Best Evidence Topics
  • Send this BET as an Email
  • Make a Comment on this BET

Does the use of a centrifugal pump offer any additional benefit for patients having open heart surgery?

Three Part Question

In [patients undergoing cardiac surgery,] does [the use of a centrifugal pump] improve [post operative outcomes].

Clinical Scenario

You are about to perform a Coronary arterial bypass graft on a 75-year-old gentleman with good LV function. You have recently been asking your perfusionist to use a centrifugal pump for all your high risk cases, and the perfusionist asks you if you want one for this case. You say 'yes' but you are unsure if there is evidence for benefit for these lower risk patients and, therefore, resolve to search for papers on the subject after the case.

Search Strategy

Medline 1966 to Oct 2004, Embase 1980 to Oct 2004 and CINAHL 1982 to Oct 2004 using the OVID interface
Extracorporeal Circulation/OR extracorporeal OR exp Heart-Lung Machine OR heart-lung OR exp Heart assist devices/OR Heart-assist] AND [exp centrifugation/OR centrifugal pump$.mp OR vortex pump$.mp]. and [roller OR roller head pump$.mp].

Search Outcome

A total of 93 papers were identified of which 26 papers were either large cohort studies or randomised controlled trials in patients undergoing cardiac surgery, comparing roller pumps with centrifugal pumps. Fifteen papers were selected representing the best evidence for the topic (Table) and an additional 11 smaller studies were rejected [Babin-Ebell, Wheeldon, Zirbel, Misoph, Murakami, Moen, Yoshikai, Pertilla, Steinbrueckner, Nishinaka, Jakob].

Relevant Paper(s)

Author, date and country Patient group Study type (level of evidence) Outcomes Key results Study Weaknesses

90 patients undergoing CABG randomised to: 1. Low-dose heparin, centrifugal pump with heparin coated circuit, ACT 150–400s, n=30 2. Standard bypass, roller pump, ACT>400s, n=30 3. Centrifugal pump with heparin coated circuit, ACT>400s, n=30PRCT (level 1b)Clinical outcomesNo differences between groupsStudy is underpowered to study most clinical outcome measures
Bleeding and transfusion requirementNo differences between groups
Lindholm et al.
41 elderly patients undergoing CABG or AVR randomised to: Closed circuit, Heparin coated system and centrifugal pump, n=21 Conventional system and roller pump, n=20 Range of inflammatory markers measured pre-op, on bypass, on rewarming, 60 min after bypass, and day 1 post-surgery Mean age 73 yearsPRCT (level 1b)Clinical outcomesBleeding, inotropic support and TnT similar

5 pts in conventional group stayed more than 1 day in ICU. Non in centrifugal group (P=0.014)
Complement activationC3a, sC5b-9, Bb significantly lower during rewarming in centrifugal group. No significant differences ata 24 h
Cytokine releaseIL-8 significantly lower on rewarming, but no differences in PMN, TNF, IL-6 at any stage
Haemostatic variablesNo differences in D-Dimer, t-PA-ag, t-PA-PAI or thrombinantithrombin at any times
Macey et al.
46 patients undergoing elective CABG randomised to Roller pump 23 Centrifugal pump 23 Exclusions criteria diabetes, PVD, CVA or steroidsPRCT (level 2b)CD11b, Cd18, CD14 Cd64 L selectinIncrease in CD11b expression but independent of pump type

No alteration in expression of other markers
Small numbers Only 2 samples collected Only male patients
Driessen et al.
Two consecutive groups of 25 patients <70 yo and EF >40% undergoing CABG 25 received standard non-pulsetile twin roller pump 25 received Sarns centrifugal pump All patients given 200 mg dexamethasone cardioplegia cold crystalloid All patients remained in ITU for 48 hCohort study (level 2b)Blood loss and transfusionNo differenceSmall study Not randomised Cross clamp time significantly longer in the centrifugal group 63 min vs 52 min P<0.05
Haemodynamic studiesRoller group needed more sodium nitropruside during CPB. No differences in SVR, PVR, Hb, platelet count or ICU stay
Baufreton et al.
France and Holland
29 patients elective CABG high risk patients included Roller pump=15 Centrifugal pump=14 Exclusion criteria, recent aspirin intake, valvular or ventricular surgery, impaired organ function except myocardial ischaemia, coagulopathy, diabetes, active inflammatory diseasePRCT (level 2b)TNF, IL6, IL8, ELAM-1, ICAM-1No significant differenceSmall study Aprotinin used but no indication in which patients it was used despite randomisation significant difference in haematocrit and haemoglobin preop
Terminal complement complex activation SC5b-9Centrifugal pump patients levels significantly higher on stopping bypass and on giving protamine 772 vs 274 (P+0.01) and 161 vs 66 (P=0.006)
Neutrophil activation (Elastase level)CFP group 161±125 mg/l. Roller pump 66±31 mg/l P=0.006
Andersen et al.
34 patients undergoing elective CABGx3 or more, randomised to: 1. Biomedicus centrifugal group n=17 2. Roller pump group=17 Exclusions LV aneurysms, valvular procedures, EF<50%, pulmonary disease, IDDM, renal failure HB<12, aspirin use less than 7 days pre-op Primary outcome measure, platelet aggregation post-operativelyPRCT (level 2b)Clinical outcomesNo difference in blood product use

