2644 RBC units (84.4%). BCCR data fields includedbaseline subject demographics, comorbidity, and PCI proceduraldata. Brachial cases accounted for 0.1% of the entire cohort andwere excluded. Thirty-day mortality was determined using theMinistry of Health Vital Statistics database and cross-referencedwith the BCCR and CTR data. The study was approved by theresearch and ethics committee of the University of BritishColumbia and each of the 4 hospitals.

Statistical analysisStatistical calculations and analyses were performed using SPSS

15 software (SPSS, Chicago, IL). Frequencies were analyzed usingthe Pearson χ2 test. Means of continuous variables werecompared using the independent-samples t test (with equal orunequal variances determined with Levine test) and by analysisof variance. Transfusion variables of interest were RBC storageduration (days) and transfusion volume (number of units, U). Wehypothesized that transfusion with older blood would beassociated with increased mortality. Demographics, transfusiondetails (RBC age and transfusion volume) and mortality data wereentered into a Cox proportional hazard regression model.Explanatory variables were entered stepwise with forwardselection (complemented by backward elimination) with P-in =.05 and P-out = .10 under the criterion of Maximum Likelihood.Hazard ratios (HR), together with 95% CIs, were estimated for theeffect of mean transfusion volume and RBC storage duration onmortality after adjustment for baseline demographics andprocedural characteristics. We used RBC storage age (days) as acontinuous variable within the Cox analysis (mean storageduration of all RBCs received for patients receiving N1

transfusion). RBCs of unknown age were identified by a separatevariable within the Cox model but this variable was notsignificant. We assessed whether patients receiving only theoldest blood might be at higher risk by then replacing mean RBCage with a categorical variable for transfusions where minimumRBC storage duration was N28 days (RBCminimum [min] age N28days), that is, upper third of the 42 day maximum permittedstorage duration. We undertook further analyses to investigatethe effect of receiving blood of intermediate (RBC min age N21days) and shorter (RBC min age N14 days) storage durations.As transfusion events were not randomly assigned to patients

undergoing PCI and the possibility that multivariate analysis maynot adequately control for confounding, the propensity fortransfusion was determined using binary logistic regression.16,17

Variables that remained in the model were used to calculate theprobability (propensity score) of receiving a transfusion for eachpatient. The Cox proportional hazards model was then repeatedwith the propensity score and transfusion variables to determinewhether transfusion volume and RBC storage age remained asindependent predictors of 30-day mortality. In view of a possiblesurvival time bias for patients receiving blood following PCI, wealso ran the Cox regression model excluding patients dyingwithin 24 hours of PCI, and by including an additional variablefor early (within 48 hours after PCI) versus later transfusion.The authors are solely responsible for the design and conduct

of this study, all statistical analyses, the drafting and editing of themanuscript, and its final contents. This work was supported byfunding received from the Victoria Foundation (registeredCanadian charity 130650898RR0001).

ResultsFrom a total of 32,580 patients, 909 patients (2.8%)

received at least 1 U of blood within 10 days of PCI.Baseline characteristics of subjects according to transfu-sion status are shown in Table I. Transfused patients had agreater burden of comorbidity than non-transfused sub-jects. Transfusion volumes ranged from 1 to 21 U of blood(Figure 1) with lower mean transfusion volumes insurvivors than transfused patients dying within 30 days(2.8 ± 2.1 vs 3.8 ± 2.9 U, P = .002).Mean RBC storage duration prior to transfusion was 25 ±

10 days (median 25 , interquartile range 18-34 days);Figure 2. Baseline characteristics of transfused subjectswere not significantly different between patients receivingblood stored for longer or shorter periods than the mean/median storage duration (Table II).

