blood product use during routine open heart surgery: the impact of the centrifugal pump

Blood Product Use During Routine Open Heart Surgery:The Impact of the Centrifugal Pump

*Michael Klein, †Chris Brown Mahoney, *Chris Probst, *Hagen D. Schulte, and*Emmeran Gams

*Department of Cardiothoracic Surgery, Heinrich-Heine University Hospital, Dusseldorf, Germany; and †Carlson Schoolof Management, University of Minnesota, Minneapolis, Minnesota, U.S.A.

Abstract: A prospective randomized study was done in-cluding 1,000 patients undergoing routine open heart sur-gery. Patients were randomly assigned to either a rollerpump or a BioMedicus centrifugal pump with identicalextracorporeal circuits.There were no significant differ-ences between study groups. Actual blood products trans-fused and predicted transfusion requirements (using Car-diac RiskMaster) were examined as was chest tubedrainage (CTD). The predicted tranfusion requirementwas 885 of 1,000 patients. Transfusions were required by472 of 1,000. Risk factors as significant predictors of in-creased CTD and use of blood products were emergency

surgery status, increased cross-clamp time, and higher pre-dicted risk of mortality. The only significant predictor ofdecreased CTD was the use of a centrifugal pump. Predic-tors of increased length of stay were myocardial infarction,preoperative urea, age, and massive transfusion. Data pro-vide evidence that use of the centrifugal pump improvespatient outcomes by decreasing CTD and decreasing therequirements for transfusion, which results in a shorterhospital stay. Key Words: Cardiopulmonary bypass—Centrifugal blood pump—Blood loss—Blood transfu-sion—Clinical outcomes.

Over the past decade, mortality and morbidity fol-lowing cardiac surgical procedures did not decreaseas might have been expected given improvements insurgical technique and the technological progress inthe field of extracorporeal circulation (1). Myocar-dial protection, anesthetic management, surgicaltechniques, and postoperative intensive care havebeen and continue to be subjected to extensive re-search. Refinements in all of these fields showedpositive effects on their own, but did not demon-strate the expected significant improvement in over-all patient outcome.

Examination of outcomes must include consider-ation of the change in patient population that hastaken place over this time period. It is widely ac-cepted that patients demonstrate significant changesin age, gender, cardiac condition, and their comor-bidity (1,2). This may be explained by improved con-servative cardiac treatment as well as improved

minimally invasive techniques such as percutaneoustransluminal coronary angioplasty (PTCA). Thisleads to a shift of indications for cardiac procedures.Different outcomes, less successful due to the in-creasing proportion of multimorbid patients present-ing for surgery, are expected. Therefore, the assess-ment of preoperative risk is essential for anyaccurate analysis of patient outcomes.

The need for further improvements in open heartsurgery management will become even more com-pelling in the future. Increased perioperative com-plications result in higher costs. This is not only lim-ited to hospital costs but also results in a subsequenteconomic burden to society and a reduced quality oflife for the patients concerned. Perioperative com-plications render patients unable to return to pro-ductive work or their normal activities of daily lifefor longer periods of time.

Cardiopulmonary bypass (CPB) represents a sig-nificant burden for the patient in addition to thesurgical procedure itself. Patients with numerous co-morbidities require additional therapeutic interven-tions to avoid perioperative complications followingCPB, including length of stay. Given the increasing

Received June 2000; revised November 2000.Address correspondence and reprint requests to Dr. Michael

Klein, Department of Cardiothoracic Surgery, Heinrich-HeineUniversity Hospital, Moorenstrasse 5, 40225 Dusseldorf, Ger-many.

Artificial Organs25(4):300–305, Blackwell Science, Inc.© 2001 International Society for Artificial Organs

300

number of multimorbid patients, further improve-ment in extracorporeal circulation technology maybe a key to improved overall patient outcome intimes of limited resources.

The arterial pump, in addition to the oxygenatorand the cardiotomy suction system, is one of themost critical items of an extracorporeal circuit. Dur-ing CPB, blood is driven through the arterial line andreturned to the patient by centrifugal forces that aregenerated within the pump (3). Substantial mechani-cal blood damage and device related complicationshave been reported with the use of a roller pump(4–10). Ample evidence has been accumulated fromin vitro, in vivo, and clinical studies that centrifugalpumps cause less blood trauma than roller pumpsduring prolonged application (10–20). Findings indi-cate that centrifugal systems are less likely to pumpgross air, generate considerably less arterial linepressure, have lower particle expulsion rates, reducecomplement activation, preserve hematologic integ-rity, and are less likely to cause coagulation disorders(9,21–23). In contrast, evidence shows that centrifu-gal pumps provide several advantages, such as im-proved blood and air handling, absence of spallation,and elimination of the risk of overpressurization(22,24,25).