No difference in hospital stay, blood loss or duration of ventilation

Significant increase in arrhythmias in centrifugal pump group 11/17 vs 5/17 (P<0.05)
Study supported by Medtronic No difference in clinical outcomes. This was a small group of low risk patients. It was underpowered to detect small differences in clinical outcome
Platelet aggregationAfter bypass number of platelet aggregates higher in roller pump group initially and after 3 days no difference between the groups in number of microvesicles or single platelets
HaemolysisNo difference
Ashraf et al.
41 elective patients undergoing CABG Randomised to 2 groups 1. Roller pump n=21 2. Centrifugal pump n=20 Exclusion criteria, unstable angina, MI within 3 months, reoperation, diabetes liver or renal failure, COAD, anticoagulant use, immunosuppression Neutrophil elastase and neutrophil counts measured blood loss and cardiac outputs were also measured A subset had S-100 Beta measured in 32 patients, reported in their second paperPRCT (level 2b)Blood loss and transfusion requirementNo significant differenceSmall number of patients
Interleukin 1 BetaNot detectable
Interleukin 6Larger peak 2 hours post-protamine administration 341 vs 260 pg/ml (P<0.05) in the centrifugal group
Neutrophil countHigher peak in centrifugal 8.15x10 to the power of 9 vs. 6.2x10 to the power of nine (P=0.03)
C5b-9 complexTotal count was higher in the centrifugal group (P<0.05)
NeutrophilsRelease during and after CPB higher in the centrifugal group (P<0.05)
Plasma leukocyte elastaseNo difference in clinical outcomes
Serum S100 betaRoller pump 2.26±0.39 ng/ml Centrifugal pump 2.11±0.27 ng/ml P=NS
Hansbro et al.
60 patients undergoing coronary arterial bypass surgery. Randomised to 3 groups: Standard roller pump n=20 dynamically set non-occlusive roller pump n=20 centrifugal pump n=20 CPB time longer in the dynamic group P<0.05PRCT (level 1b)Plasma free haemoglobin levels (Indicator of haemolysis)Standard roller pump 0.424±0.17. Non-occlusive roller pump 0.481±0.2. Centrifugal pump 0.471±0.18 P=NSShort bypass times Small numbers Two surgical teams
RBC count, WCCNo difference
Scott et al.
2001 and 2002
113 patients undergoing first time CABG Randomised to: 57 pts had centrifugal pump 56 pts had roller pump Neuropsychiatric testing on two occasions Impairment was defined as 1 S.D. deviation in previous testing 130 enrolled but 27 didn't complete 54 in centrifugal group - 49 in roller groupPRCT (level 3b)Neuropsychological outcomeTrend towards improved neuropsychological outcome at day 5 with centrifugal compared to roller but results not significantSmall study 27 pts did not complete the neuro-psychometric testing No sample size performed
Blood product usage, haemoglobin, platelet counts and chest tube drainageNo difference
Chest tube drainageCentrifugal pump 1300±92 ml. Roller pump 1117±83 ml P=0.14
Klein et al.
1998 and 2001
1000 adults undergoing elective open heart procedures from 1994 to 1995 randomised to 2 groups 1. Biopump centrifugal group n=500 2. Stockert roller pump group n=500 Exclusion criteria, CO<3 l/min, emergency cases, preoperative coagulopathies, aspirin usage less than 10 days preop, age less than 18 Block randomisation performed according to 3 risk groupsPRCT (level 2b)MortalityNo significant decrease in mortalityThis study was supported by a grant from Medtronic 163 patients excluded due to inadequate data collection. Statistical methods 1998 not given in 1998 paper Inappropriate use of multivariate analysis in an RCT in the 2001 paper Group demographics not given in 1998 paper Results poorly presented
Haemoglobin levelsSignificantly higher haemoglobin, RBC count and haematocrit corrected RBC count at day 1 and 7 in centrifugal pump group. P value not given