RBC transfusion and mortalityA Cox proportional hazards model for 30-day mortality

was created using the variables shown in Table II to adjustfor the effect of subject demographics, procedural details,RBC transfusion volume, and RBC storage age. AdjustedHRs for the entire cohort are shown in Table III. LongerRBC storage durations were significantly associated with30-day mortality (HR 1.02 [95% CI 1.01-1.04], P = .002).Patients receiving only RBCs storedN 28 days appeared tobe at greatest risk following transfusion with an adjusted

Table I. Subject characteristics according to transfusion status

Transfusedpatients(n = 909)

Patients nottransfused(n = 31 628) P

Age 70.5 ± 11.3 64.2 ± 11.5 b.001BMI, kg/m2 26.2 ± 5.6 27.9 ± 5.0 b.001Gender, male 51.8 73.6 b.001Current smoker 17.9 19.8 .07Creatinine N90 μmol/L⁎ 55.7 44.0 b.001Stable angina 11.5 30.6 b.001Past medical history

Hypertension 67.0 57.4 b.001Dyslipidemia 55.1 63.4 b.001PVD 18.7 8.3 b.001CVD 12.2 6.9 b.001Diabetes mellitus 35.3 22.1 b.001Malignancy 11.1 6.2 b.001Prior MI/CHF 45.6 32.2 b.001Prior PCI/CABG 33.3 27.4 b.001Pulmonary disease 16.9 8.2 b.001GI/liver disease 12.9 5.9 b.001

Vascular AccessRadial 11.8 20.6 b.001Femoral 88.2 79.4

Procedural dataElective 8.9 28.2 b.001Urgent 60.9 60.9Emergency 30.2 10.9Fluoroscopy time, min 17.2 ± 13.7 13.0 ± 16.5 b.001

Clinical outcomeMortality at 30 d 12.3 1.4 b.001

Values are % or mean ± SD.P, transfused versus not transfused (Pearson m2 tests for frequencies and independentsamples t-test of means); BMI, body mass index; PVD, peripheral vascular disease;CVD, cerebrovascular disease; MI, myocardial infarction; CHF, congestive heartfailure; GI, gastrointestinal.⁎Creatinine N1.02mg/dL.

Figure 1

0 1 2 3 4 5 6 7 8 9 10 12 13 14 16 18 210

10

20

30

40

50

Transfused AliveTransfused Dead 30

RBC transfusion volume, units

% o

f G

rou

p

Not transfused Dead 30

Transfusion volumes (range 1-21 U) according to survival (light bluebars) or death (dark blue bars) within 30 days. For comparison 30-day mortality within the non-transfused cohort is shown (grey shadedbar, 0 U transfused).

Figure 2

1 7 14 21 28 35 420

2

4

6

8Transfused dead 30 daysTransfused alive

RBC storage duration, days

% o

f P

atie

nts

878 Robinson et alAmerican Heart Journal

May 2010

HR of 2.49 [95% CI 1.45-4.25], P = .001). A consistentassociation between RBC storage duration and increasedmortality was also observed for RBC Min age N21 days (HR2.30 [95% CI 1.42-3.72], P = .001) and N14 days (HR 1.94[95% CI 1.28-2.94], P = .002).

Proportion of transfused patients surviving (light blue bars) and dyingwithin 30 days (dark blue bars) according to mean RBC storageduration.

Transfusion volume and mortalityIn the multivariate analysis including baseline risk and

RBC storage duration, larger transfusion volumes werealso associated with increased 30-day mortality (HR 1.26[95% CI 1.18-1.34], P b .001) (Table III). Interestinglymean transfusion volumes were lower in patients whoreceived only older blood (RBC min age N28 days)compared to those who did not (2.4 ± 1.8 versus 3.2 ±2.4, P b .001).When the propensity score from the binary logistic

model for treatment was substituted for the baselineconditions in the Cox Proportional Hazards model for 30-day mortality, both transfusion volume and RBC storageage remained as independent predictors of death(Table IV). After excluding patients dying within 24hours of PCI, the results changed only marginally. In

particular, RBC transfusion volume (HR 1.28 [95% CI 1.19-1.38], P b .001) and mean RBC storage age (HR 1.03 [95%CI 1.02-1.05], P b .001) were associated with 30-daymortality. There appeared to be no additional risk/benefitassociated with early (b48 hours post PCI) versus latertransfusion (P = .58).