Due to these advantages, the use of the centrifugalpumps in the United Sates and Canada accounts forabout 50% of the adult open heart cases today. Ap-proximately 80% of these procedures are performedwith the nonimpeller type BioMedicus centrifugalpump (as published in the San Diego Cardiopulmo-nary Bypass Symposium 1995 Survey). However, inEurope and particularly in Germany, the acceptanceof the centrifugal pump is much lower (estimated tobe less than 10%). Therefore, we decided to evaluatethe impact of different pumping technologies on pa-tient outcome in our clinical setting. The purpose ofthis study was to compare the roller pump to theBioPump centrifugal pump, targeting hematologicalparameters, transfusion requirements, postoperativeorgan function, and complications.

PATIENTS AND METHODS

A prospective, randomized study was carried outcovering 1,000 patients undergoing routine openheart surgery. All patients were randomly assignedto either a roller pump or a centrifugal pump. Allpreoperative parameters were collected at least 1day preoperatively (T−1) except for the plasma freehemoglobin, which was measured immediately at thebeginning and the end of extracorporeal circulation.Prior to the study, a pilot study was performed on

150 patients. This provided training for the perfu-sionists, assured the acceptance of the newly intro-duced centrifugal pump by the operating staff, andallowed for evaluation of the study protocol and thecomputerized data acquisition system, preventingany potential study effects.

Patients included were coronary artery bypassgrafting (CABG), valve replacements, combinedprocedures, and others such as hypertrophic obstruc-tive cardiomyopathy , ventricular aneurysms, or con-genital defects. Exclusion criteria were defined ascardiac output of less than 3 L/min, emergency cases,preoperative coagulopathies, aspirin therapy lessthan 10 days preoperatively, and patients less than 18years of age. Patients were randomly assigned to ei-ther study group until 1,000 complete data sets wereachieved. The database was then audited for com-pleteness of primary parameters (hematological andbleeding parameters) without breaking the random-ization code. A total of 163 patients were excludeddue to incomplete data for primary parameters. Pa-tient enrollment continued until 1,000 complete pa-tient records were attained.

Extracorporeal circuits were identical in bothgroups except for the arterial pump. The extracor-poreal circuit consisted of a hollow fiber oxygenatorwith an integral heat exchanger and cardiotomy res-ervoir CML Excel (Cobe Laboratories, Lakewood,CO, U.S.A.), and an arterial filter M-40 (Medtronic,Anaheim, CA, U.S.A.). In all cases, 6 mm metalarterial cannulas (custom made) and two 3⁄8 inch ve-nous cannulas (custom made) with metal tips wereused. The roller pump circuit contained a MultiflowRoller Pump (Stockert Instrumente, Munich, Ger-many), and the centrifugal pump circuit contained aBioPump Centrifugal Pump BP8O (Medtronic Bio-Medicus Inc., Eden Praine, MN, U.S.A.) as arterialpumps.

The preoperative medication and anesthetic man-agement was consistent during the study. No antifi-brinolytic agents were used during the course of theinvestigation. The protocol for postoperative carewas identical for both groups. All patients received 3IU/kg body weight heparin. Additional doses wereadministered at activated clotting time levels of lessthan 400 s. Protamine was given at a ratio of 1:1.5 ofthe initial heparin dose. Myocardial protection inopen heart procedures was performed with hypo-thermia between 24 to 29°C and additional cold crys-talloid high volume cardioplegia. All CABG caseswere done under intermittent aortic cross-clampingwith mild hypothermia (27 to 29°C). The trigger forintraoperative blood product use was a hematocrit of

IMPACT OF THE CENTRIFUGAL PUMP 301

Artif Organs, Vol. 25, No. 4, 2001

25% or a hemoglobin of less than 8.0 g/dl. Postop-eratively, a hemoglobin of 9.5 mg/dl and an albuminlevel of 2.5 mg/dl were accepted without the need fortransfusions. The routine antibiotic treatment wascephalosporine (Gramaxin, Boehringer MannheimGmbH, Mannheim, Germany).