No difference in haemolysis
Chest tube drainageCentrifugal group 666 ml±691. Roller pump 1,017 ml± 1,510 ml P<0.01
Neurological outcomesCentrifugal pump 2.4% Roller pump 5.1% P<0.05 NNT 37
Other outcomesNo difference in renal function, myocardial infarction, arrhythmias and time of recovery
Parolari et al.
4000 patients undergoing cardiac surgery between Jan 1994 and Jun 1999 2213 pts had a centrifugal pump 1787 pts had a Roller pump Patients generally had centrifugal pump when CPB>100 min anticipated Median age 65 72% maleCohort study (level 2b)Hospital mortalityCentrifugal pump 50/2213 (2.3%). Roller pump 38/1787 (2.1%) P=NSRetrospective study with unmatched groups. Benefits proven despite clear selection bias centrifugal pump preferred for longer cases
Permanent neurological deficitCentrifugal pump 34/2213 (1.5%). Roller pump 47/1787 (2.6%) P=0.020

NNT 91 Multivariate analysis demonstrated protective benefit of centrifugal pump
Perioperative comaCentrifugal pump 20/2213 (0.9%). Roller pump 32/1787 (1.8%) P=0.020
Alamanni et al.
Cohort of 3438 patients undergoing coronary± valve surgery between Jan 1994 to Dec 1998 1805 pts had a centrifugal pump (Biomedicus) 1633 pts had a roller pump (Stokert) Assessed for neurological outcomes Patients generally had centrifugal pump when CPB>100 min anticipated Patients having aneurysmectomy, carotid endarterectomy or aortic replacement excludedCohort study (level 2b)Neurological complicationsCVA centrifugal pump 29/1805 (1.6%). Roller pump 39/1633 (2.4%) P=0.127 NNT 125

Coma centrifugal pump 18/1805 (1.0%). Roller pump 26/1633 (1.6%) P=0.116

All neuro complications centrifugal pump 92/1805 (5.2%). Roller pump 85/1633 (5.1%) P=0.838
Allocation of patients to centrifugal pump non-randomised and heavily biased to more complex procedures
Neurological complications in >75 year oldsCVA centrifugal pump 25/1643 (1.5%). Roller pump 37/1492 (2.5%) P=0.054 NNT 100
Multivariate analysis for permanent neurological deficitsBypass time (P<0.01) Previous TIA (P=0.009) Age (P=0.011) Centrifugal pump (P=0.042), OR 0.77
In-hospital mortalityCentrifugal pump 41/1805 (2.3%). Roller pump 35/1633 (2.1%) P=0.692