DiscussionWe have found that patients transfused with red cells

stored for longer periods are at increased risk of deathwithin 30 days following PCI. Within this cohort,

Table II. Baseline characteristics of transfused patients accordingto mean (median) RBC storage age

Patients receivingyounger blood

(n = 352)

Patients receivingolder blood(n = 360) P

Age 70.5 ± 11.2 70.3 ± 11.4 .80BMI, kg/m2 26.2 ± 5.8 26.3 ± 5.0 .82Sex, male 52.1 53.0 .83Current smoker 18.8 15.9 .60C r e a t i n i n e N90μmol/L⁎

55.3 52.0 .38

Stable angina 12.0 11.0 .65Past medical history

Hypertension 69.3 67.9 .72Dyslipidemia 53.7 58.5 .27PVD 18.9 17.6 .67CVD 13.4 11.7 .54Diabetes 32.4 36.0 .33Malignancy 10.2 11.6 .56Prior MI/CHF 43.3 47.8 .23Prior PCI/CABG 32.4 33.7 .71Pulmonary disease 18.2 13.4 .10GI/liver disease 14.4 13.2 .66

Vascular AccessRadial 10.6 6.2 .03Femoral 89.4 93.8

Procedural dataElective 9.7 7.0Urgent 62.5 60.6 .24Emergency 27.8 32.4Fluoroscopy time,min

16.5 ± 12.2 18.0 ± 15.6 .18

Clinical outcomeMortality at 30 d† 12.0 14.3 .37

Younger blood, RBC storage age b25 days; older blood ≥25 days.Values are % or mean ± SD.P, RBC storage age younger versus older blood.Pearson m2 tests for frequencies and independent samples t test of means.⁎Creatinine N1.02mg/dL.†One hundred ninety-seven patients received ≥1 U of RBCs for which storage age wasnot entered in the CTR and this group had a 30-day mortality of 9.1%.

Table III. Risk adjusted predictors of mortality

HR (95% CI) P=

Stable angina 0.31 (0.16-0.61) .001CVD 1.80 (1.30-2.49) b.001Diabetes mellitus 1.55 (1.20-1.98) .001Urgent PCI 2.05 (1.05-4.02) .04Emergency PCI 11.39 (5.69-22.78) b.001Prior pulmonary disease 1.57 (1.15-2.15) .005Prior MI/CHF 1.46 (1.15-1.86) .002Fluoroscopy time, min 1.003 (1.001-1.005) .01Dyslipidemia 0.77 (0.61-0.97) .03Age, y 1.05 (1.04-1.06) b.001Creatinine N90 μmol/L 1.32 (1.04-1.66) .02BMI b18.5 kg/m2 2.21 (1.23-3.96) .008RBC storage age, days⁎ 1.02 (1.01-1.04) .002RBC volume, U 1.26 (1.18-1.34) b.001

⁎When mean storage age was replaced with variable for the oldest blood (all RBCsreceived min age N28 days), adjusted HR was 2.49 (95% CI 1.45-4.25), P = .001.

Table IV. Cox regression including the propensity score fortransfusion

HR (95% CI) P=

Propensity score 1.007 (1.006-1.007) b.001RBC storage age, days 1.03 (1.02-1.04) b.001RBC volume, U 1.27 (1.19-1.34) b.001

Adjusted HRs for 30-day mortality using the estimated propensity score (magnitudeadjusted) for transfusion derived from the baseline variables in Table II.

Robinson et al 879American Heart JournalVolume 159, Number 5

higher transfusion volumes were also associated withmortality in addition to the hazard observed with olderblood. Furthermore, despite lower mean transfusionvolumes overall, those patients receiving only bloodstored for N28 days appeared to be at greater risk witha HR of almost 2.5. This is consistent with the notionthat red cells undergo progressive deleterious changesduring storage which may affect mortality withinvulnerable patients.Despite experimental evidence of myocardial ischemia