DefinitionsMeans and standard deviations of patient demo-

graphics and the outcome variables for both groupsin this study are presented in Table 1. Numerousother variables were collected, but were not exam-ined for this study. Variables used in the analysis aredefined as follows: sex: 1 if male, 0 if female; myo-cardial infarction (MI): 1 if preoperative MI, 0 ifotherwise; diabetes: 1 if insulin dependent, 0 if oth-erwise; chronic obstructive pulmonary disease(COPD): 1 if COPD preoperatively, 0 if otherwise;age: age in years; risk: risk score predicted with Car-diac RiskMaster; preoperative hemoglobin: actualvalue; preoperative urea: actual value; MI enzymediagnosed: 1 if preoperative MI was diagnosed byenyzme levels, 0 if otherwise; chest tube drainage:actual amount of drainage in milliters; blood: num-ber of units of red blood cells; massive transfusion: 1if greater than 5 U of blood, 0 if otherwise; CABG:1 if CABG only, 0 if otherwise; bio: 1 if BioPump, 0if otherwise; bio*risk (BioPump use *risk); andlength of stay: number of days in hospital.

The group of patients randomly assigned to thecentrifugal pump group differed significantly (p <

0.05) from the control group preoperatively in age(significantly older) and percentage of patients withdiabetes (significantly more, 62.8% versus 57.6%).The centrifugal pump (BioPump) group also exhib-ited a higher, but not significantly different, inci-dence of COPD (Table 1).

Risk score calculationRisk scores were predicted using Cardiac Risk-

Master software (Health Data Research Inc., Port-land, OR, U.S.A.). Preoperative risk of mortality iscalculated based on the Bayes theorem (1,2). Thisrisk score can be used to predict future outcomesbased on associated previous events. The probabilityof mortality following cardiac surgery for any givenpatient with specific risk factors is computed. TheRiskMaster score is based on data from the interna-tional Merged Cardiac Registry (MCR). The MCRdatabase contains data from 45 institutions in theUnited States and Europe on more than 140,000 pa-tients over the last 3 years. Table 2 shows a list of therisk factors that were incorporated in the analysis.

Statistical analysisStatistical analysis was performed using ordinary

(nonweighted) least squares and logistic multiple re-gression analysis. Multiple linear regression was usedto identify variables that were independently associ-ated with the respective outcome variable (e.g., chesttube drainage, units of blood, massive transfusion,length of stay). Only those variables having a p value

TABLE 1. Patient demographics

VariableRoller pump

(n 4 500)BioPump(n 4 500)

pvalue

Age (years) 61.76 ± 10.98 64.38 ± 9.87 0.00Sex (%) 73 73 0.50Body surface area (cm2) 1,883.53 ± 491.90 1,849.63 ± 260.93 0.09Peripheral vascular disease 1.96 ± 0.20 1.96 ± 0.19 0.44Chronic obstructive pulmonary disease (%) 3.4 5.2 0.08Risk score 7.15 ± 10.43 7.04 ± 9.77 0.43Diabetes (%) 57.6 62.8 0.05Hypertension (%) 35.8 37.8 0.26Left ventricular function 1.82 ± 0.58 1.79 ± 0.60 0.20Preoperative PTCA (%) 57.4 53.8 0.31Preoperative red blood count 4.49 ± 119.70 4.41 ± 90.20 0.50Preoperative urea 41.60 ± 142.47 43.21 ± 126.61 0.43Preoperative hemoglobin 13.38 ± 120.78 13.25 ± 78.75 0.49Preoperative creatinine 1.11 ± 135.54 1.14 ± 119.30 0.50Preoperative white blood count 7.94 ± 382.79 7.51 ± 423.33 0.49Preoperative myocardial infarction (%) 37 38.2 0.35Aortic cross-clamp time (min) 115.50 ± 50.31 116.93 ± 100.03 0.39CABG only (%) 76.8 76.4 0.41Chest tube drainage 1,017.31 ± 1,510.40 666.08 ± 691.33 0.00Blood 2.31 ± 4.50 2.00 ± 45.02 0.44Length of stay (days) 11.16 ± 2.90 11.36 ± 2.91 0.13Reoperation (%) 3.4 3.6 0.43Death (%) 4.4 4.6 0.44

PTCA: percutaneous transluminal coronary angioplasty, CABG: coronary artery bypass grafting.