Parolari et al. performed a retrospective analysis of 4000 patients who underwent open-heart surgery either with a centrifugal pump or with a roller pump. (Results published again with fewer patients by Alamanni et al. The groups were not matched and the centrifugal pump was generally selected when a bypass time over 100 min was anticipated. A range of pre-operative variables such as TIA, MI, re-operation were also higher in the centrifugal pump group. There was no difference in hospital mortality between the groups but the centrifugal pump patients had a significantly lower permanent neurological deficit rate (1.5% vs. 2.6%). This halving of the stroke rate was despite a higher incidence of pre-op CVAs and longer bypass times in the centrifugal pump group. Klein et al. performed a prospective randomised trial of 1000 patients undergoing routine open heart surgery. They demonstrated significantly higher red blood cell counts, 30% less chest tube drainage, lower transfusion rates and significantly lower incidence of neurological events (5.2% vs. 2.4%) in the centrifugal pump group (NNT 100). Interestingly, there was no difference in the rate of haemolysis and also no mortality difference. This study was sponsored by the manufacturers of the centrifugal pump. Scott et al. randomised 113 elective coronary artery bypass patients and compared the effects of pump type on blood loss, transfusion requirements and also neuropsychometric testing. No difference in blood product usage, haemoglobin, platelet numbers and chest tube drainage were detected. In addition, no differences in neuropsychometric testing were found. Driessen et al. enrolled 50 elective patients undergoing coronary artery bypass grafting. They were unable to demonstrate any significant haemodynamic differences, or differences in ICU stay, Hb level or platelet count. However, they detected a need for greater amounts of sodium nitroprusside during cardiopulmonary bypass in the roller pump group. Hansbro et al. performed a randomised study in 60 patients, looking at standard roller pumps, non-occlusive roller pumps and centrifugal pumps. They found no increase in rates of haemolysis, haemoglobin levels, or white cell count after the study, although the numbers were small. Ashraf et al. studied 40 patients randomised to centrifugal pump or a non-occlusive roller pump. Blood loss and haemodynamics were recorded and there was no significant difference between the groups. In contrast to other studies, neutrophil count, IL-6, and elastase release were higher in the centrifugal pump group suggesting an increase in inflammatory response using the centrifugal pump. In a separate paper Ashraf also measured S-100 Beta in these patients as a marker of brain injury but found no significant difference. Andersen et al. performed a PRCT in 34 patients, looking at the inflammatory effects of cardiopulmonary bypass with particular emphasis on platelet activation. No clinical differences were found and only a marginal increase in platelet aggregates were seen in the roller pump group. No differences in haemolysis, single platelets, or platelet fragments were seen. Baufreton et al. sought to demonstrate reduced inflammation with a centrifugal pump. In fact in a randomised study of 29 patients they showed a higher neutrophil and complement activation level in the centrifugal pump group. No other differences were shown, either clinically or in levels of interleukins or TNF. Macey et al. measured the levels of CD11b, Cd18, CD14, Cd64 and L selectin in 46 patients randomised to either roller or centrifugal pumps. All measured rises in inflammatory mediators were independent of pump type used. Lindholm et al. studied 2 groups of elderly patients randomised to either a closed system, heparin bonded circuit with a centrifugal pump or a standard circuit with a roller pump. They found significantly decreased levels of 3 out of 4 complement activation measures, IL-8, and Elastase on rewarming. However, there were no differences in TNF, IL-6, C4d, TAT, D-Dimer, or t-PA at any stage, and all markers were at similar levels by 24 h. Mullen et al. randomised 90 patients to roller pump, centrifugal pump with heparin circuit and either normal or low heparinization. They found no clinical differences in bleeding, transfusion requirements or clinical outcomes between groups.

Clinical Bottom Line

The two largest studies, a large RCT of 1000 patients and a cohort study of 4000 patients, both demonstrated a halving in the incidence of neurological events with numbers needed to treat of 37 and 91. However, the remaining much smaller RCTs and cohort studies that we assessed failed to show significant differences in either clinical or biochemical markers.