occurring with mild anemia,18 a low threshold fortransfusion appears neutral, or perhaps even harmful,in critically ill patients.19 Transfusion has previouslybeen shown to predict increased early and latemortality5,14 in patients undergoing PCI though themechanism underlying this remains uncertain. Data fromover 70,000 patients enrolled within the CRUSADEregistry20 and a review of 10 randomized trials21 showeda reduction in myocardial infarction and death with a

restrictive approach to transfusion. In this current study,we have confirmed an association between highertransfusion volumes and increased mortality and, inaddition, have shown that prolonged RBC storage ageappears hazardous to patients undergoing PCI. Thesefindings reinforce the notion that physicians shouldembrace therapeutic strategies which minimize bleedingand the likelihood of transfusion after PCI. This may beparticularly relevant in patients undergoing emergencyPCI who had the highest risk of death and were morelikely to require transfusion.

Red cell storage duration and mortalitySome22,23 but not all24,25 studies have shown increased

morbidity and mortality in patients receiving blood storedfor longer periods. These studies were mostly small withdifferent patient populations and substantial variability intransfusion practices between groups. Some of the earlierreports of an adverse effect of stored blood assessed theresponse to non-leucodepleted red cell transfusion.Although leucodepletion was introduced with the aim ofreducing the risk of transfusion related infection there isevidence that white blood cells compromise the functionof red cells during and following storage.26 In the largeststudy to date on the effect of RBC age, Koch et al reported

880 Robinson et alAmerican Heart Journal

May 2010

higher in hospital mortality and postoperative morbiditywith blood stored for N14 days,13 about 50% of thesepatients received leucodepleted blood. This is in keepingwith our results showing increased mortality in thosepatients receiving older blood after correction for baselinerisk. A previous case control linking mortality andtransfusion of older blood following PCI included b150patients receiving blood.27 Our current study has includeddata on N900 transfused patients and reflects routineclinical practice across four interventional centers.Despite exclusive use of leucodepleted blood within ourstudy and comparable groups receiving older and youngerblood, older blood was associated with a greater risk ofdeath. The estimated HR of ∼2.5 associated withtransfusion of blood stored for N28 days exceeded thatof diabetes mellitus, cerebrovascular disease, and renalimpairment. This suggests accumulation of the adversechanges in stored red cells over time and that patientsreceiving older blood may be at greater risk.

Pathophysiological mechanism linking transfusion andadverse outcomesThe pathophysiological mechanism linking transfusion,

red cell storage and adverse outcomes remains uncertainthough is likely multi-factorial.7 Acceptable RBC storagedurations were originally determined by demonstratingintegrity of cells during storage and survival of donor RBCsfor 24 h following transfusion as an indication of plausibletherapeutic benefit.9 However, these observations did notaccount for the possibility that stored RBCs would beineffective oxygen carriers or be directly harmful invulnerable patients through the generation of a proin-flammatory response or direct toxicity from the by-products of storage. Some of these storage related changesoccur soon after collection such as depletion of 2,3-diphosphoglycerate and adenosine triphosphate.28 Storedred cells also become deficient in nitric oxide (NO) whichis essential to erythrocyte oxygen exchange.29 Red cellsdepleted of nitric oxide may reduce local NO bio-availability predisposing to vasoconstriction, plateletaggregation and increasing the likelihood of coronaryischemia or thrombosis.30 Renitrosylation of blood usingaqueous NO improves coronary blood flow duringinfusion of stored red cells, an effect which is augmentedin the presence of coronary hypoxia.10 This raises theprospect of developing strategies to restore function instored red cells which may be of particular benefit topatients with coronary disease who are at higher risk ofmyocardial ischemia.Current transfusion practice in North America and

Europe permits the refrigerated storage of red cells for upto 42 days with appropriate additives. The mean storageduration of 25 days for our study cohort is substantiallyb42 days but longer than the 15 days reported from theUnited States.13 Our findings, which suggest transfusion

with older blood may pose an additional hazard followingPCI, require confirmation in a prospective randomizedstudy to assess the impact of transfusing younger blood inthis population. Lowering the maximum permittedstorage duration of blood may, however, impact ontransfusion services and risk shortages in the supply ofyounger RBCs for patients requiring urgent transfusion.