M. KLEIN ET AL.302


less than 0.05 were considered to contribute signifi-cantly to predicting outcome variables. Variables in-dependently associated with amount of chest tubedrainage, actual units of blood products used, andlength of stay (in days) were examined with ordinaryleast squares multiple regression analysis. Residualswere plotted to ensure a reasonably normal distribu-tion.

The ability of the independent variables to predictthe need for a massive transfusion is examined withthe odds ratio, presented in Table 3 with the logisticregression results. The odds ratio is the number bywhich we would multiply the odds of developing theparticular outcome for each 1-unit increase in theindependent variable. An odds ratio greater than 1indicates that the odds of requiring a massive trans-fusion increase when the independent variable in-creases. An odds ratio of less than 1 indicates thatthe odds of requiring a massive transfusion decreasewhen the independent variable increases.

RESULTS

Univariate and multivariate analysesTo find the predictors of blood product usage, the

following variables were tested by univariate analy-sis followed by multivariate analysis: age, gender,body surface area, left ventricular ejection fraction,diabetes, peripheral vascular disease, chronic ob-structive pulmonary disease, hypertension, preop-erative creatinine, preoperative hemoglobin, preop-erative urea, preoperative white blood cells,preoperative platelets, preoperative PTCA, preop-erative MI, risk score, CABG only, aortic cross-clamp time, chest tube drainage, use of the

BioPump, and the interaction of the BioPump withthe risk score.

Use of blood productsThe number of units of blood as well as the need

for a massive transfusion was analyzed by examiningthe variables previously listed. Also examined wasthe amount of chest tube drainage since it also maycontribute to the use of blood products, independentof other factors. These results are presented in Table4. Significant (p < 0.05) independent predictors ofthe amount of chest tube drainage were preoperativedialysis, preoperative MI enzyme diagnosed, CABGonly, aortic cross-clamp time, and use of BioPumpcentrifugal pump. Increase in aortic cross-clamptime, the patient having emergent surgery, patientbeing on dialysis preoperatively, preoperative MIbeing enzyme diagnosed, and patients having CABGonly significantly increased chest tube drainage. Pa-tients who used the BioPump centrifugal pump hadsignificantly less chest tube drainage.

Significant (p < 0.05) independent predictors ofthe number of units of blood were preoperative he-moglobin, risk score, amount of chest tube drainage,use of BioPump centrifugal pump, and the interac-tion of the BioPump with the risk score. These re-sults are presented in Table 5. Increasing risk scoreand increasing test tube drainage increased the needfor units of blood. Increasing preoperative hemoglo-bin and use of the BioPump centrifugal pump de-creased the need for units of blood transfused. Ad-ditionally, the BioPump interacted with risk score tohave an additional significant impact on reducing the

TABLE 2. Bayesian risk factors

Age Diabetes Arterial hypertensionBody surface area Failed PTCA Pulmonary hypertensionCongestive heart failure Hemodynamic status ObesityPeripheral vascular disease Left ventricular function Left main diseaseReoperation Congestive heart failure Mitral valve diseaseCentral nervous system disease COPD

Myocardial infarctionAortic valve disease

PTCA: percutaneous transluminal coronary angioplasty, COPD: chronic obstructivepulmonary disease.

TABLE 3. Independent predictors of massive transfusion(massive transfusion is greater than 5 U)

Variable Coefficient (SD) p value Odds ratio

Constant −1.92 (0.899) 0.0322 0.15BioPump −0.46 (0.19) 0.0178 0.63Age 0.03 (0.01) 0.0018 1.03Preop hemoglobin −0.18 (0.05) 0.0001 0.84Preop urea 0.11 (0.004) 0.0028 1.11

TABLE 4. Independent predictors of chest tube drainage

Variable Coefficient (SD) p value

Constant 531 (126) 0.0000BioPump −361 (73) 0.0000CABG 227 (86) 0.0081Dialysis 905 (235) 0.0001Cross-clamp time 2.3 (0.76) 0.0033MI (enzyme diagnosed) 377 (150) 0.0116

CABG: coronary artery bypass grafting, MI: myocardial infarc-tion.



need for blood; those with an increasing risk scorerequired fewer units of blood in the BioPump.