  1. Jakob HG, Hafner G, Thelemann C, Sturer A, Prellwitz W, Oelert H. Routine extracorporeal circulation with a centrifugal or roller pump. ASAIO Trans 1991; 37:M487–M489.
  2. Nishinaka T, Nishida H, Endo M, Miyagishima M, Ohtsuka G, Koyanagi H. Less blood damage in the impeller centrifugal pump: a comparative study with the roller pump in open heart surgery. Artif Organs 1996; 20:707–710.
  3. Steinbrueckner BE, Steigerwald U, Keller F, Neukam K, Elert O, Babin-Ebell J. Centrifugal and roller pumps – are there differences in coagulation and fibrinolysis during and after cardiopulmonary bypass. Heart Vessels 1995; 10:46–53.
  4. Perttila J, Salo M, Peltola O. Comparison of the effects of centrifugal versus roller pump on the immune response in open-heart surgery. Perfusion 1995; 10:249–256.
  5. Yoshikai M, Hamada M, Takarabe K, Okazaki Y, Ito T. Clinical use of centrifugal pumps and the roller pump in open heart surgery: a comparative evaluation. Artif Organs 1996; 20:704–706.
  6. Moen O, Fosse E, Dregelid E, Brockmeier V, Andersson C, Hogasen K, Venge P, Mollnes TE, Kierulf P. Centrifugal pump and heparin coating improves cardiopulmonary bypass biocompatibility. Ann Thorac Surg 1996; 62:1134–1140.
  7. Murakami F, Usui A, Hiroura M, Kawamura M, Koyama T, Murase M. Clinical study of totally roller pumpless cardiopulmonary bypass system. Artif Organs 1997; 21:803–807.
  8. Misoph M, Babin-Ebell J, Schwender S. A comparative evaluation of the effect of pump type and heparin-coated surfaces on platelets during cardiopulmonary bypass. Thorac Cardiovasc Surg 1997; 45:302–306.
  9. Zirbel GM, Letson ME, Kauffman JN, Walker CT, Guyton RA. Hematologic derangements of cardiopulmonary bypass: a comparison of two perfusion systems. J Extra Corporeal Technol 1990; 22:15–19.
  10. Wheeldon DR, Bethune DW, Gill RD. Vortex pumping for routine cardiac surgery: a comparative study. Perfusion 1990; 5:135–143.
  11. Babin-Ebell J, Misoph M, Mullges W, Neukam K, Elert O. Reduced release of tissue factor by application of a centrifugal pump during cardiopulmonary bypass. Heart Vessels 1998; 13:147–151.
  12. Mullen JC, Bentley MJ, Gelfand ET, Koshal A, Modry DL, Guenther CR, Etches WS, Stang LJ, Lopushinsky SR. Coronary artery bypass surgery with heparin-coated perfusion circuits and low-dose heparinization. Can J Surg 2002; 45:166–172.
  13. Lindholm L, Westerberg M, Bengtsson A, Ekroth R, Jensen E, Jeppsson A. A closed perfusion system with heparin coating and centrifugal pump improves cardiopulmonary bypass biocompatibility in elderly patients. Ann Thorac Surg 2004; 78:2131–2138.
  14. Macey MG, McCarthy DA, Trivedi UR, Venn GE, Chambers DJ, Brown KA. Neutrophil adhesion molecule expression during cardiopulmonary bypass: a comparative study of roller and centrifugal pumps. Perfusion 1997; 12:293–301.
  15. Driessen JJ, Fransen G, Rondelez L, Schelstraete E, Gevaert L. Comparison of the standard roller pump and a pulsatile centrifugal pump for extracorporeal circulation during routine coronary artery bypass grafting. Perfusion 1991; 6:303–311.
  16. Baufreton C, Intrator L, Jansen PG, te Velthuis H, Le Besnerais P, Vonk A, Farcet JP, Wildevuur CR, Loisance DY. Inflammatory response to cardiopulmonary bypass using roller or centrifugal pumps Ann Thorac Surg 1999; 67:972–977.
  17. Andersen KS, Nygreen EL, Grong K, Leirvaag B, Holmsen H. Comparison of the centrifugal and roller pump in elective coronary artery bypass surgery – a prospective, randomised study with special emphasis upon platelet activation. Scand Cardiovasc J 2003; 37:356–362.
  18. Ashraf S, Bhattacharya K, Zacharias S, Kaul P, Kay PH, Watterson KG. Serum S100 beta release after coronary artery bypass grafting: roller versus centrifugal pump. Ann Thorac Surg 1998; 66:1958–1962.
  19. Ashraf S, Butler J, Tian Y, Cowan D, Lintin S, Saunders NR, Watterson KG, Martin PG. Inflammatory mediators in adults undergoing cardiopulmonary bypass: comparison of centrifugal and roller pumps. Ann Thorac Surg 1998; 65:480–484.
  20. Hansbro SD, Sharpe DA, Catchpole R, Welsh KR, Munsch CM, McGoldrick JP, Kay PH. Haemolysis during cardiopulmonary bypass: an in vivo comparison of standard roller pumps, nonocclusive roller pumps and centrifugal pumps. Perfusion 1999; 14:3–10.
  21. Scott DA, Silbert BS, Blyth C, O'Brien J, Santamaria J. Blood loss in elective coronary artery surgery: a comparison of centrifugal versus roller pump heads during cardiopulmonary bypass. J Cardiothorac Vasc Anesth 2001; 15:322–325.
  22. Scott DA, Silbert BS, Doyle TJ, Blyth C, Borton MC, O'Brien JL, de LH. Centrifugal versus roller head pumps for cardiopulmonary bypass: effect on early neuropsychologic outcomes after coronary artery surgery. J Cardiothorac Vasc Anesth 2002; 16:715–722.
  23. M E, Mahoney CB, Probst C, Schulte HD, Gams E. Blood product use during routine open heart surgery: the impact of the centrifugal pump. Artif Organs 2001; 25:300–305.
  24. Klein M, Dauben HP, Schulte HD, Gams E. Centrifugal pumping during routine open heart surgery improves clinical outcome. Artif Organs 1998; 22:326–336.
  25. Parolari A, Alamanni F, Naliato M, Spirito R, Franze V, Pompilio G, Agrifoglio M, Biglioli P. Adult cardiac surgery outcomes: role of the pump type. Eur J Cardiothorac Surg 2000; 18:575–582.
  26. Alamanni F, Parolari A, Zanobini M, Porqueddu M, Dainese L, Bertera A, Costa C, Fusari M, Spirito R, Biglioli P. Centrifugal pump and reduction of neurological risk in adult cardiac surgery. J Extra Corporeal Technol 2001; 33:4–9.