Limitations of studyWe used an electronic linkage between the BC cardiac

and transfusion registries to capture all non cardiacsurgery related transfusion events occurring within 10days of PCI across the 4 provincial interventional centers.The clinical decision for transfusion and allocation of redcells was non-randomized. Although we used multivariateregression and propensity scoring techniques to adjust forimbalances in potentially confounding risk variables, it ispossible that unobserved differences between patientsmay have influenced our results. Furthermore, despitelarger transfusion volumes appearing to confer a greaterrisk of death, our study cannot determine specific clinicalindications or a nadir hemoglobin level below whichtransfusion might be beneficial. Physician directedpremature discontinuation of anti-coagulant pharma-cotherapies may also be a factor explaining whytransfused patients are at increased risk of death as thismay increase the risk of stent thrombosis. Because data onpharmacotherapy were not mandated for inclusionwithin the BCCR, we were unable to investigate theimportance of medication usage within this cohort. Thisissue requires further study along with additionalinvestigation on how the choice of periproceduralantithrombotic therapy might influence bleeding andtransfusion rates following PCI.

ConclusionIn patients undergoing PCI, transfusion of older red cells

and higher transfusion volumes are associated withincreased 30-day mortality. Our results reinforce thenotion that the perceived benefits of transfusion mustbe carefully considered for each patient and that use ofolder red cells may pose an additional hazard to thoseundergoing PCI.

AcknowledgementsWe are indebted to Cheryl Lewis and Karin Humph-

ries and their staff at the BC Provincial BloodCoordinating Office (Central Transfusion Registry) andBC Cardiac Registry for their assistance with dataextraction. We thank the Victoria Foundation (registeredCanadian charity 130650898RR0001) for its grantsupport of this research.

Robinson et al 881American Heart JournalVolume 159, Number 5

References

1. Cannon CP, Weintraub WS, Demopoulos LA, et al. Comparison ofearly invasive and conservative strategies in patients with unstablecoronary syndromes treated with the glycoprotein IIb/IIIa inhibitortirofiban. N Engl J Med 2001;344:1879-87.

2. Fox K, Poole-Wilson P, Clayton T, et al. 5-year outcome of aninterventional strategy in non-ST-elevation acute coronary syndrome:the British Heart Foundation RITA 3 randomised trial. The Lancet 2005;366:914-20.

3. SYNERGY Trial Investigators. Enoxaparin vs unfractionated heparin inhigh-risk patients with non-ST-segment elevation acute coronarysyndromes managed with an intended early invasive strategy: primaryresults of the SYNERGY randomized trial. JAMA 2004;292:45-54.

4. Rao SV, Jollis JG, Harrington RA, et al. Relationship of bloodtransfusion and clinical outcomes in patients with acute coronarysyndromes. JAMA 2004;292:1555-62.

5. Kinnaird TD, Stabile E, Mintz GS, et al. Incidence, predictors, andprognostic implications of bleeding and blood transfusion followingpercutaneous coronary interventions. Am J Cardiol 2003;92:930-5.

6. Murphy GJ, Reeves BC, Rogers CA, et al. Increased mortality,postoperative morbidity, and cost after red blood cell transfusion inpatients having cardiac surgery. Circulation 2007;116:2544-52.

7. Doyle BJ, Rihal CS, Gastineau DA, et al. Bleeding, blood transfusion,and increased mortality after percutaneous coronary intervention:implications for contemporary practice. J Am Coll Cardiol 2009;53:2019-27.

8. Bennett-Guerrero E, Veldman TH, Doctor A, et al. Evolution of adversechanges in stored RBCs. Proc Nat Acad Sci 2007;104:17063-8.

9. Tinmouth A, Fergusson D, Yee IC, et al. Clinical consequences of redcell storage in the critically ill. Transfusion 2006;46:2014-27.

10. Reynolds JD, Ahearn GS, Angelo M, et al. S-nitrosohemoglobindeficiency: a mechanism for loss of physiological activity in bankedblood. Proc Nat Acad Sci 2007;104:17058-62.