Significant (p < 0.05) independent predictors ofthe need for a massive transfusion were age, preop-erative hemoglobin, preoperative urea, emergentsurgery, and use of the BioPump centrifugal pump.Age and increased preoperative urea increased theodds of massive transfusion. Each additional year ofage increases the odds of requiring a massive trans-fusion by 3%, and each unit increase in the preop-erative urea increases the odds of requiring a mas-sive transfusion by 11%. Increased preoperativehemoglobin and use of the centrifugal pump de-creased the odds of massive transfusion. Each unitincrease in the preoperative hemoglobin decreasesthe odds of requiring a massive transfusion by 18%.Patients who used the BioPump centrifugal pumpwere only 63% as likely to require a massive trans-fusion. These results are presented in Table 5.

Hospital stayThe length of hospital stay (in days) was analyzed

by examining the variables previously listed. Signifi-cant (p < 0.05) independent predictors of the lengthof stay were preoperative myocardial infarction, gen-der, diabetes, preoperative hemoglobin, preopera-tive urea, age, body surface area, massive transfusionrequired, and CABG only (Table 6). The use of theBioPump centrifugal pump significantly decreased

the need for massive transfusion, and massive trans-fusion, in turn, increases the length of stay.

DISCUSSION

This study was designed to answer the questionwhether the use of the BioPump centrifugal pumpaffects clinical outcomes. We demonstrated that,compared with using a roller pump as the arterialpump in the extracoporeal circuit, the use of theBioPump centrifugal pump significantly decreasesboth the magnitude of homologous transfusion aswell as the incidence of massive transfusion and is asignificant, independent predictor of decreasedlength of stay. Only the difference in arterial pumpused can explain the differences in clinical outcomesbetween the treatment groups since the circuit con-figurations, perfusion techniques, and clinical path-ways used to manage patients were identical.

Acknowledgments: The authors thank the members ofour surgery, anesthesia, and perfusion departments fortheir cooperation during the study. This study was sup-ported in part by a grant from Medtronic GmbH, Ger-many.

REFERENCES

1. Yasnoff WA, Page US. Effective use of outcomes data incardiovascular surgery. SPIE Proc 1994;2307:212–22.

2. Edwards F, Albus RA, Zajtchuk R, Graeber GM, Barry MJ,Rumisek JD, Arishita G. Use of a bayesian model for riskassessment in coronary artery surgery. Ann Thorac Surg 1988;45:437–40.

3. Lynch MF, Peterson D, Baker V. Centrifugal blood pumpingfor open heart surgery. Minnesota Medicine 1978;61:536–7.

4. Koja K, Kuniyoshy Y, Ikemura F, Iha K, Kusaba A, Shima-bukuro M. Influence of the centrifugal pump (BioPump) onblood components: Comparison with roller pump. Jpn J ArtifOrgans 1986;15:545–8.

5. Hoerr HR, Kraemer MF, Williams JL, Sherman ML, RileyJB, Crowley JC, Soronen SW. In vitro comparison of theblood handling by the constrained vortex and twin rollerblood pumps. Trans Am Soc Artif Organs 1988;34:500–4.

6. Oku T, Harasaki H, Smith W, Nose Y. Hemolysis. A com-parative study of four nonpulsatile pumps. ASAIO Trans1988;34:500–4.

7. Takeda H, Goda T, Uzawa S, Takabe T, Matsurka H, SakaiK, Tanabe T. A comparison of centrifugal and roller pumps.An experimental study. Jpn J Thorac Surg 1984;37:273–7.

8. Jakob H, Kutschera Y, Palzer B, Prellwitz W, Oelert H. In-vitro assessment of centrifugal pumps for ventricular assist.Artif Organs 1990;14:278–83.

9. Mandl JP. Comparison of embol production between a con-strained force vortex pump and a roller pump. Am SECTProc 1977:27–31.

10. Jakob H, Maass D, Kutschere Y, Palzer B, Dennebaum R,Prellwitz W, Oelert H. Ventricular assist with centrifugalpump systems: Early clinical experience and in-vitro compari-son for blood trauma (Abstract). In: Thoma H, Schima H, eds.ASAIO 33rd Annual Meeting Proceedings, Vienna, 1988.

11. Golding LAR, Stewart RW, Loop FD. Centrifugal pumps inclinical practice. In: Unger F, ed. Assisted Circulation, 3rd ed.Berlin: Springer, 1989:160–6.

TABLE 5. Independent predictors of use ofblood products


Constant 3.61 (0.76) 0.0000BioPump −0.69 (0.28) 0.0137Preop hemoglobin −0.26 (0.053) 0.0000Risk score 0.82 (0.017) 0.0000Chest tube drainage 0.16 (0.0001) 0.0000Bio*risk (interaction term) −0.68 (0.023) 0.0030

TABLE 6. Independent predictors of length ofstay (days)


Constant 9.05 (0.71) 0.0000MI 1.25 (0.15) 0.0000Sex (male) −0.95 (0.16) 0.0000Diabetes −0.56 (0.15) 0.0001Preop hemoglobin −0.10 (0.032) 0.0020Preop urea 0.022 (0.003) 0.0000Age 0.09 (0.007) 0.0000BSA −0.0006 (0.00017) 0.0004Massive transfusion 0.63 (0.20) 0.0014CABG only −2.45 (0.17) 0.0000

MI: myocardial infarction; BSA: body surface area, CABG:coronary artery bypass grafting.

M. KLEIN ET AL.304


12. Driessen JJ, Fransen G, Rondelez L, Schelstraete E, GevaertL. Comparison of the standard roller pump and a pulsatilecentrifugal pump for extracorporeal circulation during rou-tine coronary artery bypass grafting. Perfusion 1991;6:303–11.

13. Pertilla J, Sab M, Peltola O. Comparison of the effects ofcentrifugal versus roller pump on the immune response inopen-heart surgery. Perfusion 1995;10:249–56.

14. Nowak J, Maass D, Lichtensteiner M, Oelert H. Comparativestudy on cost-benefit of Medtronic BioMedicus centrifugalpump versus roller pump curing cardiopulmonary bypass(CPB)—Preliminary results (Abstract). XXI World CongressInternational Society for Cardiovascular Surgery, 1993.

15. Maas C, Kok R, Segers P, Boogaart A. Comparison betweena centrifugal and a roller pump (Abstract). Pathophysiologyand Techniques of Cardiopulmonary Bypass, San Diego, CA,1993.

16. Klein M, Dauben H-P, Schulte HD, Gams E. Centrifugalpump during routine open heart surgery improves clinicaloutcome. Artif Organs 1998;22:326–36.

17. Hansbro SD, Sharpe DA, Catchpole R, Welsh KR, et al. Hae-molysis during cardiopulmonary bypass: An in vivo compari-sion of standard roller pumps, nonocclusive roller pumps andcentrifugal pumps. Perfusion 1999;14:3–10.

18. Yasuda T, Sawa S, Ishikawa N, Tanaka N, Manaka K. Clinicaleffect of high-flow pulsatile cardiopulmonary bypass in coro-nary artery bypass grafting. Kyobu Geka 1997;50:1091–4.

19. Ashraf S, Butler J, Tian Y, Cowan D, Lintin S, Saunders NR,Watterson KG, Martin PG. Inflammatory mediators in adultsundergoing cardiopulmonary bypass: Comparison of centrifu-gal and roller pumps. Ann Thorac Surg 1998;65:480–4.

20. Steines D, Westphal D, Gobel C, Reul H, Rau G. Plateletfunction and hemolysis in centrifugal pumps: in vitro investi-gations. Int J Artif Organs 1999;22:559–65.

21. De Bois WJ, Brennan R, Wein E, Isom OW, Gold JP. Cen-trifugal pumping: The patient outcome benefits followingcoronary artery bypass surgery. J Extra Corpor Technol 1995;27:77–80.

22. Parault BG, Conrad SA. The effect of extracorporeal circu-lation time and patient age on platelet retention during car-diopulmonary bypass. J Extra Corpor Technol 1991;23:34–8.

23. Salama A, Hugo F, Heinrich D, et al. Deposition of intimalcsb-9 complement complexes on erythrocytes and leukocytesduring cardiopulmonary bypass. N Engl J Med 1988;318:408–14.

24. Jakob HG, Hafner G, Thelemann C, Sturer A, Prellwitz W,Qelert H. Routine extracorporeal circulation with a centrifu-gal or roller pump. ASAIO Trans 1991;37:M487–9.

25. Berki T, Gurbuz A, Isik O, Akkaya H, Bayezid O, Yakut C.Cardiopulmonary bypass using centrifugal pump. Vasc Surg1992;26:123–34.



blood product use during routine open heart surgery: the impact of the centrifugal pump

Documents