11. Sielenkamper AW, Chin-Yee IH, Martin CM, et al. Diaspirin cross-linked hemoglobin improves systemic oxygen uptake in oxygensupply-dependent septic rats. Am J Respir Crit Care Med 1997;156:1066-72.

12. Vamvakas EC, Carven JH. Transfusion and postoperative pneumoniain coronary artery bypass graft surgery: effect of the length of storageof transfused red cells. Transfusion 1999;39:701-10.

13. Koch CG, Li L, Sessler DI, et al. Duration of red-cell storage andcomplications after cardiac surgery. N Engl J Med 2008;358:1229-39.

14. Chase AJ, Fretz EB, Warburton WP, et al. Association of the arterialaccess site at angioplasty with transfusion and mortality: the M.O.R.T.A.L study (Mortality benefit Of Reduced Transfusion after percutane-ous coronary intervention via the Arm or Leg). Heart 2008;94:1019-25.

15. Byrne J, Spence MS, Fretz E, et al. Body mass index, periproceduralbleeding, and outcome following percutaneous coronary intervention(from the British Columbia cardiac registry). Am J Cardiol 2009;103:507-11.

16. Rosenbaum PR, Rubin DB. The central role of the propensity score inobservational studies for causal effects. Biometrika 1983;70:41-55.

17. D'Agostino RBJ. Propensity score methods for bias reduction in thecomparison of a treatment to a non-randomized control group. StatMed 1998;17:2265-81.

18. Hebert PC, Yetisir E, Martin CM, et al. Is a low transfusion thresholdsafe in critically ill patients with cardiovascular diseases? Crit CareMed 2001;29:227-34.

19. Hebert PC, Wells G, Blajchman MA, et al. A multicenter,randomized, controlled clinical trial of transfusion requirements incritical care. Transfusion Requirements In Critical Care investigators,Canadian Critical Care Trials Group. N Engl J Med 1999;340:409-17.

20. Alexander KP, Chen AY, Wang TY, et al. Transfusion practice andoutcomes in non ST-segment elevation acute coronary syndromes. AmHeart J 2008;155:1047-53.

21. Hill SR, Carless PA, Henry DA, et al. Transfusion thresholds and otherstrategies for guiding allogeneic red blood cell transfusion. CochraneDatabase Syst Rev 2002:CD002042.

22. Purdy FR, Tweeddale MG, Merrick PM. Association of mortality withage of blood transfused in septic ICU patients. Can J Anaesth 1997;44:1256-61.

23. Basran S, Frumento RJ, Cohen A, et al. The association betweenduration of storage of transfused red blood cells and morbidity andmortality after reoperative cardiac surgery. Anesth Analg 2006;103:15-20.

24. van de Watering L, Lorinser J, Versteegh M, et al. Effects of storagetime of red blood cell transfusions on the prognosis of coronary arterybypass graft patients. Transfusion 2006;46:1712-8.

25. Yap CH, Lau L, Krishnaswamy M, et al. Age of transfused red cells andearly outcomes after cardiac surgery. Ann Thorac Surg 2008;86:554-9.

26. Shanwell A, Kristiansson M, Remberger M, et al. Generation ofcytokines in red cell concentrates during storage is prevented byprestorage white cell reduction. Transfusion 1997;37:678-84.

27. Kim P, Dixon S, Eisenbrey AB, et al. Impact of acute blood loss anemiaand red blood cell transfusion on mortality after percutaneouscoronary intervention. Clin Cardiol 2007;30:II35-II43.

28. Wolfe LC. The membrane and the lesions of storage in preserved redcells. Transfusion 1985;25:185-203.

29. McMahon TJ, Moon RE, Luschinger BP, et al. Nitric oxide in the humanrespiratory cycle. Nat Med 2002;8:711-7.

30. Stamler JS, Jia L, Eu JP, et al. Blood flow regulation by S-nitrosohemoglobin in the physiological oxygen gradient. Science1997;276:2034-7.